Abstract
Gallstones are common disorders associated with several cardiovascular risk factors. Gallstone formation and atherosclerosis may share key pathways, but studies on putative associations between gallstones and the risk of cardiovascular disease are sparse and non-conclusive. We studied the relationship between gallstones and the risk of subsequent cardiovascular diseases in the German arm of the European Prospective Investigation into Cancer and Nutrition (EPIC).
The study comprises 46,468 participants from EPIC-Potsdam and EPIC-Heidelberg aged 35-65 years, free of cardiovascular diseases and diabetes at baseline. Information about the gallstone status at baseline was ascertained via questionnaires. For all incident cases of myocardial infarction and stroke confirmation was obtained from the treating physician. Relative risks were estimated using Cox proportional hazards regression.
During eight years of follow-up, 919 participants suffered a stroke or myocardial infarction. After multivariable adjustment for established risk factors, subjects with reported gallstones (n = 4828) had an increased risk of cardiovascular diseases (hazard rate ratio (HR) = 1.24, 95% confidence interval (CI): 1.02, 1.50). In individuals, who underwent a cholecystectomy before baseline a 1.32-fold increase in risk was observed (95%CI: 1.05, 1.65). HRs differed depending on the presence of selected established risk factors (e.g. HR for cardiovascular diseases regarding gallstones in smokers = 1.66, 95%CI: 1.20, 2.30, and non-smokers = 1.09, 95%CI: 0.86, 1.38).
Our results indicate an increased cardiovascular risk for gallstone formers, which cannot be counteracted by gallbladder removal and opens up perspectives for individualized prevention strategies.
Introduction
Gallstones and cardiovascular diseases are common diseases worldwide, especially in Western countries.1–3 The available epidemiologic evidence indicates a positive association between a number of cardiovascular risk factors and the development of gallstone disease. In particular, a higher presence of biliary calculi has been reported in subjects with obesity, hypertension, hyperlipidemia, diabetes mellitus and carotid atherosclerosis.4–9 A relationship between the metabolic syndrome and stone diseases has been documented and gallstones are discussed as one part of this multifactorial disorder.10,11 In a previous study we observed that gallstones predicted the risk of developing diabetes independently from other established cardiovascular risk factors.12 Increasing evidence from rodent studies suggests that pathways contributing to gallstone development may intersect with regulatory pathways for atherosclerosis and hence play a role in cardiovascular diseases.13–16 In particular, Farnesoid X receptor, which is an essential regulator of bile salt synthesis, supposed to suppress both gallstone formation13 and the onset of atherosclerosis.14–16 Presumably, this receptor induces a rise in phospholipids and biliary bile salts leading to enhanced lipid metabolism and a restored solubility of cholesterol in bile.
Although first reports of a possible link between gallstones and the risk of coronary heart disease go back to the middle of the last century,17 this area is still sparsely explored. Evidence is limited to four American studies,18–21 two of which being retrospective in nature,18,19 and one of them being a small case-control study restricted to women.19 Furthermore, only one study provided results on the risk of stroke.19 Interestingly, positive associations between gallstones and cardiovascular diseases seem to attenuate or even reverse for individuals undergoing gallbladder removal.18,21
Altogether, the current evidence on a putative relationship between gallstone formation and the development of cardiovascular diseases is equivocal. Therefore, we examined the association between gallstones and the risk of cardiovascular diseases, based on two prospective German cohort studies, providing data on both myocardial infarction and stroke. In addition, we analysed the risk of cardiovascular diseases specifically in subjects with and without gallbladder removal.
Materials and methods
Study population
The European Prospective Investigation into Cancer and Nutrition (EPIC) is a large-scale multi-center cohort study with a total of nearly 520,000 participants from 10 European countries.22 EPIC-Germany comprises two cohorts of initially 35-65 year old subjects at centres in Heidelberg and Potsdam. Using general population registries, 25,440 participants (11,928 men and 13,612 women) were recruited in Heidelberg and 27,548 participants (16,644 men and 10,904 women) were recruited in Potsdam between 1994 and 1998.23 Physical examination at recruitment included standardized measurements of blood pressure and anthropometry. Furthermore, information on diet, lifestyle and prevalent comorbidities was assessed using self-administered questionnaires and computer-guided interviews.23,24 All participants gave their written, informed consent and the study’s ethical approval was obtained from the Ethics Committee of Medical Society of the federal state of Brandenburg (Cottbus) and the medical faculty of the University of Heidelberg, respectively.
