Consumption of Cruciferous Vegetables and Glucosinolates in a Spanish Adult Population

Home , Brassicaceae, Cauliflower, Cruciferous vegetables, Glucosinolate

ORIGINAL ARTICLE Consumption of cruciferous vegetables and glucosinolates in a Spanish adult population

A Agudo1, R Iba´n˜ez1, P Amiano2, E Ardanaz3, A Barricarte3, A Berenguer1, M Dolores Chirlaque4, M Dorronsoro2, P Jakszyn1, N Larran˜aga2, C Martinez5, C Navarro4, G Pera1, JR Quiro´s6, MJ Sanche´z5, MJ Tormo4 and CA Gonza´lez1

1Unit of Epidemiology, IDIBELL, Catalan Institute of Oncology, L’Hospitalet de Llobregat, Spain; 2Direccioń de Salud de Guipu´zcoa, San Sebastiań, Spain; 3Instituto de Salud Pu´blica de Navarra, Pamplona, Spain; 4Departamento de Epidemiologıá, Conserjerıáde Sanidad, Consumo, Murcia, Spain; 5Escuela Andaluza de Salud Pu´blica, Granada, Spain and 6Consejerıá de Sanidad y Servicios Sociales de Asturias, Oviedo, Spain

Objective: To assess the intake of glucosinolates and cruciferous vegetables among Spanish adults. Design: Cross-sectional analysis of a prospective cohort study. Setting: The Spanish cohort of the European Prospective Investigation into Cancer and Nutrition (EPIC). Subjects: We analysed data from 40 684 men and women aged 35–64 years from the EPIC-Spain cohort. The usual diet was assessed by means of the dietary history method, and glucosinolate intake was calculated using a published food composition database. Results: The average intake of cruciferous vegetables was 11.3 g/day, accounting for about 5% of total vegetable consumption, whereas the daily intake of total glucosinolates was 6.5 mg, among which 35% were of indole type. The absolute intake of glucosinolates was in average higher in men than in women (6.8 vs 6.2 mg/day), whereas glucosinolate density per energy unit was higher in women’s diet (3.4 vs 2.7 mg/4200 kJ). Northern regions consumed in average 36% more glucosinolates than Southern regions (7.3 vs 5.4 mg/day). There was a positive association of glucosinolate intake with body mass index, physical activity, educational level and an inverse relationship with alcohol consumption. Conclusions: Contrary to the pattern seen for total vegetable intake, our estimate of consumption of cruciferous vegetables, and hence of glucosinolates, is relatively low within Europe, which in turn is lower than in North America and several Asian populations. European Journal of Clinical Nutrition (2008) 62, 324–331; doi:10.1038/sj.ejcn.1602750; published online 4 April 2007

Keywords: Cruciferous vegetables; glucosinolates; dietary intake; EPIC (European Prospective Investigation into Cancer and Nutrition).

