Vegetables, fruit and risk of non-gallstone-related acute pancreatitis: a population-based prospective cohort study
- 1Unit of Nutritional Epidemiology, Institute of Environmental Medicine, Karolinska Institutet, Stockholm, Sweden
- 2Unit of Upper Gastrointestinal Research, Department of Molecular Medicine and Surgery, Karolinska Institutet, Stockholm, Sweden
- Correspondence to Mr Viktor Oskarsson, Unit of Nutritional Epidemiology, Institute of Environmental Medicine, Karolinska Institutet, SE-171 76 Stockholm, Sweden;
Contributors Concept and design: all authors; acquisition of data: AW; statistical analysis: VO, NO; analysis and interpretation of data: all authors; drafting the manuscript: VO; critical revision: all authors; obtained funding: AW; study supervision: AW.
- Revised 6 April 2012
- Accepted 4 May 2012
- Published Online First 27 June 2012
Objective To examine the association of vegetable and fruit consumption with the risk of non-gallstone-related acute pancreatitis.
Design A population-based prospective cohort of 80 019 women and men, aged 46–84 years, completed a food-frequency questionnaire at baseline and was followed up for incidence of non-gallstone-related acute pancreatitis from 1 January 1998 to 31 December 2009. Participants were categorised into quintiles according to consumption of vegetables and consumption of fruit. Cox proportional hazards models were used to estimate RRs and 95% CIs.
Results In total, 320 incident cases (216 men and 104 women) with non-gallstone-related acute pancreatitis were identified during 12 years of follow-up (891 136 person-years). After adjustment for potential confounders, the authors observed a significant inverse linear dose–response association between vegetable consumption and risk of non-gallstone-related acute pancreatitis; every two additional servings per day were associated with 17% risk reduction (RR=0.83; 95% CI 0.70 to 0.98; p=0.03). Among participants consuming >1 drink of alcohol per day and among those with body mass index ≥25 kg/m2, the RR for the highest compared with the lowest quintile of vegetable consumption was 0.29 (95% CI 0.13 to 0.67) and 0.49 (95% CI 0.29 to 0.85), respectively. Fruit consumption was not significantly associated with the risk of non-gallstone-related acute pancreatitis; the RR comparing extreme quintiles of consumption was 1.20 (95% CI 0.81 to 1.78).
Conclusions Vegetable consumption, but not fruit consumption, may play a role in the prevention of non-gallstone-related acute pancreatitis.
Significance of this study
What is already known on this subject?
▸ Oxidative stress and reactive species are associated with the pathogenesis of acute pancreatitis.
▸ The effect and possible benefit of antioxidant-rich food on the risk of acute pancreatitis have been scarcely investigated.
What are the new findings?
▸ This large population-based prospective cohort study is the first cohort study to examine the association of vegetable and fruit consumption with the risk of non-gallstone-related acute pancreatitis.
▸ A significant inverse association between vegetable consumption and risk of non-gallstone-related acute pancreatitis was observed; the risk declined in a linear dose–response fashion for every additional serving per day.
▸ There was no significant association between fruit consumption and risk of non-gallstone-related acute pancreatitis.
How might it impact on clinical practice in the foreseeable future?
▸ Recommendation of increased vegetable consumption may be beneficial for the prevention of non-gallstone-related acute pancreatitis.
The pathogenesis of acute pancreatitis is not fully elucidated, but the initial step in the development of the disease seems to be an uncontrolled activation of the proteolytic enzyme trypsin in the pancreatic acinar cells.1 The resulting autodigestion of the pancreas leads to an inflammatory cascade in which reactive species are of major importance. Reactive oxygen and nitrogen species exert damage on the pancreas through direct effects on cellular components and by recruitment of inflammatory cells.2 In the course of an attack of acute pancreatitis, markers for superoxide dismutase (an endogenous antioxidant enzyme) are increased while levels of antioxidant molecules (eg, vitamin C) are decreased.3 ,4
Although the role of antioxidants in the clinical management of acute pancreatitis seems to be unclear,5 ,6 an existing imbalance in the antioxidant status due to dietary factors may facilitate development of acute pancreatitis by making the pancreas more sensitive to oxidative stress. Consumption of vegetables and fruit, which are rich in antioxidants, can hypothetically create an optimal redox balance and protect against the development of the disease. The association of vegetables and fruit with the risk of acute pancreatitis has been scarcely investigated. One previous case-control study observed an inverse association between fruit consumption and risk of acute pancreatitis.7
Using a large population-based prospective, the cohort of Swedish men (COSM) and women, we aimed to examine the association of vegetable and fruit consumption with the risk of non-gallstone-related acute pancreatitis. Furthermore, we evaluated whether alcohol consumption, overweight and smoking—factors associated with oxidative stress—may modify this association.
