Association of chili pepper consumption, low socioeconomic status and longstanding gallstones with gallbladder cancer in a Chilean population
We explored the risk factors for gallbladder cancer and explanations for its sharp and constant incidence increase in Chile since the 1970s. We compared 114 consecutive patients with verified gallbladder cancer, diagnosed 1992–1995, to 114 matched hospital patients with gallstones, using conditional logistic regression analysis. Low education showed a nonsignificant positive relationship with gallbladder cancer [odds ratio (OR) = 2.3, 95% confidence interval (CI) 0.8–6.2], and low socioeconomic level showed a significant relationship (OR = 5.0, 95% CI 1.5–17.3). A very long history of gallstone disease was significantly more prevalent among cases (OR = 11.0, 95% CI 1.4–85.2). Significant red chili pepper consumption was observed in gallbladder cancer patients (OR = 2.9, 95% CI 1.6–5.2). Low intake of both fresh fruit and sugar as soft drinks was associated with gallbladder cancer, with ORs of 6.4 (95% CI 1.4–30.3) and 3.6 (95% CI 1.3–10.1), respectively. Multivariate analysis kept only a very low socioeconomic status and red chili pepper consumption as significant independent risk factors for gallbladder cancer, ORs of 6.3 (95% CI 1.7–23.0) and 3.2 (95% CI 1.7–5.9). Longstanding gallstone cases were removed from the multivariate model because all were in the low and very low socioeconomic groups, reinforcing the association. Patients with gallbladder cancer differed from matched controls by exhibiting lower socioeconomic levels, having a much longer history of gallstone disease and presenting a dietary pattern characterized by high red chili pepper consumption and low fresh fruit intake. © 2002 Wiley-Liss, Inc.
Gallbladder cancer is uncommon in most countries and falling in incidence worldwide, with the relevant exceptions of Chile, Japan, Sweden and Finland.1, 2 The Chilean trend toward increasing crude mortality rates for gallbladder cancer has been sharp and uninterrupted over the last 3 decades, going from 3.7/100,000 for both sexes combined in 1970 to 11.7/100,000 in 1999.3 Age adjustment does not change this trend, and the Chilean mortality rate for this cancer ranks first in the world for the period 1981–1986 by far (for both men and women), above that of Japan and some European countries.4 Chile has also a very high and increasing proportion of gallbladder cancers (code 156.0 ICD-9 or C23 ICD-10) compared to choledocal, ampullar and unspecified sites (156.1, 156.2 and 156.9 ICD-9 or C24 ICD-10) and to other countries.5 Cancers originating in the gallbladder have increased from 79% to 94% in the last 3 decades in Chile, instead of the constant 50–60% shown in most countries.6
Several hypotheses have been suggested to explain the high prevalence and high gallbladder location of biliary tract cancers in Chile. There was a sharp and persistent decrease in cholecystectomy rates during the 1970s and the 1980s,7, 8, 9 as has also been shown for some developed countries.10, 11 Other postulated contributing factors are a longer life span, a possible increase in the prevalence of gallstones, deterioration of sanitary conditions in the 1980s with an increase in Salmonella typhi carriers and environmental carcinogens.8
So far, 15 different case-control studies on gallbladder cancer have been published (12 in English) in the world literature.12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29 Our study is the final report of a collaborative project between Japan and Chile, and we explored eventual carcinogenic factors in gallstone carriers, with special emphasis on food.
MATERIAL AND METHODS
Our hospital-based case-control study was carried out between January 1992 and March 1995 at the Sótero del Río Hospital. This is a teaching facility of 850 beds serving a population of 1.1 million people. The study population comprised 114 patients defined as cases after a surgically and histologically verified gallbladder cancer. Pathologic diagnosis was also verified at the First Department of Pathology, School of Medicine, University of Niigata, by one of the authors (H.W.). Of the patients with diagnosed gallbladder cancer, 96% were also gallstone carriers, as confirmed by ultrasonography, by surgery or at the moment of autopsy. The remaining patients had classical clinical features suggestive of gallstones. A control subject of the same age (3-year bands), gender and hospital was matched to each case. Eligible controls were contemporary cholecystectomized gallstone patients at the same hospital. Hence, cases and controls were matched not only for age and sex but also for gallstone disease with the aim of detecting additional risk factors other than the 3 that are generally accepted.
Hospital controls were selected because they were considered more likely than the general population, which includes healthy subjects, to be aware of antecedent exposures or events, reducing the potential for recall bias. Also, gallstone carriers should have suffered the same intangible selection factors that influenced gallbladder cancer patients to come to our hospital. The institution in which the study was carried out is committed to a specific population, with very similar social, economic and cultural backgrounds, minimizing bias in this matter.