Introduction

A recent comprehensive review (IARC, 2004) has determined Correspondence: Dr A Agudo, Unit of Epidemiology, IDIBELL, Catalan Institute that there is limited evidence that cruciferous vegetables of Oncology. Avda Gran Via s/n km 2.7, 08907 L’Hospitalet de Llobregat, Spain. reduce the risk of gastric and lung cancer, whereas it is not E-mail: [email protected] clear whether they reduce risk of tumours at other sites; Contributors: AA generated the idea for this topic, contributed to data analysis, furthermore, there seems to be an inverse association and wrote the draft of the paper. RI and PJ assisted AA in the analysis and between consumption of cruciferous vegetables and coron- production of the paper and carried out the literature search and retrieval. TB, GP and AA carried out the data managing and statistical analysis. AA, PA, EA, ary heart disease (Joshipura et al., 2001) and ischaemic stroke AB, MDC, MD, NL, CM, CN, PJ, JRQ, GP, MJS, MJT and CAG contributed to (Joshipura et al., 1999). Thus, intake of cruciferous vegetables the design of the study, data collection, data quality control and data analysis. is promoted, although as a part of a diet rich in a variety of All the contributors reviewed the draft, provided comments and gave approval fruits and vegetables, for cancer reduction and health to the final version of the article. Received 31 May 2006; revised 22 February 2007; accepted 2 March 2007; promotion. Cruciferous vegetables belong to the botanical published online 4 April 2007 family Brassicaceae, formerly known as Cruciferae owing to Consumption of cruciferous vegetables and glucosinolates A Agudo et al 325 the cruciform arrangement of their petals (Judd et al., 2002). cancer registries: three were from the north (Asturias, Despite the great diversity within the Brassicaceae, only Guipu´zcoa and Navarra) and two from the south (Granada members of a few genera are used in the human diet (Fahey and Murcia). Overall the participation rate varied among et al. 2001). The most commonly consumed cruciferous centres from 55 to 60%. At baseline, a blood sample, belong to the genus Brassica and many are within the several anthropometric measurements, a questionnaire on lifestyle varieties of Brassica oleracea, including cauliflower, cabbage factors and information on usual diet were obtained from and broccoli. each subject. The current analysis is restricted to 25 236 The protective effect of cruciferous vegetables is attributed women and 15 448 men aged from 35 to 64 years for to isothiocyanates (ITC) and indoles. Both groups of comparability reasons, as many surveys report data referred compounds have been shown to reduce the occurrence of to this age group; anyway, only 756 subjects (o2%) were cancer in experimental animals (IARC, 2004), and may have actually aged below 35 or above 64 in the original cohort. anticarcinogenic effects by several mechanisms (Verhoeven et al., 1997). ITC and indoles are both derived from degradation of glucosinolates by the action of myrosinase, Dietary information the plant thioglucosidase released after tissue disruption Information on usual intake over the previous year was owing to processing, cooking or eating vegetables (Mithen, collected by means of a computerized questionnaire based 2001); myrosinase-like activity has also been demonstrated on the diet history method validated previously (EPIC Group in the gut microflora, although glucosinolate hydrolysis is of Spain, 1997). It was structured by meals and recorded the much less efficient than the breakdown by the plant frequency of consumption of foods eaten at least twice a myrosinase (Getahun and Chung, 1999; Conaway et al., month, taking into account the seasonal variation; although 2000; Rouzaud et al., 2004). Glucosinolates may be found in the questionnaire was open, it contained a list of the most several plant families, but Brassicaceae is by far the largest common foods and recipes. In addition to frequency, the one, and thus the presence of glucosinolates is considered as method of preparation and amount consumed for each food a distinctive feature of cruciferous vegetables. Despite the item were reported; the portion size was assessed by means of great diversity of glucosinolates only a limited number are photo series, natural units and household measures. The commonly consumed within the human diet (Fahey et al., final amount consumed for each food was reported as daily 2001). intake in grams. In addition, a single 24-h dietary recall was Dietary intake of ITC and indoles is often assessed using collected for a subset of 3220 subjects randomly sampled in the consumption of cruciferous vegetables as a proxy. each centre to be used as a calibration study. The interviews However, crucifers contain other bioactive compounds with were balanced by days of the week and by seasons along 1 anticarcinogenic potential, such as vitamins and minerals. year. Detailed information on all foods and beverages As glucosinolates are the precursors of ITC and indoles, consumed during the 24-h period before the interview were quantification of glucosinolates provides an improvement in collected, entered and coded by means of a computer- the measurement of exposure to ITC and indoles and thus a assisted 24-h dietary recall (Slimani et al., 2002). Although better estimate of its association with cancer risk in both diet history and 24-h dietary recalls share drawbacks of epidemiological studies. methods based on subject’s memory, the latter seems to be We have already reported on fruit and vegetable intake in less prone to systematic errors; the principal limitation of 24- Spain (Agudo et al., 1999), showing a high consumption of h recall is that it does not provide a reliable estimate of vegetables as a whole, although intake of cabbage and other individual’s intake because of day-to-day variation, but this foods of the genus Brassica was relatively low. The aim of this may not significantly influence the mean (Nelson, 1991). study is to assess the amount and types of cruciferous Twenty-four hour recalls are thus appropriate for measuring vegetables as well as the intake of glucosinolates in the adult current diet in groups of subjects; in calibration studies, Spanish population, and its relationship with some demo- when the sub-population is representative of the whole graphic and lifestyle factors. cohort and weekly and seasonal variability is well balanced 24-h recall are used as a reference method and must provide a correct estimate of mean population intakes (Slimani et al., Methods 2002). The total intake of cruciferous (g/day) was obtained by Study population adding the consumption of all food items reported, either as The population is the Spanish cohort of the European single foods or ingredients of a recipe, classified as vegetables Prospective Investigation into Cancer and Nutrition (EPIC), belonging to the family Brassicaceae. We estimated the total a large prospective study in Europe devised to investigate the glucosinolate content for every food item classified as relationship between diet and cancer (Bingham and Riboli, cruciferous vegetable mentioned either in the diet history 2004). The EPIC-Spain cohort included 41440 healthy questionnaire or in the 24-h dietary recall using a food volunteers aged 29–69 years recruited between 1992 and composition database (McNaughton and Marks, 2003), 1996 in five regions in Spain covered by population-based which compilated published data from 18 studies. In