The study population consisted of participants in the Swedish Mammography Cohort (SMC) and the COSM. The SMC was established between 1987 and 1990 when all women residing in central Sweden (Västmanland and Uppsala counties) and born between 1914 and 1948 received a mailed questionnaire on diet, body size and educational level. An expanded questionnaire was sent to all surviving participants in the late fall of 1997 and included 350 items concerning diet, other lifestyle factors and medical history; 39 227 women (70% of eligible) completed the questionnaire. The COSM was initiated at the same time in 1997 when 48 850 men (49% of the source population) born between 1918 and 1952 and residing in Västmanland and Örebro counties in central Sweden answered an identical questionnaire, except for sex-specific questions, to the SMC questionnaire from 1997. The cohorts have been proven to represent the Swedish population well in terms of age distribution, educational level and relative body weight.8 ,9 Incidence rates of acute pancreatitis in 1998–2003 are also comparable; for example, the incidence rates per 100 000 individuals aged 60–69 years in Sweden10 and in our cohorts were 52.2 versus 53.2 for women and 66.1 versus 73.6 for men, respectively.
Eligible participants for the current study were women and men who completed the 1997 questionnaire. We excluded participants with erroneous or missing personal identity number and those with a diagnosis of cancer (except for non-melanoma skin cancer) or history of exocrine pancreatic disease before 1 January 1998. Further exclusions were made of participants who were diagnosed with pancreatic cancer during follow-up, those with implausible energy intake (>3 SDs from the natural logarithm transform mean in men and women, separately), and those with missing information on vegetable consumption or fruit consumption. After these exclusions, the study cohort consisted of 80 019 participants (44 103 men and 35 916 women) aged 46–84 years in 1998. The study was approved by the Regional Ethical Board at Karolinska Institutet, and informed consent was obtained from the participants.
Assessment of vegetable consumption and fruit consumption
The food frequency questionnaire used for the baseline dietary assessment in 1997 included 96 food items. On this questionnaire, participants were asked to report their average frequency of consumption of each item during the previous year. The eight predefined responses ranged from ‘never’ to ‘three or more times per day’. These frequency responses were converted to average consumption of each vegetable and fruit item (servings per day). The consumption of individual vegetables (n=12) and fruit (n=6) was then combined to obtain estimates of total vegetable consumption and total fruit consumption (table 1, footnotes). The validity of estimates of vegetable and fruit consumption has been investigated previously. The Spearman correlation coefficients between the food frequency questionnaire and four 1-week weighted diet records ranged from 0.4 to 0.6 for vegetable items and from 0.5 to 0.7 for fruit items (Wolk A, unpublished data).
Assessment of other variables
The questionnaire collected information on age, educational level, history of diabetes, smoking status, weight, height and alcohol consumption. We calculated the energy intake using food composition data from the Swedish National Food Administration.11
Case ascertainment and follow-up of the cohorts
Participants were followed from 1 January 1998 to 31 December 2009 through linkage to the Swedish Patient Register, the Swedish Death Register and the Swedish Cancer Register. The national coverage of the Patient Register has been nearly 100% since 1987.12 Each patient has a personal identity number and a record contains discharge diagnoses (one primary and up to seven secondary) and surgical codes assigned according to the International Classification of Diseases (ICD) and the NOMESCO Classification of Surgical Procedures, respectively.
Incident cases of acute pancreatitis (ICD-10: K85) and other exocrine pancreatic diseases (ICD-10: K86, K87) were identified. Only those with non-gallstone-related acute pancreatitis, without a previous history of exocrine pancreatic disease during follow-up, were classified as cases. In order to differentiate between the subgroups of acute pancreatitis, gallstone-related pancreatitis was ascribed to patients with biliary pancreatitis (ICD-10: K85.1) and to those with discharge diagnosis of cholelithiasis (ICD-10: K80) or surgical code for gallbladder surgery—cholecystectomy (JKA20, JAK21), endoscopic-retrograde cholangiography (JKE00, JKE02, JKE12, JKE18) or percutaneous cholecystectomy (JKB30)—within 3 months after the diagnosis of acute pancreatitis. The remaining patients with acute pancreatitis were classified as non-gallstone-related.