For the present analyses, we excluded 2041 subjects with unknown cardiovascular disease status after enrollment. Furthermore, 1436 individuals with prevalent cardiovascular diseases at baseline were not considered for analyses. Because subjects with diabetes carry a cardiovascular risk equivalent to that of non-diabetics with a prior myocardial infarction,25 and diabetes is suspected to be an intermediate in the possible pathway between gallstones and cardiovascular diseases,12 2067 persons with a medical history of diabetes were excluded. Another 1045 participants had to be excluded as covariate data were missing. Thirteen subjects were not considered for analyses because information on gallstone status was not given at baseline. Finally, 46,486 participants entered the analyses and 919 cases of incident cardiovascular diseases were identified during a mean follow-up period of 8.2 ± 1.6 years. Of 24 individuals with both myocardial infarction and stroke, we considered only the first event, leaving 507 cases of myocardial infarction and 412 strokes for the analyses.
Ascertainment of myocardial infarction and stroke
The procedures of case ascertainment in the two cohorts have already been described in detail.26,27 Briefly, potential incident cases of cardiovascular diseases were identified either by self-report, death certificate or a link to the hospital information system of the major hospital in the area. Biennially, the subjects were asked about several newly diagnosed diseases via postal questionnaires. Reported cardiovascular outcomes were verified by reviews of medical records and death certificates or by contacting the patients' attending physician. The International Classification of Diseases, 10th revision (ICD-10), was used to code all verified incident cases of cardiovascular disease. Myocardial infarction was defined as ICD-10 I21, incident stroke as ICD-10 I60.0–I64.0. Only confirmed cases of myocardial infarction and stroke after baseline were considered as incident cardiovascular diseases.
Assessment of exposure and covariates
At baseline, computer-guided interviews were used to obtain information about certain prevalent diseases including gallstones. Thus, participants were asked if they had ever been diagnosed with gallstones, and if their gallbladder had been removed. Subjects had the options to answer with ‘yes’, ‘no’ or ‘don’t know’. Validated self-administered food frequency questionnaires were used to collect data on the participants’ habitual nutritional intake and alcohol consumption over the past 12 months.28 Possible answers were scaled from ‘once a month or less’ to ‘five times a day or more’ and were converted into grams/day and milliliters/day according to the German Nutrient Data Base.29 Lifestyle questionnaires and computer-guided interviews covered different non-dietary potential confounders (e.g. smoking habits, physical activity, and educational achievement). The physical examination at recruitment provided anthropometric data on waist circumference, weight and height, which was used to calculate individual body mass indices by dividing body weight (kg) by height (m) squared. All interviews and physical examinations, including blood pressure measurements, were standardized and accomplished by trained personnel.30 Prevalent hypertension and prevalent hyperlipidemia were defined by self-reporting of a confirmed diagnosis.
Statistical analysis
For statistical analyses SAS software package, release 9.2 (SAS Institute, NC, USA) was used. The baseline characteristics were adjusted for age and sex and compared between subjects with and without prevalent gallstone disease using analysis of covariance.
In order to examine the association between present gallstone formations and the risk of cardiovascular disease development, we calculated hazard rate ratios (HR) using Cox proportional hazards regression. Age was the underlying time variable with the subject’s age at recruitment as start-time and stop-time at the age of cardiovascular disease diagnosis or censoring, respectively. We examined the association of overall gallstones, and both gallstones with and without performed cholecystectomy with risks of myocardial infarction, stroke, and total cardiovascular diseases (including both endpoints).
Competing risk analysis was performed to test whether the associations of gallstone disease with cardiovascular events differed between myocardial infarction and stroke, as described by Lunn and McNeil.31
We allowed the baseline function to vary by age at recruitment (one-year integers) and study centre. Two adjustments were performed: 1) a sex-adjusted model and 2) a multivariable adjusted model taking into account a priori selected potential confounders: educational achievement (‘none or primary school completed’ (reference), ‘technical or secondary school completed’, ‘longer education/university degree’, used as indicator variables), physical activity (continuously in hours/week), smoking habits (‘never smoker’ (reference), ‘past smoker’, ‘current smoker ≤15 units/day’, ‘current smoker >15 units/day’, used as indicator variables), alcohol intake (men: ‘0 grams/day’, ‘>0-12 grams/day’ (reference), ‘>12-24 grams/day’,‘>24 grams/day’; women: ‘0 grams/day ‘, ‘>0-6 grams/day’ (reference), ‘>6-12 grams/day’, ‘>12 grams/day’, used as indicator variables), body mass index (continuously in kg body weight/body height (m2)), waist circumference (continuously in cm), and prevalent diseases (hypertension and hyperlipidemia (present/absent)).
To examine the homogeneity within strata of various risk factors, we calculated stratum-specific risk estimates and tested for interaction by inclusion of cross product terms in the multivariable adjusted models. Given a significant interaction or strongly divergent risk estimates within strata, we performed endpoint-specific analyses across available categories of the respective risk factor. In these models a likelihood-ratio test was accomplished comparing models with and without interaction between the two factors in the multivariate model. For this purpose, product terms of indicator variables and gallstone status have been computed.