European Journal of Clinical Nutrition Consumption of cruciferous vegetables and glucosinolates A Agudo et al 326 addition to the glucosinolate content for each food item, it aged 35–64 years was carried out using the paired t-test was considered whether it applied to a fresh or frozen (Armitage and Berry, 1994). As most dietary variables were vegetable, or whether it was consumed as raw or cooked. The right-skewed, all the analyses were repeated with log- amount of indole-type glucosinolates for each cruciferous transformation of the variables, but the conclusions vegetable was estimated as a proportion of total glucosino- remained unchanged. lates (Kushad et al., 1999; Ciska et al., 2000); the remaining proportion was considered as aliphatic glucosinolates, although this term actually includes several side chain Results structures (Fahey et al., 2001). The values applied in our study for the estimation of glucosinolate intake are summar- In our population, four out of five subjects consumed ized in the Table 1. The total amount of glucosinolates, cruciferous vegetables regularly (at least twice a month); overall and by type (indole or aliphatic) was obtained by overall the average intake was 11.3 g/day (adjusted by the adding the glucosinolate content of all cruciferous vegeta- age–sex structure of the Spanish population 35–64 years), bles reported by the subject, expressed as mg/day. which represents slightly o5% of total vegetable consump- In addition to cruciferous vegetables and glucosinolates, tion (Table 2). Women consumed less cruciferous vegetables other dietary variables used in this analysis were daily energy than men, and consumption was markedly higher in intake (EI), alcohol consumption and total vegetable intake. Northern than in Southern regions, whereas there were no material differences in consumption by age. Regarding total glucosinolates (Table 3), the average intake was 6.5 mg/day Data analysis or 3.1 mg/day per 4200 kJ (1000 kcal); slightly more than one-third were indole-type glucosinolates (mean 2.3 The consumption of cruciferous vegetables and glucosino- mg/day), whereas the remaining were aliphatic. Men con- late was described by their corresponding mean, standard sumed more glucosinolates than women in average, but the deviation and median. In addition, glucosinolate consump- difference reversed when it was considered as nutrient tion was adjusted by EI using the nutrient density approach, density: glucosinolate intake per 4200 kJ (1000 kcal) was expressed as mg/4200 kJ (mg/1000 kcal). All the means higher in women than in men (3.4 vs 2.7 mg/day). As for reported were standardized by the age–sex structure of the cruciferous, Northern regions consumed in average 36% Spanish population aged 35–64 years (INE, 2005); further more total glucosinolates than Southern regions (7.3 vs adjustments, when relevant, are specified in the correspond- 5.4 mg/day), and this patterns holds as well for density ing tables. The estimations with 95% confidence intervals intake and for both types of glucosinolates (indole- and (CI), as well as comparisons of means, were carried out by aliphatic-type). analysis of variance, whereas association between contin- Among the 3172 subjects aged 35–64 years included in the uous variables was analysed by linear regression (Armitage calibration study, the mean intakes estimated from their 24- and Berry, 1994). Comparison of cruciferous and glucosino- h recalls were 12.3 g/day of cruciferous vegetables and late intakes estimated from the diet history and the 24-h 6.6 mg/day glucosinolates; the corresponding figures for dietary recall in the 3172 subjects from the calibration study the diet history were 11.9 and 6.8. Thus, taking the mean

Table 1 Food composition data for the glucosinolate content of cruciferous vegetables consumed in the EPIC-Spain cohort

Common name Scientific name Rawa (mg/ Range (mg/ Frozena (mg/ Cooked, % Indole 100 g) 100 g) 100 g) reduction typeb(%)

Cauliflower Brassica oleracea var. botrytis 43.2 11.7–78.6 40.5 32.3 45.7 Cabbage, white cabbage, green Brassica oleracea var. capitata 46.9 8.4–108.9 NA 27.8 30.6 cabbage Brussels sprouts Brassica oleracea var. 236.6 80.1–445.5 90.5 40.1 31.3 gemmifera Broccoli Brassica oleracea var. italica 61.7 19.3–127.5 50.7 39.1 14.8 Red cabbage Brassica oleracea var. rubra 64.2 26.5–76.6 NA 18.1 36.8 Turnip Brassica rapa 93.0 20.4–140.5 NA 48.0 28.3 Radish, red radish Raphanus sativus 67.6 44.8–76.8 NA NA 7.8

Abbreviation: NA, not available. Sources: Food composition data and influence of preservation and cooking in the glucosinolate content: McNaughton and Marks, 2003. Data on proportion of indole type: Kushad et al., 1999, and Ciska et al. 2000 (see references). aMedian of values from different studies (raw) as provided in the reference (McNaughton and Marks, 2003) except for cabbage and radish; in these two cases, the values correspond, respectively, to the average of the medians reported for white and unspecified cabbage, and the average of medians reported for red, white and European radish. For frozen, the three values reported correspond to estimates from one study for each food. bProportion of indole glucosinolates; the remaining percent corresponds to the overall class aliphatic, including mainly aromatic and oleofins.

European Journal of Clinical Nutrition Consumption of cruciferous vegetables and glucosinolates A Agudo et al 327 Table 2 Consumption of cruciferous vegetables in the population aged 35–64 years of the EPIC-Spain cohort

N Intake (g/day)a Proportion of consumersb Proportion of total vegetables

Mean (s.d.) Median

Total 40 684 11.3 (21.2) 3.89 80% 4.8%

Males 15 448 11.8 (22.2) 4.08 78% 4.8% Females 25 236 10.8 (20.1) 3.84 81% 4.8%

35–44 years 13 870 11.0 (21.1) 3.86 80% 4.6% 44–54 years 16 107 11.4 (21.8) 4.06 79% 4.8% 55–64 years 10 707 11.6 (20.6) 3.86 79% 5.1%

Southern centres 15 933 8.9 (19.8) 3.29 83% 2.9% Northern centres 24 751 12.9 (22.2) 4.88 78% 6.1%

Differences between means significant for sex and region (Po0.0001) and nonsignificant for age groups. aStandardized according to the age–sex structure of the Spanish population 35–64 years. bProportion of subjects reporting consumption of cruciferous vegetables different from zero in the diet history.