The Swedish Patient Register has been shown to have a high validity in general.13 In a recently published study, the specific validity of acute pancreatitis was also evaluated to be high; comparing the register to hospital charts the accuracy of the register was 98%.14 Regarding surgical procedures (excluding endoscopies and biopsies), the codes were incorrect in 2% and missing in 5.3% of the records.13
Each person accumulated follow-up time from 1 January 1998 to date of diagnosis of acute pancreatitis, date of diagnosis of other exocrine pancreatic diseases, date of death or 31 December 2009, whichever occurred first. Participants were categorised into quintiles according to consumption of vegetables and consumption of fruit. Cox proportional hazard models were used to estimate RRs and 95% CIs for each quintile of vegetable and fruit consumption compared with the lowest quintile. In the Cox model, we adjusted for potential confounders such as age (years) (≤50, 51–60, 61–70, >70), sex, educational level (less than high school, high school, university), smoking status (current, former, never), monthly alcohol consumption (grams) (≤100, 101–200, 201–400, >400), body mass index (BMI, kg/m2) (quartiles), history of diabetes (yes/no) and energy intake (kcal/day) (quartiles). Missing data for educational level (0.2%), smoking status (1.5%), alcohol consumption (2.4%) and BMI (3.5%) were modelled as separate categories. In order to calculate a p value for trend across the quintiles, participants were assigned the median value of consumption for each quintile and the variable was entered as a continuous term in the Cox model. The proportional hazard assumption was tested by regressing scaled Schoenfeld's residuals against survival time.15 No evidence of departure from this assumption was found. We next investigated the dose–response relation by flexibly modelling the quantitative exposure using restricted cubic splines with 3 knots at fixed percentiles of the distribution.16 A formal p value against the hypothesis of a linear-response model was obtained by testing the coefficients of the second spline transformation equal to zero.
Sensitivity analysis was performed to test whether the associations were similar for primary discharge diagnoses compared with primary and secondary discharge diagnoses of non-gallstone-related acute pancreatitis. We also analysed men and women separately to test whether the results varied by sex. Furthermore, we stratified our analyses by factors associated with oxidative stress—alcohol consumption17 (≤1 standard drink (≤13 g of alcohol) per day, >1 standard drink per day), BMI18 (<25 kg/m2, ≥25 kg/m2) and smoking status19 (never or ever)—and tested the statistical significance of the interactions with the likelihood ratio test.
Statistical analyses were performed using Stata, V.12.0 (StataCorp). A two-sided p value <0.05 was considered statistically significant. This manuscript follows the recommendations of the Strengthening the Reporting of Observational Studies initiative.20
During a mean follow-up of 11.1 years (891 136 person-years), 320 incident cases of non-gallstone-related acute pancreatitis were identified (281 primary discharge diagnoses and 39 secondary discharge diagnoses). Of these cases, 216 (67.5%) were men and 104 (32.5%) were women. The mean (± SD) servings of vegetables and fruit consumed per day were 2.6±1.8 and 1.7±1.3, respectively. The IQR for daily vegetable servings was 1.2–3.0 for men and 1.7–3.8 for women. For daily fruit servings, the IQR was 0.7–2.0 for men and 1.1–2.6 for women. The distribution of age-standardised baseline characteristics by vegetable consumption and fruit consumption is shown in table 1. In general, those with a low consumption of vegetables were men, current smokers and less likely to have a postsecondary education. Similar characteristics were seen for those in the lowest quintile of fruit consumption. However, they were also more likely to consume alcohol and to have a history of diabetes.
