Association of ampullary and colorectal malignancies
Because of the similarities in terms of carcinogenesis and natural history between cancer of the ampulla of Vater and colorectal cancer, the authors examined whether ampullary and colorectal malignancies occur in the same individuals at increased rates.
We used data from the Surveillance, Epidemiology, and End Results (SEER) Program of the National Cancer Institute for the period from January 1973 through December 1999. Person-years of follow-up for patients with ampullary (or colorectal) cancer were used to calculate the expected number of cases of colorectal (or ampullary) cancer as a second primary malignancy. Standardized incidence ratios (SIRs) with 95% confidence intervals (CIs) were calculated using Byar limits and assuming a Poisson distribution.
The authors identified 2043 white patients with ampullary cancer who were included in the SEER registry between 1973 and 1999. Over an aggregate 5674 person-years of follow-up, 30 patients, compared with an expected 14, developed colorectal cancer, yielding an overall SIR of 2.14 (95% CI, 1.45–3.06). Similarly, 57 of 262,066 white patients with colorectal cancer developed ampullary cancer over an aggregate 1,270,255 person-years of follow-up, yielding an SIR of 2.18 (95% CI, 1.69–2.85).
Patients with ampullary cancer are at increased risk for a second primary colorectal malignancy, and patients with colorectal cancer are at increased risk for a second primary ampullary malignancy. These findings suggest that ampullary and colorectal malignancies share common environmental and/or genetic risk factors. Cancer 2004. © 2003 American Cancer Society.
Malignancies of the ampulla of Vater are relatively uncommon but quite distinct. Their overall age-adjusted incidence rate is 0.5 per 100,000 persons in the United States.1 They are the most curable malignancies of the pancreaticobiliary tree, because they present at an early stage with clinical signs or symptoms of biliary obstruction or pancreatitis. The observed 1-year and 5-year survival rates for patients with ampullary malignancy are 59% and 21%, respectively, compared with 16% and 2%, respectively, for patients with pancreatic malignancy.1
There are reasons to suspect that ampullary and colorectal neoplasms may be associated. First, molecular changes that have been identified in the progression from normal colonic epithelium to adenoma to adenocarcinoma also have been identified in the carcinogenic pathway for ampullary malignancies, and reports suggest that the sequence of progression from adenoma to carcinoma for ampullary neoplasms closely resembles the sequence observed in colonic malignancies.2–6 Second, experimental studies of bile from patients with familial adenomatous polyposis (FAP) suggest that exposure to genotoxic compounds in bile may play a role in the carcinogenesis of both ampullary and colorectal malignancies, possibly by inducing second-hit genetic mutations.7, 8 Third, epidemiologic studies suggest a possible bidirectional association between these two types of malignancies. Ampullary cancer as a second primary malignancy in patients with colorectal cancer has been reported several times, and although most of these reports involve patients with FAP,9, 10 there are some reports that suggest that ampullary cancer is associated with hereditary nonpolyposis colon carcinoma11, 12 and also with sporadic colon cancer.13 A number of population-based studies of second primary malignancies occurring after colorectal cancer have been performed.14–16 Although none of these studies examined the association between ampullary cancer and colorectal cancer exclusively, some did report an increased risk of small intestinal or bile duct malignancy in patients with colorectal cancer.15, 16 The number of anecdotal reports of subsequent primary malignancy in patients with cancer of the ampulla of Vater is limited.17, 18 Nonetheless, one study evaluating the risk of second primary malignancy in a large cohort of patients with ampullary cancer did not demonstrate an increased risk.19 The identification of an association between colorectal and ampullary malignancies would be clinically significant, because of the potential implications for colorectal screening of individuals with ampullary neoplasms and their family members.
The Surveillance, Epidemiology, and End Results (SEER) database of the National Cancer Institute has been used extensively to investigate associations between different pairs of primary malignant neoplasms.15, 16, 20–23 Hypothesizing that ampullary and colorectal malignancies may share common genetic and environmental risk factors, we used the SEER database to examine the association between these two types of cancer in the general population.