Table 3 Consumption of glucosinolates in the population aged 35–64 years of the EPIC-Spain cohort

N Intake of glucosinolates (mg/day)a Aliphatic type (mg/day)b Indole type (mg/day)

Mean (s.d.) Median Per 1000 kcal (mean) Mean Percent of Mean Percent of total glucosinolates total glucosinolates

Total 40 684 6.53 (11.9) 2.32 3.07 4.26 (65.2) 2.27 (34.8)

Male 15 448 6.83 (12.4) 2.41 2.70 4.46 (65.3) 2.37 (34.7) Female 25 236 6.24 (11.4) 2.24 3.43 4.06 (65.2) 2.17 (34.8)

35–44 years 13 870 6.46 (12.0) 2.28 2.96 4.20 (65.1) 2.26 (34.9) 44–54 years 16 107 6.62 (12.2) 2.41 3.08 4.32 (65.3) 2.30 (34.7) 55–64 years 10 707 6.54 (11.3) 2.28 3.21 4.27 (65.4) 2.26 (34.6)

Southern centres 15 933 5.38 (11.2) 1.95 2.51 3.53 (65.6) 1.85 (34.4) Northern centres 24 751 7.31 (12.4) 2.82 3.47 4.76 (65.1) 2.55 (34.9)

Differences between means significant for sex and region (Po0.0001) and no significant for age groups. aStandardized according to the age–sex structure of the Spanish population 35–64 years. bAliphatic type includes any non-indole type glucosinolates, mainly aromatic and oleofins.

of 24-h recalls as the reference method, the diet history tends Table 5 shows the relationship of average glucosinolate to underestimate the consumption of cruciferous vegetables intake with some anthropometric and lifestyle factors; by about 3% in average, whereas it overestimates the intake in addition to age, sex and centre, these estimates are of glucosinolates by approximately the same amount. In any further adjusted by total EI, and mutually adjusted by case, the differences between the average intakes estimated all the variables in the table. Average daily intake of by the two methods were not statistically significant glucosinolates is positively associated with body mass according to the paired t-test (P ¼ 0.72 for cruciferous index. In the same way, a positive association was vegetables and 0.71 for glucosinolates, respectively). observed with physical activity (both as leisure activity Cauliflower in first place and cabbage afterwards are the or practicing sports) and educational level; the latter is most consumed cruciferous vegetables in our population mainly due to the fact that subjects with the highest (Table 4). Both together accounts for 84% of total consump- level (university degree) consumed more than other educa- tion of cruciferous vegetables and 71% of total intake of tional groups. Former smokers consumed more glucosino- glucosinolates; cauliflower alone accounts for half the total lates than never and current smokers, and there is a clear intake of indole-type glucosinolates. Brussels sprouts ac- inverse trend between glucosinolate intake and alcohol count only for o4% of consumption of crucifers, whereas consumption; despite small absolute differences by cate- they provide the 13% of glucosinolate intake, reflecting the gories of alcohol consumption the trend was clearly high content of glucosinolates in such foods. significant.

European Journal of Clinical Nutrition Consumption of cruciferous vegetables and glucosinolates A Agudo et al 328 Table 4 Proportion of total consumption of cruciferous vegetables and glucosinolates contributed by each individual food classified as cruciferous among subjects from the EPIC-Spain cohort aged 35–64 years

Common name Scientific name Proportion of total consumption of:

Cruciferous Glucosinolates Glucosinolates, aliphatic Glucosinolates, indole vegetables (%) (total) (%) type (%) type (%)

Cauliflower Brassica oleracea var. 46.1 37.2 30.5 50.0 botrytis Cabbage (white, green, non- Brassica oleracea var. 37.9 33.3 34.9 30.0 specified type) capitata Brussels sprouts Brassica oleracea var. 3.6 13.1 13.6 12.1 gemmifera Broccoli Brassica oleracea var. 4.5 5.4 6.9 2.3 italica Red cabbage Brassica oleracea var. 0.3 0.4 0.4 0.5 rubra Turnip Brassica rapa 2.6 4.4 4.8 3.7 Radish (red radish) Raphanus sativus 5.0 6.3 8.8 1.4