We observed a significant inverse association between vegetable consumption and risk of non-gallstone-related acute pancreatitis (table 2). After adjustment for age, sex and other potential confounders, the risk was reduced by 44% (RR=0.56; 95% CI 0.37 to 0.84; p for trend=0.01) for the highest compared with the lowest quintile of vegetable consumption. In a restricted cubic spline model, we found no evidence of departure from a linear dose–response relation between vegetable consumption and risk of non-gallstone-related acute pancreatitis (p for non-linearity=0.18) (figure 1). Every two additional servings per day of vegetables were associated with 17% risk reduction in the multivariable model (RR=0.83; 95% CI 0.70 to 0.98; p=0.03). The RR was similar when restricting the analysis to participants with primary discharge diagnosis. Compared with the lowest quintile of vegetable consumption, the RR of non-gallstone-related acute pancreatitis in the multivariable model was 0.49 (95% CI 0.32 to 0.76). Separate analyses of women (RR=0.49; 95% CI 0.25 to 0.96) and men (RR=0.59; 95% CI 0.35 to 1.00) also showed similar associations. There was no significant association between fruit consumption and risk of non-gallstone-related acute pancreatitis (table 2); the RR in the multivariable model for the highest compared with the lowest quintile of fruit consumption was 1.20 (95% CI 0.81 to 1.78; p for trend=0.43). The results did not vary substantially among those with primary discharge diagnosis (RR=1.24; 95% CI 0.81 to 1.89) or in separate analyses of women (RR=1.05; 95% CI 0.53 to 2.09) and men (RR=1.31; 95% CI 0.80 to 2.13). Associations between any specific fruit item and risk of non-gallstone-related acute pancreatitis were similar to those observed for total fruit consumption (data not shown).
Stratified analyses of vegetable consumption by alcohol consumption, BMI and smoking status—factors associated with oxidative stress—are shown in table 3. The observed inverse association between vegetables and the risk of non-gallstone-related acute pancreatitis seemed to be stronger among participants with alcohol consumption of >1 drink (>13 g of alcohol) per day and among those with high BMI. Comparing the highest with the lowest quintile of vegetable consumption, the risk in the multivariable model was decreased by 71% among those who consumed >1 drink (median=21.0 g of alcohol) per day and by 51% among overweight participants (BMI ≥25 kg/m2). When further stratified by alcohol consumption of >1.5 drink (median=27.9 g of alcohol) per day and by BMI ≥30 kg/m2, the RR comparing extreme quintiles was 0.17 (95% CI 0.05 to 0.52) and 0.30 (95% CI 0.07 to 1.27), respectively. The RRs were similar among never-smokers and ever-smokers. There was no interaction of vegetable consumption with alcohol drinking (p for interaction =0.90) or BMI (p for interaction=0.75) in predicting the risk of non-gallstone-related acute pancreatitis; nor was there an interaction with smoking status (p for interaction=0.52). We observed no association between fruit consumption and risk of non-gallstone-related acute pancreatitis in any strata of alcohol consumption, BMI or smoking status (data not shown).
In this prospective population-based cohort of Swedish men and women followed up for 12 years, there was a significant inverse linear dose–response association between vegetable consumption, but not fruit consumption, and risk of non-gallstone-related acute pancreatitis. The risk reduction seemed to be more pronounced among alcohol drinkers and among overweight participants.
Reactive oxygen and nitrogen species, although unable to induce the pathological process, are important in the pathogenesis of acute pancreatitis.2 Intake of dietary antioxidants can hypothetically create an optimal redox balance and diminish the effect of oxidative stress in the pancreas. The most likely mechanism whereby vegetables may protect against the development of non-gallstone-related acute pancreatitis is therefore the high content of antioxidants, for example, vitamin C and β-carotene. Of clinical interest are the observed associations when stratifying the analysis by alcohol consumption and BMI. The reduction in the risk of the disease seemed to be more pronounced among participants with a daily alcohol consumption of >1 drink as compared with those consuming less alcohol. Similarly, the inverse association appeared to be stronger among overweight and obese participants than among those with normal weight. Excessive alcohol consumption as well as overweight and obesity are factors associated with systemic oxidative stress.17 ,18 Alcohol has also been shown to increase oxidative stress in the pancreas.21 Thus, the inverse association between vegetables and the risk of non-gallstone-related acute pancreatitis might be stronger among those with underlying oxidative stress. Even though smoking is another factor associated with systemic oxidative stress and inflammatory changes in the pancreas,19 ,22 the RRs were similar when stratifying the analysis by smoking status. We acknowledge however the limited number of cases available in our study when assessing interactions and performing subgroup analyses.