MATERIALS AND METHODS
Data for the current analysis were obtained from the 1973–1999 public use data file from the SEER Program of the National Cancer Institute, which is an authoritative source of information on cancer incidence and survival in the United States.1 Information was obtained from nine SEER registries, including the Atlanta, Georgia; Connecticut; Detroit, Michigan; Hawaii; Iowa; New Mexico; San Francisco–Oakland, California; Seattle–Puget Sound, Washington; and Utah registries. The SEER Program currently collects and publishes incidence and survival data from 11 population-based cancer registries and 3 supplemental registries, covering approximately 14 percent of the United States population. Information on more than 3 million cases of in-situ and invasive cancer is included in the SEER database, and approximately 170,000 new cases are accessioned each year within the SEER catchment areas. The SEER registries routinely collect data on patient demographics, primary tumor site, morphology, disease stage at diagnosis, first course of treatment, and follow-up for vital status. The SEER Program is the only comprehensive source of population-based information in the United States that reports disease stage at the time of diagnosis and survival rates within each stage. The mortality data reported by SEER are provided by the National Center for Health Statistics (Hyattsville, MD).
The association of ampullary cancer in patients with FAP and other familial polyposis syndromes is well described in the literature.9, 10, 24 Because our aim was to study the association between ampullary cancer and sporadic colorectal cancer, all patients who had colon cancer with International Classification of Diseases site/histology code 8220-1 (adenocarcinoma in familial polyposis coli) were excluded from the current analysis. In addition, because almost all patients with FAP are diagnosed with colon cancer before the fifth decade of life, we restricted the analysis to patients age ≥ 50 years.24
Colorectal Cancer as a Second Primary Malignancy following Ampullary Cancer
Ampullary lesions that were included in the current study were primary neoplasms that 1) had an ICD-O-2 (second edition of the International Classification of Diseases for Oncology25) classification of C24.1; 2) were microscopically confirmed; 3) were malignant (fifth digit of morphology code, 3); and 4) were newly diagnosed between January 1, 1973, and December 31, 1999. Cases of malignancy of the extrahepatic bile duct, overlapping lesions of the biliary tract, and malignancy of the head of the pancreas (ICD-O-2 classification: C24.0, C24.8–24.9, and C25.0, respectively) were excluded. Cases in which diagnosis of colorectal cancer occurred before diagnosis of ampullary cancer also were excluded.
The study cohort was followed within the SEER system to identify subsequent occurrences of primary colorectal cancer. Second primary lesions that were included were neoplasms of the colon and rectum (ICD-O-2 classification: C180–189, C199, C209, or C260; SEER site recode: 21041–21049 or 21051–21052) that were 1) malignant; 2) microscopically confirmed; 3) diagnosed subsequent to the index ampullary malignancy; and 4) diagnosed on or before December 31, 1999.
The follow-up time for each patient was calculated from the date of diagnosis of the index ampullary malignancy to the first of the following dates: the date of diagnosis of the second primary colorectal malignancy, the date of death (or last contact, if the patient was lost to follow-up), or December 31, 1999. Follow-up time was calculated in person-years of observation and was stratified by 5-year age group, race (white, black, or other), gender, and 5-year calendar interval in which diagnosis was made. The total number of person-years within each specific age, race, gender, and calendar interval category was multiplied by the category-specific incidence rate of microscopically confirmed, malignant colorectal disease in the SEER population to calculate the expected number of colorectal malignancies in the study cohort.
Ampullary Cancer as a Second Primary Malignancy following Colorectal Cancer
Using the same methods and study criteria, a cohort of individuals with colorectal cancer was identified and followed in the SEER population to determine the observed and expected incidence of ampullary cancer.
Calculation of the Standardized Incidence Ratio (SIR)
The SIR was calculated as the ratio of the observed number of second primary malignancies to the expected number of second primary malignancies. Statistical significance was assessed based on the assumption that the observed number of cases followed a Poisson distribution. Ninety-five percent confidence intervals were calculated for each SIR using the Byar approximation to the exact Poisson test.26 SEER data were analyzed using SEER*Stat 4.2 statistical software (National Cancer Institute, Bethesda, MD). All other statistical tests were performed using the Statistica software package (Statsoft, Tulsa, OK).