Discussion explicit definition of cruciferous vegetables applied in each study. We have observed a daily intake of approximately 11 g of On the contrary, very few estimates of glucosinolate intake cruciferous vegetables and 6.5 mg of glucosinolates in an at the population level have been reported, probably because adult Spanish population aged 35–64 years. Compared with of lacking of food composition tables; in most cases, they are other European countries within the EPIC study (Agudo rough estimates based upon overall amount of cruciferous et al., 2002), Spain has the lowest consumption, quite close consumption and limited data on glucosinolate content in to Greece (13 g/day), and slightly lower than in Sweden, some foods. The average daily consumption of total Denmark and Italy, all of them with daily intakes below 20 g. glucosinolates in the UK was estimated to be about 50 mg France and the Netherlands consumed between 20 and 30 g (Wattenberg et al., 1986), whereas in the Netherlands, the per day, whereas Germany, Norway and the UK had daily estimated average intake of glucosinolates was 22 mg/day consumption above 30 g, almost three times the estimated (Jongen, 1996; Kistenmaker et al., 1998). As compared with consumption in Spain. This pattern is quite different from our average estimate of 6.5 mg, they are higher than what the one of overall vegetable intake, which showed a clear could be expected from the cruciferous vegetable consump- South–North gradient; this discrepancy is well appreciated tion in these countries, two to three times higher than the when consumption of cruciferous is expressed as a propor- intake reported in Spain. Other estimates available relate to tion of total vegetable intake: in countries with high indole glucosinolates (IARC, 2004): the daily intake (mg) of consumption of cruciferous vegetables, they account for glucobrasicin and neoglucobrasicin (the two main dietary more than 20%, whereas in countries with low consump- indole glucosinolates), was 22.5 in the UK, 5.5 in Denmark tion, they account only for 5–10% of total vegetable intake. and 2.8 in Finland. In the USA (Broadbent and Broadbent, Similar results were reported in a recent review of cruciferous 1998), the intake of indole glucosinolates was about 22.5 mg/ intake (IARC, 2004), showing a variable consumption in day per capita. Again, all these estimates are above our Europe, ranging from 5 to 30 g/day. In this review, the estimated daily intake of 2.3 mg for indole glucosinolates. An greatest consumption was reported in China, above 100 g/ important consideration when comparing intakes of gluco- day, whereas other Asian populations and Australia had sinolates across countries is the type of cruciferous vegetables relatively high daily intakes as well, ranging from 40 to 80 g. most commonly consumed, which may vary according to The average intake in North America was around 25–30 g per cultural and taste preferences. This can have important day, whereas relatively low daily intakes, of 15 g or less, are impacts on the actual intake of glucosinolates; for instance, reported for South Africa and some countries from South in our study, although Brussels sprouts accounted by o4% of America. As already seen, in countries with high intake of cruciferous vegetables consumption they contributed with crucifers, they constitute a relatively high proportion of total 13% to total glucosinolate intake; this discrepancy is vegetable intake. It must be considered that these estimates explained because Brussels sprouts are by far the richest come mainly from analytical studies, most of them case– source of glucosinolate among cruciferous vegetables con- control studies, although some intervention studies were sumed in our population. included as well; there was a remarkable diversity in methods Apart from between countries variation, the glucosinolate for dietary assessment used, and often there was not an consumption is also likely to vary according to socio-