Since both vegetables and fruit are rich in antioxidants, the lack of inverse association between fruit and risk of non-gallstone-related acute pancreatitis was unexpected. One case-control study from South Africa has previously observed an inverse association between fruit and risk of acute pancreatitis.7 However, that study was small (n=30 cases with acute pancreatitis) and with a non-quantitative assessment of dietary consumption (adequate vs inadequate). One possible explanation to our findings might be the content of fructose in fruit. Several experimental studies on animals have investigated the association between fructose and oxidative stress. Administration of fructose-rich diets to rats has led to an increase of oxidative markers, for example, indicators for lipid peroxidation and protein oxidation.23 In addition, as a sign of direct effect of fructose on the redox balance, treatment with antioxidants has decreased the generation of reactive oxygen species among fructose-fed rats.24 Alteration of the oxidative status by fructose might therefore counteract the potential beneficiary effects provided by antioxidants in fruit.
The recommended combined consumption of vegetables and fruit is five servings per day in most countries. The findings from our study suggest that a vegetable consumption in line with the general recommendation (the median value in the highest quintile of vegetable consumption was 4.6 servings per day) can be protective against the development of non-gallstone-related acute pancreatitis. Although our results for vegetables may indicate an important role of redox balance in the primary prevention of non-gallstone-related acute pancreatitis, the treatment with high doses of mixed antioxidants (eg, n-acetylcysteine, vitamin C and selenium) as a part of the clinical management of acute pancreatitis has not significantly decreased complications or mortality of the disease.5 ,6 However, antioxidants may under certain conditions, for example, in high doses, act as pro-oxidants.25 It is therefore theoretically possible that treatment with antioxidants may fail to be beneficial during an attack of acute pancreatitis.
The strengths of our study are the population-based prospective design, large sample size and nearly complete follow-up of the study population by linkage to different registers. The diagnosis of acute pancreatitis has also been shown to have a high validity in the Swedish Patient Register. An additional strength is the detailed information on diet and potential confounders. However, the present study also has some limitations. The use of self-administered food-frequency questionnaires is related to some misclassification of vegetable and fruit consumption, but such misclassification is expected to be non-differential with respect to the disease and should most likely underestimate any true association. Moreover, we cannot rule out the possibility that some of the participants with acute pancreatitis might have had chronic pancreatitis that clinically presented itself as acute pancreatitis. However, by excluding those with a previous history of exocrine pancreatic disease this possibility was reduced. Also, the definition of gallstone-related—and non-gallstone-related—acute pancreatitis is in the focus of an ongoing discussion.26 Finally, we could not adjust our analyses for the occurrence of other known risk factors of non-gallstone-related acute pancreatitis: for example, cystic fibrosis, hypercalcaemia, hypertriglycaemia and drug-induced pancreatitis.1 The potential residual confounding by these factors is however unlikely to fully explain the magnitude of the observed associations in this study. These comorbidities explain only a small amount of the cases with non-gallstone-related acute pancreatitis, and the prevalence of cystic fibrosis in the study cohort was 5 per 100 000 individuals at the start of follow-up.
In conclusion, results from this prospective cohort of men and women show that a high vegetable consumption may reduce the risk of non-gallstone-related acute pancreatitis. The inverse association seems to be stronger among participants consuming >1 drink of alcohol per day and among overweight participants. Our findings, if confirmed by other studies, indicate a potential benefit of increasing the consumption of vegetables for the prevention of non-gallstone-related acute pancreatitis.
Funding This work was supported by research grants from the Swedish Research Council/Committee for Infrastructure, the Swedish Cancer Foundation and the Board of Research at Karolinska Institutet (Distinguished Professor Award) (AW). Further support was received from the Board of Postgraduate Education at Karolinska Institutet (Clinical Scientist Training Program) (VO) and Olle Engkvist Byggmästare Foundation (OSA). The funding sources had no role in the design and conduct of the study; had no role in the collection, management, analysis or interpretation of the data; and had no role in the preparation, review or approval of the manuscript.
Competing interests None.
Patient consent Obtained written information and return of the completed questionnaire were considered to imply informed consent.
Ethics approval Ethics approval was provided by The Regional Ethical Board at Karolinska Institutet.
Provenance and peer review Not commissioned; externally peer reviewed.