We identified 2397 individuals (1266 men and 1131 women) registered in the SEER population who were older than age 50 years and were diagnosed with ampullary cancer between January 1, 1973, and December 31, 1999. The overall age-adjusted (to the 2000 United States standard) incidence rates for ampullary cancer in this cohort were 2.1 and 1.3 per 100,000 persons for males and females, respectively. The percentages of patients with ampullary cancer in the sixth, seventh, and eighth decades of life were 13.6%, 27.1%, and 34.3%, respectively; 25% of all patients in the cohort were age ≥ 80 years. The median follow-up time for the 2397 patients identified was 1.16 years (25th–75th percentile, 0.33–3.17 years).
Most of the 2397 patients with ampullary cancer were white (n = 2043 [85%]). Because only a small proportion of ampullary malignancies were diagnosed in the nonwhite population, further analysis was restricted to the white population. Of these 2043 white patients, a total of 30 developed colorectal cancer over 5674 person-years of follow-up, compared with an expected 14 patients; these results yielded an overall SIR of 2.14 (95% confidence interval [CI], 1.45–3.06) (Table 1). The SIRs for white males and white females were 1.99 (95% CI, 1.14–3.23) and 2.32 (95% CI, 1.27–3.90), respectively. Of the 30 patients who developed colorectal cancer as a second primary malignancy, 11 had localized disease, 9 had regional disease, and 5 had metastatic disease; staging information was not available for 5 patients. The site-specific distribution of malignancies among these 30 patients was as follows: cecum, 5 patients; ascending colon, 2 patients; hepatic flexure, 3 patients; transverse colon, 1 patient; descending colon, 2 patients; splenic flexure, 1 patient; sigmoid colon, 6 patients; rectum, 8 patients; and unspecified exact site, 2 patients. This distribution was similar to the one observed in the overall SEER population of individuals with colorectal cancer. Nine patients were diagnosed with colorectal cancer as a second primary malignancy within 1 year of diagnosis of the initial primary ampullary malignancy. When these 9 patients were excluded from the analysis, the calculated SIR was 1.5 (95% CI, 1.01–2.15).
Table 1. Data on Colorectal Cancer as a Second Primary Malignancy in Patients with Ampullary Cancer
|White males|| || || || || |
| 50–54||52||297||0.17||0|| |
| 55–59||100||501||0.46||0|| |
| 60–64||151||517||0.89||1|| |
| 65–69||183||570||1.50||3|| |
| 70–74||199||444||1.61||5|| |
| 75–79||170||313||1.46||2|| |
| 80–84||116||179||1.02||4|| |
| 85+||113||153||0.93||1|| |
| Total|| || ||8.04||16||1.99 (1.14–3.23)|
|White females|| || || || || |
| 50–54||44||252||0.11||0|| |
| 55–59||69||280||0.22||2|| |
| 60–64||100||413||0.50||3|| |
| 65–69||127||456||0.80||2|| |
| 70–74||160||391||0.97||1|| |
| 75–79||171||467||1.54||0|| |
| 80–84||141||265||1.08||5|| |
| 85+||147||176||0.81||1|| |
| Total|| || ||6.03||14||2.32 (1.27–3.90)|
|Entire cohort|| || ||14.07||30||2.14 (1.45–3.06)|
A total of 292,763 patients (149,108 men and 143,656 women) were diagnosed with colorectal cancer during the study period. The overall age-adjusted (to the 2000 United States standard) incidence rates for colon cancer in this cohort were 243.4 and 174.2 per 100,000 persons for males and females, respectively. The median follow-up time for patients in this cohort was 4.81 years (25th–75th percentile, 0.75–7.33 years).
Most patients with colon cancer were white (n = 262,066 [89.5%]). Of these 262,066 patients, 57 developed ampullary cancer as a second primary malignancy over a cumulative follow-up period of 1,270,255 person-years. The expected number of ampullary malignancies in this cohort was 26, yielding an overall SIR of 2.18 (95% CI, 1.69–2.85). The SIRs for white males and white females were 2.15 (95% CI, 1.45–3.08) and 2.21 (95% CI, 1.43–3.27), respectively (Table 2). Ten patients were diagnosed with ampullary cancer within 1 year of diagnosis of an initial primary colorectal malignancy. When these patients were excluded, the SIR was 1.81 (95% CI, 1.33–2.41).