All subjects were interviewed by trained medical students. All interviewers were blinded to the diagnosis on the pathology report as well as to each other's diagnosis. Special care was taken to ensure that the same student who interviewed a case also interviewed the respective matched control, controlling in this way for possible bias. The questionnaire included information about socioeconomic characteristics, family history, lifestyle habits (e.g., smoking, alcohol consumption, intestinal habit) and a problem-oriented medical history. A long-term dietary intake history of the subjects was investigated through an interviewer-administered food-frequency questionnaire (FFQ).
Medical and family background included information about infectious diseases in childhood and prevalence of gallstone disease and/or history of cancer in first-degree relatives. Exposure to agents presumed to be risk factors for gallbladder cancer was also explored. For women, additional questions were added to explore use of oral contraceptives, menstrual and reproductive factors and hormone-replacement therapy. Intestinal habit was classified as normal (daily evacuation), diarrhea (liquid evacuations at least once a day) or constipation (intestinal evacuation after 3 days or more with hardened stools). Socioeconomic status was assessed by the Graffar index, which has 5 categories from V (the highest) to I and considers education, occupation, household and neighborhood characteristics.30 Length of biliary colic was the time between the first episode in life and the current diagnosis, measured in years. Race was established by self-report of the interviewed person or relatives; thus, it is assumed that this factor was biased toward whites, given the Chilean culture.
The FFQ was aimed at detecting the usual diet over the past 15 years either before the diagnosis of cancer (cases) or the surgery (controls). Special emphasis was added to ensure that answers were related to long-term dietary intake; consequently, recent changes in diet were ignored. Typically, most epidemiologic studies estimate intake only over the past 12 months, which could be problematic, particularly when cases are interviewed several months after cancer diagnosis. Estimates of food intake were obtained from cases and controls themselves or from relatives. Subjects were asked to indicate the average frequency of consumption of 60 food items per month, week or day in standard home-portion sizes, following the Chilean guidelines given by the Ministry of Health.
Fat intake was also assessed. Lipid intake was estimated by taking into account not only the intake of fried foods but also the content of seasoning lipids (oils) in different dishes, the frequency of consumption and portion size of each dish, the oil used for cooking and individual fat-intake patterns. We analyzed the data in terms of daily intake (grams) of foods by multiplying the frequency of consumption by the weight of standard portion sizes.
Epiinfo software was used to examine descriptively all variables in the questionnaires. All statistical analyses were done using Stata package, version 7.0 (Stata Corporation, College Station, TX). Primary exposures of interest were fried foods, chili pepper and fresh fruit. Secondary exposures of interest included 60 kinds of vegetable. We used logistic regression to analyze dietary associations with gallbladder cancer. Univariate and conditional multivariate logistic regression analyses were also run, employing variables that were significant (p < 0.05) or of borderline significance. Frequencies were obtained for all variables, and cross-tabulations for each potential risk factor vs. case-control status were built. Variables were retained for further analysis if they had univariate p values < 0.05 (for dichotomous or multilevel variables).
The study population consisted of 114 patients with verified gallbladder cancer (cases) and 114 patients with verified gallstones by contemporary cholecystectomy at the same hospital and living in the same neighborhoods, characterized as predominantly low medium and low (gallstone controls). As expected, most cases were women, with a male:female ratio of 1:6, mainly middle-aged or older, though almost 20% were aged 49 years and less (Table I). This male:female ratio is common in Chilean surgical data but different from the national mortality ratio, which is about 1:3. Mean age in male cases (65. 8 years) was 4.8 years greater than in female cases.
Table I. Proportion of Study Subjects by Age and Type of Interview
|Gender|| || |
|Age (years)|| || |
After exploring potential risk factors for gallbladder cancer, socioeconomic status had statistically significant elevated adjusted odds ratios (ORs). Gallbladder cancer cases categorized as having a very low socioeconomic status had an OR of 5.0 [95% confidence interval (CI) 1.5–17.3] compared to those at higher levels. Few cases and controls were in the high socioeconomic categories, and no one was at the highest level. Low educational level was not significant but predominant among gallbladder cancer patients, with an OR of 2.3 (95% CI 0.8–6.2) for less educated people (1–3 schooling years) compared to those with no education. A gradient was observed with descending ORs of 1.8 (95% CI 0.8–4.2) and 0.8 (95% CI 0.3–2.0) for people with 4–6 schooling years and 7 or more, respectively (Table II).