European Journal of Clinical Nutrition Consumption of cruciferous vegetables and glucosinolates A Agudo et al 329 Table 5 Intake of glucosinolates (mg/day) in the population 35–64 month. Thus, in spite of the being statistically significant, years of the EPIC-Spain cohort according to selected anthropometric and these differences are not likely to be of great physiological lifestyle factors relevance. N (%) Meana 95% CI P-value There are several factors that may influence the exposure to ITC and indoles: the genetics of glucosinolate biosynthesis Body mass index within the plant (Kushad et al., 1999; Mithen, 2001), 20.00 kg/m2 271 0.7 5.55 (4.16–6.93) 0.0001 o o environmental factors, such as soil and climate conditions 20.00–24.99 kg/m2 8651 21.5 5.95 (5.68–6.22) 25.00–29.99 kg/m2 19 249 47.8 6.27 (6.11–6.44) (Ciska et al., 2000; Pereira et al., 2002; Vallejo et al., 2003), X30.00 kg/m2 12 107 30.1 7.05 (6.83–7.27) and storage, processing and cooking. Overall up to 50% of 2 Continuous (kg/m ) o0.0001 total glucosinolate content may be lost after cold storage (McNaughton and Marks, 2003), whereas chopping may lead Leisure activity p3.5 h/week 9050 22.5 5.62 (5.37–5.87) o0.0001 to remarkable increase in the level of indole glucosinolates 3.5–7.0 h/week 11 602 28.8 6.20 (5.98–6.42) (Verkerk et al., 1997). An important determinant of ITC 7.1–14.0 h/week 11 451 28.4 6.84 (6.62–7.05) exposure is the combined influence of cooking and micro- 414.0 h/week 8175 20.3 7.10 (6.84–7.36) flora in the intestinal tract: mild-cooking (o75–801C) results Continuous (h/week) o0.0001 in enhanced ITC production by releasing myrosinase, but Sports further cooking prevent further ITC formation, as myrosi- 0 h/week 30 258 75.2 6.29 (6.15–6.43) 0.0001 nase is quickly denaturated when temperature reaches 901C 0–3 h/week 6744 16.7 6.85 (6.57–7.14) (Dekker et al., 2002). However, ITC may still be formed after 43 h/week 3276 8.1 6.91 (6.51–7.32) Continuous (h/week) 0.008 eating different types of cruciferous vegetables cooked by different methods (boiled, steamed, microwaved), showing School level that glucosinolates may be converted to their breakdown None 13 904 34.5 6.26 (6.04–6.48) 0.0001 products by the myrosinase-like activity of colonic micro- Primary 15 797 39.2 6.34 (6.15–6.53) Secondary/technical 5930 14.7 6.49 (6.19–6.80) flora (Getahun and Chung, 1999; Conaway et al., 2000; University 4647 11.6 7.20 (6.86–7.55) Rouzaud et al., 2004). However, the hydrolysis of glucosino- Score, 1–4 o0.0001 lates by the microflora appears to be much less efficient than that catalysed by the plant myrosinase. On the other hand, Smoking status Never smoker 22 336 55.5 6.32 (6.14–6.50) 0.0004 several cruciferous vegetables, including broccoli and cab- Former smoker 7111 17.7 6.95 (6.67–7.23) bage, have been shown to form nitriles (without known Current smoker 10 831 26.9 6.33 (6.10–6.56) anticarcinogenic properties) as breakdown products of glucosinolates (Bones and Rossiter, 2006; Matusheski et al., Alcohol consumption Non-consumer 12 189 30.3 6.60 (6.36–6.84) 0.15 2006; Rungapamestry et al., 2006). These crucifers contain a o5.0 g/day 10 528 26.1 6.55 (6.31–6.80) protein called epithiospecifier protein (ESP) or ESP-like 5.0–14.9 g/day 5620 14.0 6.41 (6.11–6.72) protein, which appears to be responsible for the formation 15.0–29.9 g/day 4734 11.8 6.31 (5.97–6.64) of epithionitriles. ESP does not catalyse glucosinolates by X30.0 g/day 7207 17.9 6.07 (5.76–6.38) Continuous (g/day) 0.005 itself but acts as a cofactor of myrosinase, directing unstable products of glucosinolate hydrolysis toward epithionitriles Analysis carried out over 40 278 subjects with complete information in all the rather than ITC. ESP is more heat labile than myrosinase, variables included. and therefore mild and short heat treatment may lead to aStandardized according to the age–sex structure of the Spanish population aged 35–64 years, further adjusted by centre and energy intake, and mutually high production of ITC, as it inactivates ESP leaving some adjusted for all the variables in the table. active myrosinase (Bones and Rossiter, 2006). Several limitations must be considered when interpreting our estimates of cruciferous vegetables and glucosinolate economic and lifestyle factors within populations. We found intake. First, the EPIC-Spain cohort was not based on a that glucosinolate intake is higher among more educated representative sample of Spanish population. However, the people, those with higher physical activity, and former number of participants was large, the participation rate was smokers, and it tends to decrease with increasing alcohol relatively high, and subjects came from different social consumption. This is not surprising, as a similar pattern has backgrounds and differential geographical areas. The validity already been shown for consumption of vegetables (Agudo of glucosinolate intake is based upon the quality of two et al., 2002). However, although our study found statistically components: the assessment of consumption of cruciferous significant differences in the consumption of glucosinolates vegetables and the food composition database. Regarding given its huge sample size, the absolute differences are the latter (McNaughton and Marks, 2003), the authors actually quite small. For instance, the largest difference in applied quality criteria to include papers reporting glucosi- average daily intake of glucosinolate by individual lifestyle nolate content in foods, taking into account the detail of factors is 1.5 mg, which corresponds to approximately 20 g of information provided and the use of appropriate methods of cauliflower or cabbage per week, or about one serving per analysis. In addition to overall amount of glucosinolates,