The validity of the proportional hazards assumption was explored by calculating Schoenfeld residuals and plotting them against time. Statistical tests were 2-sided and p-values below 0.05 were considered to indicate statistical significance. Because the variance of an interaction tends to be very large and lower statistical power in detection of such an effect can be assumed, a cut-off p-value of 0.1 was chosen.32
Results
At baseline, 10.4% of 46,486 study participants reported the presence of gallstones (6.7% men (n = 1279), 13.0% women (n = 3549)). The reported prevalence was about two times higher in East Germany (Potsdam: 8.8% males and 15.6% females) than in Southern Germany (Heidelberg: 4.8% males and 9.7% females). Furthermore, 66.5% of subjects with gallstones also reported cholecystectomy (54.0% of subjects with gallstones in Heidelberg and 72.5% in Potsdam). During an average follow-up period of 8.2 ± 1.6 years (1994-2008), 919 incident cases of cardiovascular diseases (264 myocardial infarctions and 188 strokes in Heidelberg and 243 and 224 in Potsdam) occurred. Incidence rates for cardiovascular diseases were slightly higher in Potsdam, but almost comparable to Heidelberg (2.53 cases per 1000 person years versus 2.32 cases per 1000 person years).
Baseline characteristics according to gallstone status and treatment by cholecystectomy are presented in Table 1. Individuals with gallstones were on average 4.5 years older and more likely to be female than persons without gallstones. Therefore, further characteristics are presented age- and sex-adjusted. Subsequently, participants with gallstones had a higher body mass index and waist circumference than individuals who were free of stones. Besides, gallstone formers were less physically active and less well educated as their stone-free counterparts. The proportion of prevalent hypertension and hyperlipidemia was higher in subjects having gallstones compared to gallstone free individuals. There was no substantial difference in smoking habits and nutritional and dietary intakes between the groups after adjustment for age and sex. Gallstone formers who underwent cholecystectomy were even more frequently women, tend to be slightly more overweight, consumed less alcoholic beverages and were less often diagnosed with hyperlipidemia than gallstone formers without surgery.
Baseline characteristicsa of 46,486 participants according to gallstone status, additionally subdivided into gallstones with and without cholecystectomy
| Characteristics | Gallstones | |||
|---|---|---|---|---|
| No | Yes | |||
| (n = 41,658) | (n = 4828) | Cholecystectomy | ||
| Yes (n = 3194) | No (n = 1634) | |||
| Ageb | 49.7 (8.4) | 54.2 (8.0) | 54.3 (8.0) | 54.1 (8.1) |
| Sex, % maleb | 42.7 | 26.5 | 24.1 | 31.1 |
| Body mass indexc | 25.9 (0.0) | 27.7 (0.1) | 27.8 (0.1) | 27.4 (0.1) |
| Waist circumference, cmd | ||||
| Men | 94.7 (0.1) | 97.3 (0.3) | 97.7 (0.4) | 96.6 (0.4) |
| Women | 79.8 (0.1) | 84.5 (0.2) | 84.6 (0.2) | 84.1 (0.3) |
| Sports, hours/week | 1.36 (0.0) | 1.08 (0.0) | 1.07 (0.0) | 1.10 (0.1) |
| Educational achievement, % | ||||
| None/ primary school | 22.4 | 25.7 | 26.0 | 25.0 |
| Technical/secondary school | 42.9 | 44.1 | 44.4 | 43.4 |
| Longer education/university | 34.7 | 30.2 | 29.6 | 31.7 |