Table 2. Data on Ampullary Cancer as a Second Primary Malignancy in Patients with Colorectal Cancer
|White males|| || || || || |
| 50–54||7157||557,324||0.19||0|| |
| 55–59||11,963||914,229||0.67||2|| |
| 60–64||17,339||1,279,960||1.61||2|| |
| 65–69||22,234||1,420,527||2.55||3|| |
| 70–74||23,749||1,303,248||3.31||5|| |
| 75–79||21,287||960,663||2.98||5|| |
| 80–84||15,311||554,246||2.00||10|| |
| 85+||10,796||292,367||1.55||5|| |
| Total|| || ||14.86||32||2.15 (1.45–3.08)|
|White females|| || || || || |
| 50–54||5950||44,541||0.15||0|| |
| 55–59||9364||68,323||0.39||1|| |
| 60–64||13,624||91,585||0.83||2|| |
| 65–69||17,835||110,376||1.38||2|| |
| 70–74||21,589||115,954||2.12||4|| |
| 75–79||23,053||104,756||2.55||5|| |
| 80–84||20,276||73,684||2.10||4|| |
| 85+||20,539||54,157||1.80||7|| |
| Total|| || ||11.32||25||2.21 (1.43–3.27)|
|Entire cohort|| || ||26.18||57||2.18 (1.69–2.85)|
Among patients with ampullary cancer who developed colorectal cancer as a second primary malignancy, the median time from diagnosis of the first primary malignancy to diagnosis of the second primary malignancy was 5.25 years (25th–75th percentile, 2.42–10.25 years). Similarly, among patients with colorectal cancer who developed ampullary cancer as a second primary malignancy, the median time between diagnoses was 7.58 years (25th–75th percentile, 2.33–13.3 years).
Patients age ≥ 50 years who previously were diagnosed with colorectal cancer were found to have a relative risk of approximately 2 for the development of ampullary cancer as a second primary malignancy; similar results were observed for the development of colorectal cancer as a second primary malignancy in patients with ampullary cancer. It should be noted that the relative risk of developing colon cancer is 2.4 when a first-degree relative has been affected with a large bowel malignancy, 1.9 when a first-degree relative has had an adenomatous polyp, and approximately 1.5 when a second-degree relative has had malignant disease of the colon.27 In comparison, the risk of developing periampullary cancer is increased by more than 100 times for patients with FAP, among whom the prevalence rate of periampullary malignancy is 1–12%.28
In a previous study in which similar analytic techniques were used, Su et al.15 reported a significantly increased risk of bile duct malignancy following colorectal cancer in both men and women. Only women were found to have an increased risk of developing colorectal cancer after being diagnosed with bile duct malignancy. To our knowledge, the current study is the first to analyze ampullary cancer alone, separate from malignancies of the extrahepatic biliary tree; this feature probably accounts for the stronger bidirectional association between ampullary cancer and colorectal cancer for both genders. In addition, the current study followed the SEER population to the end of 1999 and thus had a longer cumulative observational follow-up period compared with the study conducted by Su et al.,15 who analyzed the SEER data up to 1993.
Patients with malignant disease are more likely than the average person to develop a subsequent malignancy, because of genetic susceptibility or a common etiologic agent, or as a consequence of treatment.29 The statistical methods used to investigate multiple primary malignant neoplasms in large populations, such as the SEER cohort, are well established.30 The biologic plausibility of a significant association between a pair of primary malignancies is better established if the association is bidirectional. In examining any two primary malignancies (A and B), two relevant statistical parameters must be evaluated: the SIR of A following B (SIRA/B) and the SIR of B following A (SIRB/A). Mathematical modeling has demonstrated that under relatively general assumptions regarding the number of common risk factors, the prevalence of these factors, and the interaction (synergism) between them, the two SIRs should be nearly equal (i.e., SIRA/B ≈ SIRB/A), provided that the lifetime risk profiles of the individuals in the study do not change.31 Thus, because the two SIRs in the current study were nearly equal, it can be inferred that the risk factor(s) common to both ampullary and colorectal malignancies are likely to be acquired early in life. Therefore, risk factors such as the carcinogenic effects of the therapy used to treat the first primary malignancy and short-term changes in lifestyle (e.g., cessation of smoking, diet modification, etc.) are unlikely to explain the association between the two primary malignancies. Nonetheless, it should be noted that in this particular mathematical model, the significance of the equivalence of SIRs was developed under the assumption that chronic smoking is a powerful common risk factor for all malignancies studied, and thus the conclusions of the model may not be entirely applicable to the current study.