European Journal of Clinical Nutrition Consumption of cruciferous vegetables and glucosinolates A Agudo et al 330 further data were considered to account for the influence of Centros de Cańcer (RTICCC, C03/10), the participating factors such as preservation and storage (frozen vs fresh) and Regional Governments, and the International Agency for cooking. Concerning the dietary assessment, we used a diet Research on Cancer (Agreement AEP/93/02). The participa- history method validated previously (EPIC Group of Spain, tion of RI was supported by a grant of the ‘Institut 1997). Furthermore, we also estimated the consumption of d’Investigacio´ Biome`dica de Bellvitge’ (IDIBELL). Some cruciferous vegetables and glucosinolate intake using the authors (AA, CAG) are members of ECNIS (Environmental 24-h dietary recalls from the calibration study, showing a Cancer Risk, Nutrition and Individual Susceptibility), difference of 3% in average as compared with the estimates a Network of Excellence of the 6th EU Framework Pro- from the diet history. The average intake from the calibration gramme (FP6, FOOD-CT-2005-513 943). study is supposed to be an accurate estimate of the mean consumption at the population level (Nelson, 1991). How- ever, in our calibration study under- or overreporting of EI was assessed by means of the ratio between EI and basal References metabolic rate; as compared with the reference ratio of 1.55, Agudo A, Amiano P, Barcos A, Barricarte A, Beguiristain JM, Spanish men tended to be overreporters (ratio 1.63), whereas Chirlaque MD et al. (1999). Dietary intake of vegetables and fruits women tended to be underreporters, with a ratio of 1.49 among adults in five regions of Spain. Eur J Clin Nutr 53, 174–180. (Ferrari et al., 2002). Agudo A, Slimani N, Ocke MC, Naska A, Miller AB, Kroke A et al. An alternative method to assess ITC exposure is the use of (2002). Consumption of vegetables, fruit and other plant foods in the European Prospective Investigation into Cancer and Nutrition biomarkers. Analytical methods based upon the cyclocon- (EPIC) cohorts from 10 European countries. Public Health Nutr 5, densation reaction of dithiocarbamates (DTC) with benzene- 1179–1196. dithiol (Zhang et al., 1996) enable the quantification of ITC- Armitage P, Berry G (1994). Statistical Methods in Medical Research,3rd edn. Oxford: Blackwell Scientific Publications. equivalents. As all ITC are mainly metabolized through Bingham S, Riboli E (2004). Diet and cancer – the European conjugation with glutathione to form DTC, the total urinary Prospective Investigation into Cancer and Nutrition. Nat Rev ITC measured by this method have been shown to be a Cancer 4, 206–215. reliable marker of dietary intake of ITC, and correlate Bones AM, Rossiter JT (2006). The enzymic and chemically induced decomposition of glucosinolates. Phytochemistry 67, 1053–1067. positively with the consumption of cruciferous vegetables Broadbent TA, Broadbent HS (1998). The chemistry and pharmacol- (Fowke et al., 2002). The cyclocondensation assay has been ogy of indole-3-carbinol(indole-3-methanol) and 3-(methoxy- also developed to quantify DTC in human plasma, serum methyl)indole. Part I. Curr Med Chem 5, 337–352. and erythrocytes (Ye et al., 2002). Ciska E, Martyniak-Przybyszewska B, Kozlowska H (2000). Content of glucosinolates in cruciferous vegetables grown at the same site for In conclusion, our estimates of consumption of crucifer- two years under different climatic conditions. J Agric Food Chem ous vegetables, and hence of glucosinolates, among Spanish 48, 2862–2867. adults are relatively low within Europe, and lower than in Conaway CC, Getahun SM, Liebes LL, Pusateri DJ, Topham DK, North America and several Asian populations. The usefulness Botero-Omary M et al. (2000). Disposition of glucosinolates and of assessing glucosinolate intake relies on the assumption sulforaphane in humans after ingestion of steamed and fresh broccoli. Nutr Cancer 38, 168–178. that the total glucosinolate content of cruciferous vegetables Dekker M, Verkerk R, Jongen WMF (2002). Predictive modelling of is related to the exposure to hydrolysis products with health aspects in the food production chain: a case study on anticarcinogenic potential ITC and indoles. However, to glucosinolates in cabbage. Trends Food Sci Technol 11, 174–181. EPIC Group of Spain (1997). Relative validity and reproducibility of a get a better estimate of the amount of ITC and indoles diet history questionnaire in Spain. I. Foods. Int J Epidemiol 26 available at the intestinal tract, it is important to know (Suppl 1), S91–S99. whether the crucifers were eaten raw or cooked, as well as Fahey JW, Zalcmann AT, Talalay P (2001). The chemical diversity and details about the cooking method, including the time and distribution of glucosinolates and isothiocyanates among plants. Phytochemistry 56, 5–51. the temperature. Such information should be included in Ferrari P, Slimani N, Ciampi A, Trichopoulou A, Naska A, Lauria C questionnaires aimed to assess the intake of ITC and indoles. et al. (2002). Evaluation of under- and overreporting of energy Another approach to overcome such limitations is the use of intake in the 24-hour diet recalls in the European Prospective biomarkers. Biological assays are available to measure ITC in Investigation into Cancer and Nutrition (EPIC). Public Health Nutr 5, 1329–1345. plasma and serum, and it would be desirable in the future to Fowke JH, Hebert JR, Fahey JW (2002). Urinary excretion of assess the relationship between the estimated glucosinolate dithiocarbamates and self-reported Cruciferous vegetable intake: intake and the ITC blood levels in our population. application of the ‘method of triads’ to a food-specific biomarker. Public Health Nutr 5, 791–799. Getahun SM, Chung FL (1999). Conversion of glucosinolates to isothiocyanates in humans after ingestion of cooked watercress. Acknowledgements Cancer Epidemiol Biomarkers Prev 8, 447–451. IARC (International Agency for Research on Cancer (2004). IARC The EPIC study received financial support from the European Handbooks of Cancer Prevention Volume 9 Cruciferous vegetables, Isothiocyanates and Indoles. IARC Press: Lyon. Commission (Agreement SO 97 200302 05F02), the Health INE (Instituto Nacional de Estadı´stica (INE) (2005). Poblacioń 2001 Research Fund (FIS) of the Spanish Ministry of Health (Exp. por provincias, edad y sexo. http://www.ine.es/inebase/index.html 96-0032), the Red Tema´tica de Investigacioń Cooperativa de (accessed April).