| Smoking, % | ||||
| Never smoker | 46.6 | 47.6 | 47.3 | 48.1 |
| Former smoker | 32.5 | 32.4 | 32.8 | 31.5 |
| Smoker (≤15 cigarettes/d) | 12.9 | 12.1 | 12.1 | 12.2 |
| Smoker (>15 cigarettes/d) | 8.0 | 7.9 | 7.7 | 8.2 |
| Alcohol intake, g/d | 17.3 (0.1) | 14.3 (0.3) | 14.1 (0.3) | 15.0 (0.5) |
| Medical history, % | ||||
| Prevalent hypertension | 39.8 | 47.0 | 47.5 | 45.4 |
| Prevalent hyperlipidemia | 29.8 | 32.0 | 29.8 | 36.1 |
| Nutritione | ||||
| Total energy intake, Kcal/d | 2117 (3.4) | 2136 (10.1) | 2127 (12.3) | 2152 (17.0) |
| Total fat, g/d | 81.0 (0.1) | 81.4 (0.2) | 81.5 (0.3) | 81.1 (0.4) |
| Cholesterol, mg/d | 312 (0.4) | 313 (1.3) | 314 (1.5) | 313 (2.1) |
| Red meat, g/d | 31.0 (0.1) | 30.8 (0.3) | 30.6 (0.4) | 31.3 (0.6) |
| Processed meat, g/d | 61.6 (0.2) | 65.8 (0.6) | 66.5 (0.7) | 64.2 (1.0) |
| Fruit, g/d | 135 (0.5) | 140 (1.4) | 141 (1.7) | 137 (2.4) |
| Vegetables, g/d | 50.0 (0.2) | 49.8 (0.6) | 50.5 (0.7) | 48.5 (1.0) |
| Characteristics | Gallstones | |||
|---|---|---|---|---|
| No | Yes | |||
| (n = 41,658) | (n = 4828) | Cholecystectomy | ||
| Yes (n = 3194) | No (n = 1634) | |||
| Ageb | 49.7 (8.4) | 54.2 (8.0) | 54.3 (8.0) | 54.1 (8.1) |
| Sex, % maleb | 42.7 | 26.5 | 24.1 | 31.1 |
| Body mass indexc | 25.9 (0.0) | 27.7 (0.1) | 27.8 (0.1) | 27.4 (0.1) |
| Waist circumference, cmd | ||||
| Men | 94.7 (0.1) | 97.3 (0.3) | 97.7 (0.4) | 96.6 (0.4) |
| Women | 79.8 (0.1) | 84.5 (0.2) | 84.6 (0.2) | 84.1 (0.3) |
| Sports, hours/week | 1.36 (0.0) | 1.08 (0.0) | 1.07 (0.0) | 1.10 (0.1) |
| Educational achievement, % | ||||
| None/ primary school | 22.4 | 25.7 | 26.0 | 25.0 |
| Technical/secondary school | 42.9 | 44.1 | 44.4 | 43.4 |
| Longer education/university | 34.7 | 30.2 | 29.6 | 31.7 |
| Smoking, % | ||||
| Never smoker | 46.6 | 47.6 | 47.3 | 48.1 |
| Former smoker | 32.5 | 32.4 | 32.8 | 31.5 |
| Smoker (≤15 cigarettes/d) | 12.9 | 12.1 | 12.1 | 12.2 |
| Smoker (>15 cigarettes/d) | 8.0 | 7.9 | 7.7 | 8.2 |
| Alcohol intake, g/d | 17.3 (0.1) | 14.3 (0.3) | 14.1 (0.3) | 15.0 (0.5) |
| Medical history, % | ||||
| Prevalent hypertension | 39.8 | 47.0 | 47.5 | 45.4 |
| Prevalent hyperlipidemia | 29.8 | 32.0 | 29.8 | 36.1 |
| Nutritione | ||||
| Total energy intake, Kcal/d | 2117 (3.4) | 2136 (10.1) | 2127 (12.3) | 2152 (17.0) |
| Total fat, g/d | 81.0 (0.1) | 81.4 (0.2) | 81.5 (0.3) | 81.1 (0.4) |
| Cholesterol, mg/d | 312 (0.4) | 313 (1.3) | 314 (1.5) | 313 (2.1) |
| Red meat, g/d | 31.0 (0.1) | 30.8 (0.3) | 30.6 (0.4) | 31.3 (0.6) |
| Processed meat, g/d | 61.6 (0.2) | 65.8 (0.6) | 66.5 (0.7) | 64.2 (1.0) |
| Fruit, g/d | 135 (0.5) | 140 (1.4) | 141 (1.7) | 137 (2.4) |
| Vegetables, g/d | 50.0 (0.2) | 49.8 (0.6) | 50.5 (0.7) | 48.5 (1.0) |
Baseline characteristics are expressed as means (standard error) or percentages and adjusted for age and sex; bage and sex of participants are expressed as unadjusted mean (standard deviation) or percentage; cbody mass index = body weight (kg)/ body height (m2), dparticipants' waist circumference is expressed as age-adjusted mean (standard error), enutritional intakes are expressed as means (standard error), controlled for age, sex and total energy intake. Total energy intake and alcohol intake are adjusted for age and sex only.