One possible source of bias in the current study population is increased awareness of and exposure to screening. In particular, after a diagnosis of ampullary cancer, individuals may be more likely to be screened for colorectal cancer, resulting in increased detection of colon cancer among these individuals relative to the general population. This limitation can be addressed by the exclusion of patients with second primary malignancies that were detected within 1 year of diagnosis of the initial primary malignancy. In the current analysis, when such patients were excluded, the calculated SIRs remained greater than 1, indicating that the increased incidence of second primary malignancies in the study population was not entirely attributable to a possible screening bias. In addition, in the current analysis, 36% of patients who developed colorectal cancer as a second primary malignancy had localized disease. Because this figure is comparable to the rate of localized disease among all patients with colorectal cancer in the SEER population, and given that screening is expected to be associated with localized disease, it is unlikely that increased screening alone can account for the increased relative risk in the current study population. It is more likely that shared genetic factors or environmental factors introduced early in life are responsible for the observed bidirectional association. Patients with second primary malignancies in the current study had a longer follow-up period than did patients with single primary malignancies (either ampullary or colorectal); thus, the possibility that differential follow-up may partially account for our findings cannot be excluded.
It is clear that the current study has significant clinical implications for the surveillance of patients with primary ampullary or colorectal cancer. Colorectal cancer screening should be strongly considered for individuals with ampullary cancer, because these individuals are at an elevated risk. In contrast, it is more difficult to determine whether patients with colorectal cancer should be screened endoscopically for ampullary abnormalities, particularly because the incidence of ampullary cancer is rather low and because focused examination of the ampulla requires side-viewing endoscopy, a specialized technique. Clinical trials are unlikely to resolve this issue, and therefore, decision analysis techniques should be applied to determine whether such a strategy would be cost-effective in a defined subset of patients with the highest risk of developing ampullary cancer. Additional studies should be performed to assess clinical predictors, such as family history of either type of malignancy, that may help to distinguish patients with colon cancer who subsequently develop primary ampullary cancer from those who do not.
In any analysis of the association between two primary malignancies that is based on limited numbers of observed second primary malignancies in a large population-based cohort, there are at least two potential sources of error: missing data and misclassified data.20 The SEER database has several recognized advantages, such as population-based case identification with case finding procedures (including review of medical and pathology records), rigorous data collection procedures, and quality control standards that ensure a patient follow-up rate of greater than 95%. In the SEER Program, even registered patients who move to a geographic location that is not included in a SEER registry are followed actively for their mortality status through multiple sources. Thus, it reasonably can be assumed that the data on person-years of follow-up and expected number of second primary malignancies in the current analysis are fairly accurate. It should be noted, however, that the SEER Program does not routinely record the occurrence of second primary malignancies in registered patients who have moved from SEER areas to non-SEER areas. It therefore is possible that the observed number of second primary malignancies, and consequently the observed relative risk as measured by the SIR, is an underestimate. Misclassification of second primary malignancies is possible but unlikely, due to the rigorous case finding and coding procedures incorporated into the SEER database. Also, because we included only patients with histologically confirmed second primary malignancies in the current analysis, the possibility of misclassification of data is diminished. Ampullary cancer often is confused with malignancies arising from other periampullary structures, such as the distal biliary tree and the head of the pancreas. We specifically excluded such malignancies by using ICD-O-2 code-based exclusion criteria. Although we attempted to exclude patients with classic FAP from the current analysis, attenuated forms of FAP cannot be excluded entirely.
In summary, we observed that patients with ampullary cancer have an increased risk of developing a subsequent primary colorectal cancer and that patients with colorectal cancer have an increased risk of developing a subsequent primary ampullary cancer. These findings suggest that ampullary cancer and colorectal cancer may share common environmental and/or genetic risk factors. Given the benefits of colorectal cancer screening, screening colonoscopy should be recommended to patients who have received curative therapy for ampullary cancer.