Table II. Risk Factors for Gallbladder Cancer
|Education (years)|| || || || |
| 0 y||11.4||15.8||Ref|| |
|Socioeconomic status|| || || || |
| Very low||20.2||9.7||Ref|| |
|Length of biliary colic (years)1|| || || || |
| ≤24||84.5||95.5||Ref|| |
|Parity2|| || || || |
| ≤5 pregnancies||47.3||62.4||Ref|| |
| >5 pregnancies||52.7||37.6||1.5||0.8–2.7|
|Intestinal habit|| || || || |
| Diarrhea/normal||64.9||76.3||Ref|| |
|Obesity (BMI)|| || || || |
| Up to 24.9||46.5||41.2||Ref|| |
|Typhoid fever|| || || || |
| Negative self-report||90.3||92.5||Ref|| |
| Positive self-report||9.7||7.5||0.5||0.2–1.2|
|Sugar intake (as soft drinks)|| || || || |
| 10–15 g/day||35.1||44.7||Ref|| |
| 3–4 units/week||50.0||50.0||1.2||0.7–2.2|
| Rarely (<10 g/week)||14.9||5.3||3.6||1.3–10.1|
|Green chili pepper|| || || || |
| <20 g/day||41.2||61.4||Ref|| |
| >20 g/day||58.8||38.6||2.9||1.5–5.6|
|Red chili pepper|| || || || |
| <20 g/day||42.1||66.7||Ref|| |
| >20 g/day||57.9||33.3||2.9||1.6–5.2|
|Fried foods|| || || || |
| <200 g/day||69.3||84.2||Ref|| |
| >200 g/day||30.7||15.8||2.1||1.1–3.8|
|Fresh fruit|| || || || |
| Frequent consumption (2 or more fruits/day)||36.8||44.7||Ref|| |
| 3–4 Fruits per week||47.4||49.1||1.4||0.7–2.8|
| Rarely (less than 1 fruit/week)||15.8||6.2||6.4||1.4–30.3|
High and similar proportions of both cases and gallstone controls with symptomatic gallstone disease were observed (74% and 77%, respectively), but differences appeared when a comparison was done between groups on the length of history of biliary colic. Some 15.5% of gallbladder cancer cases with diagnosed biliary colic reported a gallstone disease diagnosis more than 24 years earlier compared to 4.5% of gallstone carriers with diagnosed biliary colic but without cancer, yielding an OR of 11.0 (95% CI 1.4–85.2), a wide range and a single cut point that may have maximized the observed statistical power. Grouping below 24 years did not yield significance and was not considered for further analysis (Table II).
Variables found in other studies to be risk factors were only nominally statistically significant for gallstone carriers developing gallbladder cancer in our study. Obesity, parity and sugar consumption as soft drinks were not associated with the odds of gallbladder cancer. Constipation was more frequent in cases but did not reach statistical significance. History of typhoid fever was a nonsignificant, protective factor (Table II).
High consumption of fried foods in gallbladder cancer patients had a weakly significant OR of 2.1 (95% CI 1.1–3.8). A difference was observed with low fresh fruit consumption, which appeared as a strong risk factor for gallbladder cancer. Logistic regression analysis showed that low consumption of fresh fruit was the single most important independent predictor for development of gallbladder cancer (OR = 6.4, 95% CI 1.4–30.3), a wide confidence interval for a random influence (Table II).
Red and green chili peppers conferred significantly higher risk in cases compared to gallstone carriers, with ORs of 2.9 (95% CI 1.5–5.6) and 2.9 (95% CI 1.6–5.2), respectively (Table II).
When a conditional multivariate logistic regression analysis was applied, considering those risk factors that were univariately significant, only very low socioeconomic status and red chili pepper consumption remained as risk factors (Table III).
Table III. Adjusted ORs in a Conditional Logistic Regression Multivariate Analysis (114 Cases and 114 Controls)1
|Low socioeconomic status||5.6||1.4–20.5||0.014|
|Red chili pepper consumption||2.5||1.2–5.2||0.016|
|Green chili pepper consumption||1.4||0.6–3.5||0.792|
|Fried foods (once or more/day)||1.4||0.7–2.8||0.921|
|Schooling (6 years or less)||1.0||0.9–1.1||0.994|
In the final model, which included only the 2 significant variables, the OR for low socioeconomic status was 6.3 (95% CI 1.7–23.0) and that for red chili pepper consumption was 3.2 (95% CI 1.7–5.9). Despite the significant influence of longstanding gallstones (OR = 11.0, 95% CI 1.4–85.0), this risk factor was removed from the final model because all cases with low and very low socioeconomics status were longstanding gallstones carriers; therefore, the longstanding gallstone OR tends to be endless.
Our finding of excessive chili pepper consumption as a risk factor had not been hypothesized, but it agrees (perhaps randomly) with another case-control study of a different digestive tract cancer that showed a positive relationship with chili pepper consumption.31 Experimental research has shown that capsaicin, an active ingredient of chili pepper, behaves as a carcinogen32 and a mutagen.33 A pathophysiologic explanation for gallbladder cancer could be the association of capsaicin to an increase of bile concentration by increasing the flow of water into the mucosa34 and, thus, the concentration of eventual carcinogens in the bile. Another possibility, not explored in our study, could be that fungus aggregated during the red chili pepper production.