Baseline characteristicsa of 46,486 participants according to gallstone status, additionally subdivided into gallstones with and without cholecystectomy
| Characteristics | Gallstones | |||
|---|---|---|---|---|
| No | Yes | |||
| (n = 41,658) | (n = 4828) | Cholecystectomy | ||
| Yes (n = 3194) | No (n = 1634) | |||
| Ageb | 49.7 (8.4) | 54.2 (8.0) | 54.3 (8.0) | 54.1 (8.1) |
| Sex, % maleb | 42.7 | 26.5 | 24.1 | 31.1 |
| Body mass indexc | 25.9 (0.0) | 27.7 (0.1) | 27.8 (0.1) | 27.4 (0.1) |
| Waist circumference, cmd | ||||
| Men | 94.7 (0.1) | 97.3 (0.3) | 97.7 (0.4) | 96.6 (0.4) |
| Women | 79.8 (0.1) | 84.5 (0.2) | 84.6 (0.2) | 84.1 (0.3) |
| Sports, hours/week | 1.36 (0.0) | 1.08 (0.0) | 1.07 (0.0) | 1.10 (0.1) |
| Educational achievement, % | ||||
| None/ primary school | 22.4 | 25.7 | 26.0 | 25.0 |
| Technical/secondary school | 42.9 | 44.1 | 44.4 | 43.4 |
| Longer education/university | 34.7 | 30.2 | 29.6 | 31.7 |
| Smoking, % | ||||
| Never smoker | 46.6 | 47.6 | 47.3 | 48.1 |
| Former smoker | 32.5 | 32.4 | 32.8 | 31.5 |
| Smoker (≤15 cigarettes/d) | 12.9 | 12.1 | 12.1 | 12.2 |
| Smoker (>15 cigarettes/d) | 8.0 | 7.9 | 7.7 | 8.2 |
| Alcohol intake, g/d | 17.3 (0.1) | 14.3 (0.3) | 14.1 (0.3) | 15.0 (0.5) |
| Medical history, % | ||||
| Prevalent hypertension | 39.8 | 47.0 | 47.5 | 45.4 |
| Prevalent hyperlipidemia | 29.8 | 32.0 | 29.8 | 36.1 |
| Nutritione | ||||
| Total energy intake, Kcal/d | 2117 (3.4) | 2136 (10.1) | 2127 (12.3) | 2152 (17.0) |
| Total fat, g/d | 81.0 (0.1) | 81.4 (0.2) | 81.5 (0.3) | 81.1 (0.4) |
| Cholesterol, mg/d | 312 (0.4) | 313 (1.3) | 314 (1.5) | 313 (2.1) |
| Red meat, g/d | 31.0 (0.1) | 30.8 (0.3) | 30.6 (0.4) | 31.3 (0.6) |
| Processed meat, g/d | 61.6 (0.2) | 65.8 (0.6) | 66.5 (0.7) | 64.2 (1.0) |
| Fruit, g/d | 135 (0.5) | 140 (1.4) | 141 (1.7) | 137 (2.4) |
| Vegetables, g/d | 50.0 (0.2) | 49.8 (0.6) | 50.5 (0.7) | 48.5 (1.0) |
| Characteristics | Gallstones | |||
|---|---|---|---|---|
| No | Yes | |||
| (n = 41,658) | (n = 4828) | Cholecystectomy | ||
| Yes (n = 3194) | No (n = 1634) | |||
| Ageb | 49.7 (8.4) | 54.2 (8.0) | 54.3 (8.0) | 54.1 (8.1) |
| Sex, % maleb | 42.7 | 26.5 | 24.1 | 31.1 |
| Body mass indexc | 25.9 (0.0) | 27.7 (0.1) | 27.8 (0.1) | 27.4 (0.1) |
| Waist circumference, cmd | ||||
| Men | 94.7 (0.1) | 97.3 (0.3) | 97.7 (0.4) | 96.6 (0.4) |
| Women | 79.8 (0.1) | 84.5 (0.2) | 84.6 (0.2) | 84.1 (0.3) |
| Sports, hours/week | 1.36 (0.0) | 1.08 (0.0) | 1.07 (0.0) | 1.10 (0.1) |
| Educational achievement, % | ||||
| None/ primary school | 22.4 | 25.7 | 26.0 | 25.0 |
| Technical/secondary school | 42.9 | 44.1 | 44.4 | 43.4 |
| Longer education/university | 34.7 | 30.2 | 29.6 | 31.7 |
| Smoking, % | ||||
| Never smoker | 46.6 | 47.6 | 47.3 | 48.1 |
| Former smoker | 32.5 | 32.4 | 32.8 | 31.5 |
| Smoker (≤15 cigarettes/d) | 12.9 | 12.1 | 12.1 | 12.2 |
| Smoker (>15 cigarettes/d) | 8.0 | 7.9 | 7.7 | 8.2 |
| Alcohol intake, g/d | 17.3 (0.1) | 14.3 (0.3) | 14.1 (0.3) | 15.0 (0.5) |
| Medical history, % | ||||
| Prevalent hypertension | 39.8 | 47.0 | 47.5 | 45.4 |
| Prevalent hyperlipidemia | 29.8 | 32.0 | 29.8 | 36.1 |
| Nutritione | ||||
| Total energy intake, Kcal/d | 2117 (3.4) | 2136 (10.1) | 2127 (12.3) | 2152 (17.0) |
| Total fat, g/d | 81.0 (0.1) | 81.4 (0.2) | 81.5 (0.3) | 81.1 (0.4) |
| Cholesterol, mg/d | 312 (0.4) | 313 (1.3) | 314 (1.5) | 313 (2.1) |
| Red meat, g/d | 31.0 (0.1) | 30.8 (0.3) | 30.6 (0.4) | 31.3 (0.6) |
| Processed meat, g/d | 61.6 (0.2) | 65.8 (0.6) | 66.5 (0.7) | 64.2 (1.0) |
| Fruit, g/d | 135 (0.5) | 140 (1.4) | 141 (1.7) | 137 (2.4) |
| Vegetables, g/d | 50.0 (0.2) | 49.8 (0.6) | 50.5 (0.7) | 48.5 (1.0) |
Baseline characteristics are expressed as means (standard error) or percentages and adjusted for age and sex; bage and sex of participants are expressed as unadjusted mean (standard deviation) or percentage; cbody mass index = body weight (kg)/ body height (m2), dparticipants' waist circumference is expressed as age-adjusted mean (standard error), enutritional intakes are expressed as means (standard error), controlled for age, sex and total energy intake. Total energy intake and alcohol intake are adjusted for age and sex only.
No decreasing or increasing trend in risk ratio was observed over time, indicating compliance with the proportional hazards assumption. Since the competing risks analyses showed no significantly different risk estimates for myocardial infarction and stroke (HR = 1.29, 95%CI: 0.99, 1.67 and HR = 1.19, 95%CI: 0.90, 1.58, respectively) with Wald test p-values equal to 0.70 for overall gallstones, equal to 0.33 for gallstones with cholecystectomy, and equal to 1.00 for gallstones without cholecystectomy, we provide further risk estimates for both cardiovascular endpoints combined. Moreover, similar multivariable risk estimates were observed in both cohorts (Potsdam: HR = 1.25, 95%CI: 0.98, 1.60, and Heidelberg: HR = 1.25, 95%CI: 0.92, 1.70); thus these estimates were pooled to facilitate the presentation of results.
Table 2 depicts the estimated HRs of cardiovascular diseases regarding the presence of gallstones, additionally divided according to gallstones with and without performed cholecystectomy. After controlling for age and sex only, a moderately increased risk of cardiovascular diseases was observed in subjects with gallstone disease. After additional adjustment for smoking status, alcohol consumption, educational achievement, physical activity, anthropometric data, and prevalent diseases the association was only slightly attenuated and remained significant (HR = 1.24, 95%CI: 1.02, 1.50). Further adjustment for dietary intakes also did not attenuate the results to non-significant (data not shown).
Multivariable adjusted hazard rate ratios of cardiovascular diseases regarding the presence of gallstones (n = 46,486), additionally subdivided into gallstones with and without cholecystectomy
| n | Cases (%) | Person years | Model 1a | Model 2b | |
|---|---|---|---|---|---|
| HR (95% CI) | HR (95% CI) | ||||
| No gallstones | 41,658 | 785 (1.9) | 340,085 | Reference | Reference |
| All gallstones | 4828 | 134 (2.8) | 39,267 | 1.34 (1.11, 1.62) | 1.24 (1.02, 1.50) |
| Gallstones with CCE | 3194 | 92 (2.9) | 26,059 | 1.41 (1.13, 1.76) | 1.32 (1.05, 1.65) |
| Gallstones without CCE | 1634 | 42 (2.6) | 13,208 | 1.20 (0.88, 1.64) | 1.09 (0.80, 1.50) |
| n | Cases (%) | Person years | Model 1a | Model 2b | |
|---|---|---|---|---|---|
| HR (95% CI) | HR (95% CI) | ||||
| No gallstones | 41,658 | 785 (1.9) | 340,085 | Reference | Reference |
| All gallstones | 4828 | 134 (2.8) | 39,267 | 1.34 (1.11, 1.62) | 1.24 (1.02, 1.50) |
| Gallstones with CCE | 3194 | 92 (2.9) | 26,059 | 1.41 (1.13, 1.76) | 1.32 (1.05, 1.65) |
| Gallstones without CCE | 1634 | 42 (2.6) | 13,208 | 1.20 (0.88, 1.64) | 1.09 (0.80, 1.50) |
CCE: cholecystectomy; CI: confidence interval; HR: hazard rate ratio
Adjusted for sex and stratified for age and study centre, b model 1, additionally adjusted for educational achievement, physical activity, smoking habits, alcohol intake, body mass index, waist circumference, and prevalent hypertension and hyperlipidemia.