The association of a longer period (>24 years) of symptomatic gallstone disease in cancer patients compared to symptomatic gallstone carriers suggests that gallbladder cancer could be facilitated by physical trauma from gallstones.35 Periods shorter than 24 years showed no significant relationship. Our finding should be related to the inverse relationship between cholecystectomy rates and gallbladder cancer mortality shown in Sweden,10 the United States, England and Wales, Scotland, Canada11 and Chile.8, 9
It is well known that not all gallstone carriers who develop gallbladder cancer are symptomatic. In our study, it appeared that 28% of gallbladder cancer patients were asymptomatic despite an advanced disease that had emerged as acute cholecystitis, cholangitis or a very short period of weight loss. This fact may be related to results showing a strong positive relationship between larger diameter and lasting gallstones with gallbladder cancer.11, 15
Poverty tended to be prevalent and more marked in cases. Education showed an even gradient with descending ORs of 1.8 and 0.8 for people with 4–6 years and 7 or more, respectively. Our finding fits with a Polish study where low education was so important among cases compared to controls that all rates were adjusted for this variable.14
Parity has appeared in the literature as a risk factor for gallbladder cancer;18, 19, 20, 23 but in our study, the OR for gallbladder cancer among women with more than 5 pregnancies was at most borderline, with a value of 1.5 (95% CI 0.8–2.7). Constipation was not significant either (OR = 1.8, 95% CI 1.0–3.2). Obesity was associated with gallbladder cancer in a case-control study in Bolivia;21 but in our study, both overweight and obesity were unrelated to cancer risk, with ORs of 0.8 (95% CI 0.4–1.4) and 0.9 (95% CI 0.4–1.8), respectively.
Many other variables or potential risk factors, mainly foods, were examined; but no differences were seen when cases were compared to controls matched by age, sex and gallstone disease. The only exception was fresh fruit, which appeared indirectly as protective for gallbladder cancer (OR = 6.4, 95% CI 1.4–30.3) when consumption was low, in agreement with a growing literature that shows antioxidants as protective for many cancers.36, 37 Sugar, measured by soft drink consumption, did not appear to be a risk factor for gallbladder cancer, in agreement with all but 1 previous case-control study.17 Of course, lifetime history of food consumption has to be regarded cautiously because it is influenced by memory bias, which is typical in retrospective studies.
Aymara ethnic background appeared recently as a powerful risk factor for gallbladder cancer.21 In our study, cases who defined themselves as of Indian (Mapuche) origin were slightly more frequent than matched controls, but numbers were too small (5 vs. 2 people) to achieve statistical significance. This was possibly due to biased self-reporting of ethnicity as “white”. Genetic studies have established that the Chilean population is composed of about 70% mestizos (admixture of whites and Indians), 25% Caucasians and only 5% pure Indians. Indigenous origin might, in our perspective, also mean poverty, delayed recognition of disease or lack of accessibility to medical care, reflecting environmental and not genetic differences.38 Furthermore, in the case of Chile, it seems difficult to explain the sharp and constant increase in gallbladder cancer incidence during the last 3 decades based on ethnic changes. Of course, there could be a host susceptibility of the Chilean population to environmental factors.
Typhoid fever is a risk factor for gallbladder cancer,39, 40 but in our analysis it appeared to be protective, though not statistically significant (OR = 0.5, 95% CI 0.2–1.2). A possible explanation is subclinical and undiagnosed typhoid fever cases.
Proxy respondents appeared in relatively high proportion among cases (51.8%) but infrequently in gallstone controls (11.4%), reflecting the poor condition of cancer patients at hospital discharge. Despite the facts that our proportion of direct respondents was higher than in other studies14 and that data obtained from proxy respondents have shown good correlation with past history of patients,41, 42 this is a limitation of our study and errors may exist in some of our findings.
In conclusion, our study, which reports on a relatively large number of cases, suggests that gallbladder cancer is related to longstanding gallstone disease, sometimes asymptomatic, and is more common in poor and less educated people. High consumption of red chili pepper was in our study a significant risk factor for gallbladder cancer. This does not necessarily mean that capsaicin, the active ingredient in chili pepper, is a risk factor; other factors associated with high chili pepper consumption may be important. One of these potential associations could be a high intake of fried foods, as observed in our study. Also, our data suggest that consumption of fresh fruit is a protective factor for gallbladder cancer. Further examination of every potential risk factor is warranted.
We thank Dr. A.K. Diehl (University of Texas, Health Science Center, San Antonio, TX) and Dr. S.I. Bangdiwala (University of North Carolina, Chapel Hill, NC) for careful review of the manuscript.