Multivariable adjusted hazard rate ratios of cardiovascular diseases regarding the presence of gallstones (n = 46,486), additionally subdivided into gallstones with and without cholecystectomy
| n | Cases (%) | Person years | Model 1a | Model 2b | |
|---|---|---|---|---|---|
| HR (95% CI) | HR (95% CI) | ||||
| No gallstones | 41,658 | 785 (1.9) | 340,085 | Reference | Reference |
| All gallstones | 4828 | 134 (2.8) | 39,267 | 1.34 (1.11, 1.62) | 1.24 (1.02, 1.50) |
| Gallstones with CCE | 3194 | 92 (2.9) | 26,059 | 1.41 (1.13, 1.76) | 1.32 (1.05, 1.65) |
| Gallstones without CCE | 1634 | 42 (2.6) | 13,208 | 1.20 (0.88, 1.64) | 1.09 (0.80, 1.50) |
| n | Cases (%) | Person years | Model 1a | Model 2b | |
|---|---|---|---|---|---|
| HR (95% CI) | HR (95% CI) | ||||
| No gallstones | 41,658 | 785 (1.9) | 340,085 | Reference | Reference |
| All gallstones | 4828 | 134 (2.8) | 39,267 | 1.34 (1.11, 1.62) | 1.24 (1.02, 1.50) |
| Gallstones with CCE | 3194 | 92 (2.9) | 26,059 | 1.41 (1.13, 1.76) | 1.32 (1.05, 1.65) |
| Gallstones without CCE | 1634 | 42 (2.6) | 13,208 | 1.20 (0.88, 1.64) | 1.09 (0.80, 1.50) |
CCE: cholecystectomy; CI: confidence interval; HR: hazard rate ratio
Adjusted for sex and stratified for age and study centre, b model 1, additionally adjusted for educational achievement, physical activity, smoking habits, alcohol intake, body mass index, waist circumference, and prevalent hypertension and hyperlipidemia.
With regard to gallstone treatment, a 1.32-fold increased risk (95%CI: 1.05, 1.65) for developing cardiovascular diseases was observed among participants with gallbladder removal (comparison group = stone-free participants). In contrast, in individuals with prevalent gallstones but no performed cholecystectomy, the risk for cardiovascular diseases was not significantly higher than for participants free of stones (HR = 1.09, 95%CI: 0.80, 1.50). But in direct comparison, this difference in risk turned out to be non-significant (HR for cholecystectomy versus no cholecystectomy in participants with gallstones = 1.24, 95% CI: 0.85, 1.81; p = 0.27).
Figure 1 shows the HRs of cardiovascular diseases according to prevalent gallstones (including participants with and without cholecystectomy) in specific subgroups defined by the presence or absence of selected cardiovascular risk factors. Apparently, HRs varied between some strata, although no significant interaction was detected. Actually, the highest risk of cardiovascular diseases regarding gallstones was observed in current smokers (HR = 1.66, 95%CI: 1.20, 2.30). For this reason, an endpoint-specific sensitivity analysis was conducted with smoking used as categorical variable (see Materials and methods section).
Associations of gallstones and cardiovascular diseases stratified by selected risk factors.
Hazard rate ratios (HR) and 95% Confidence Intervals (CI) are adjusted for age, sex, educational achievement, physical activity, smoking habits, alcohol intake, body mass index (BMI), waist circumference, and prevalent hypertension and hyperlipidemia. Note: the HRs (y-axis) are plotted on a logarithmic scale; pint: p for interaction.
There was indication for interaction with smoking for myocardial infarction (pint = 0.07) but not for stroke (pint = 0.96). Compared to never smokers without gallstones, heavy smokers had an expected 4-fold risk of myocardial infarction, whereas the risk of myocardial infarction increased about 8-fold if heavy smokers had gallstones (see Figure 2).
Association of gallstones and myocardial infarction across categories of smoking.
Hazard rate ratios (HR) and 95% confidence intervals (CI) are adjusted for age, sex, educational achievement, physical activity, smoking habits, alcohol intake, body mass index (BMI), waist circumference, and prevalent hypertension and hyperlipidemia. Note: the HRs (y-axis) are plotted on a logarithmic scale; pint: p for interaction.
Discussion
The present study provides evidence for an independent association between prevalent gallstones and the development of cardiovascular diseases including both myocardial infarction and stroke. This is the first large prospective cohort study examining this association.
So far only the Walnut Creek Contraceptives Drug Study in 1979, a cohort of more than 15,000 women, examined gallstone disease and the risks of myocardial infarction and stroke separately.19 The authors of that study used a retrospective case-control-design, including a small number of cases (26 myocardial infarctions and 51 strokes divided into three subtypes) indicating the substantial uncertainty of the findings. They reported a significant increased risk of myocardial infarction (more than 6-fold), but they detected no significant associations with stroke subgroups due to gallstone disease.
Based on previous studies by Strom et al.18 and Ruhl et al.21 it was assumed that the removal of the gallbladder is associated with a lower risk of cardiovascular endpoints compared to gallstone formers without surgery. It was argued that the gallbladder removal stimulates an increased cholesterol excretion through the gastrointestinal tract and therefore might lead to an improvement of cardiovascular risk.18,33–36 Interestingly, our data do not support this theory. On the contrary, an increased risk of cardiovascular disease compared to individuals without stones was particularly observed in the two thirds of gallstone formers with cholecystectomy, but not in those without surgical gallstone treatment. Our results are in line with the findings from the Framingham Heart Study,20 where the investigators defined gallstone disease based on an undertaken cholecystectomy and a risk increase was seen in men. A higher risk of cardiovascular disease incidence due to gallbladder removal versus no removal could be explained by post-operative rises in hepatic triglyceride concentrations and thus a favored accumulation of fat in the liver.37 Moreover, cholecystectomy probably represents a more precise exposure confirmation with a smaller proportion of false-positives than purely self-reporting of gallstones. Gallbladder removal may also indicate a longer and more severe course of disease. Both could explain a decreased power in detecting a potentially independent association in gallstone carriers without cholecystectomy.
In a second step, we investigated the impact of important risk factors on the associations between gallstones and cardiovascular risk. Our results indicate a higher risk of cardiovascular diseases due to gallstones in smokers compared to non-smokers, particularly for myocardial infarction. Presumably, there is a joint multiplicative association between inflammatory mechanisms triggered by smoking,38 and metabolic constraints, as present in individuals with gallstones. However, to date no other study has reported a larger influence of gallstones on the risk of cardiovascular diseases among smokers. Furthermore, smoking is one of the strongest modifiable risk factors of myocardial infarction39 and residual confounding may to some extent explain the strong risk increase among smokers with gallstones. But since smoking habits are comparable between participants with and without gallstones this bias might be less influencing. Although tests for interaction with sex failed significance, risk estimates may suggest sex differences in risk. The weaker association of gallstones and cardiovascular diseases in women compared to men is in line with results from the Framingham Heart Study,20 where a 1.75-fold risk increase of coronary heart disease was only seen in men. In contrast, Petitti and colleagues19 observed significant risk-increasing associations between myocardial infarction and gallstone disease (odds ratio = 6.6, 95%CI: 3.5, 12.6), though the investigation was limited to American women. Our results indicate that the role of gallstones with regard to cardiovascular risk is less pronounced in women than in men.
Some limitations need to be discussed. First, it was not possible to distinguish between types of stones. However, since most gallstones are composed mainly of cholesterol40 and the association between pigment stones and cardiovascular diseases didn’t appear to be different,41 we expect only a minor impact on the results. Second, we had to rely on self-reported information on health conditions including the presence of gallstones, thus raising the potential for misclassification. However, the high socioeconomic status of participants of the EPIC-Germany study may be associated with a sufficient quality of self-reports. Moreover, our reported gallstone prevalence is comparable to that observed in another German survey.42 Furthermore, we expect possible misclassification to be non-differential, as both cases and non-cases may have had silent gallstones. An overestimation of the true effect could have been induced by a higher frequency of medical consultations and a higher probability of gallstone detection among individuals with less favorable health conditions. Nevertheless, this should not differ systematically between cases and non-cases after adjustment for cardiovascular comorbidities.
A particular strength of our study was its prospective design which allowed us to investigate time-dependent associations between gallstones and cardiovascular diseases. Second, by means of the large number of participants, we were able to detect even small associations, and by providing access to two independent cohorts, we had the opportunity to test the individual results for consistency and minimize results due to chance. In fact, the overall risk estimates of cardiovascular diseases due to gallstones were almost identical in both cohorts, despite the differences in gallstone prevalence. Third, numerous cardiovascular risk factors could be taken into account, including anthropometric data.
In our opinion, the diagnosis of gallstones is an optimal occasion to discuss lifestyle modifications with the patient. Gallstones may impact the etiology of cardiovascular diseases in different ways: by the increase in risk as an independent cardiovascular risk factor, and by the increase in risk of diabetes.12 Additionally, gallstone disease may serve as a marker of an unfavorable risk-profile and an early sign of a generally increased risk. Results from our stratified analyses could further be useful for the improvement of patient-oriented measures in health promotion. Subjects with gallstones may considerably reduce their risk of cardiovascular diseases, especially myocardial infarction, by the cessation of smoking, which goes far beyond the already known degree of improvement.
In summary, in the German population persons with gallstones were at increased risk of myocardial infarction and stroke. Our results further substantiate that subjects with gallbladder removal do not benefit from a lowered cardiovascular risk. Further, the strength of the association may be a function of the status of other risk factors. While the presence of gallstones and the habit of smoking seemed to affect each other synergistically, weaker associations were seen in women than in men, although women usually are more exposed to gallstone disease.
Acknowledgements
The authors thank all study participants for their cooperation. We furthermore are indebted to Wolfgang Fleischhauer and Marie-Luise Groß for case ascertainment, and to Ellen Kohlsdorf for data management.
Funding
The recruitment phase of the EPIC-Potsdam Study was supported by the Federal Ministry of Science, Germany (01 EA 9401), and the European Union (SOC 95201408 05F02). The follow-up was supported by the German Cancer Aid (70-2488-Ha I) and the European Community (SOC 98200769 05F02). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Conflict of interest
The authors declare that there is no conflict of interest.
Comments