European Journal of Clinical Nutrition Consumption of cruciferous vegetables and glucosinolates A Agudo et al 331 Jongen WMF (1996). Glucosinolates in Brassica: occurrence Rouzaud G, Young SA, Duncan AJ (2004). Hydrolysis of glucosino- and significance as cancer-modulating agents. Proc Nutr Soc 55, lates to isothiocyanates after ingestion of raw or microwaved 433–446. cabbage by human volunteers. Cancer Epidemiol Biomarkers Prev 13, Joshipura KJ, Ascherio A, Manson JE, Stampfer JJ, Rimm EB, Speizer FE 125–131. et al. (1999). Fruit and vegetable intake in relation to risk of Rungapamestry V, Duncan AJ, Fuller Z, Ratcliffe B (2006). Changes in ischemic stroke. JAMA 282, 1233–1239. glucosinolate concentrations, myrosinase activity, and production Joshipura KJ, Hu FB, Manson JE, Stampfer JJ, Rimm EB, Speizer FE of metabolites of glucosinolates in cabbage (Brassica oleracea Var. et al. (2001). The effect of fruit and vegetable intake on risk for capitata) cooked for different durations. J Agric Food Chem 54, coronary heart disease. Ann Intern Med 134, 1106–1114. 7628–7634. Judd WS, Campbell CS, Kellogg EA, Stevens PF, Donoghue MJ (2002). Slimani N, Kaaks R, Ferrari P, Casagrande C, Clavel-Chapelon F, Lotze G Plant Systematics, A Phylogenetic Approach. MA: Sinauer Associates et al. (2002). European Prospective Investigation into Cancer and Inc.: Sunderland. Nutrition (EPIC) calibration study: rationale, design and population Kistenmaker C, Bouman M, Hulshof KFAM (1998). Food Consumption characteristics. Public Health Nutr 5,1125–1145. Standard 1997–1998, report no 98812 Zeist. TNO Nutrition and Vallejo F, Garcia-Viguera C, Tomas-Barberan FA (2003). Changes in Food Research: The Netherlands. broccoli (Brassica oleracea L. Var. italica) health-promoting com- Kushad MM, Brown AF, Kurilich AC, Juvik JA, Klein BP, Wallig MA pounds with inflorescence development. J Agric Food Chem 51, et al. (1999). Variation of glucosinolates in vegetable crops of 3776–3782. Brassica oleracea. J Agric Food Chem 47, 1541–1548. Verhoeven DT, Verhagen H, Goldbohm RA, van den Brandt PA, Matusheski NV, Swarup R, Juvik JA, Mithen R, Bennett M, Jeffery EH van Poppel G (1997). A review of mechanisms underlying (2006). Epithiospecifier protein from broccoli (Brassica oleracea anticarcinogenicity by brassica vegetables. Chem Biol Interact L. ssp. italica) inhibits formation of the anticancer agent 103, 79–129. sulforaphane. J Agric Food Chem 54, 2069–2076. Verkerk R, van der Gaag MS, Dekker M, Jongen WM (1997). Effects of McNaughton SA, Marks GC (2003). Development of a food processing conditions on glucosinolates in cruciferous vegetables. composition database for the estimation of dietary intakes of Cancer Lett 114, 193–194. glucosinolates, the biologically active constituents of cruciferous Wattenberg LW, Hanley AB, Barany G, Sparrins VL, Lam LKT, vegetables. Br J Nutr 90, 687–697. Fenwick GR (1986). Inhibition of carcinogenesis by some minor Mithen R (2001). Glucosinolates – biochemistry, genetics and dietary constituents. In: Hayashi Y (ed). Diet, Nutrition and Cancer, biological activity. Plant Growth Regulat 34, 91–103. pp 193–203. Japan Scientific Society Press: Tokyo. Nelson M (1991). Assessment of food composition and nutrient Ye L, Dinkova-Kostova AT, Wade KL, Zhang Y, Shapiro TA, Talalay P intake Past intake. In: Margetts BM and Nelson M (eds). Design (2002). Quantitative determination of dithiocarbamates in human Concepts in Nutritional Epidemiology. Oxford University Press: plasma, serum, erythrocytes and urine: pharmacokinetics of pp 167–191. broccoli sprout isothiocyanates in humans. Clin Chim Acta 316, Pereira FM, Rosa E, Fahey JW, Stephenson KK, Carvalho R, Aires A 43–53. (2002). Influence of temperature and ontogeny on the levels of Zhang Y, Wade KL, Prestera T, Talalay P (1996). Quantitative glucosinolates in broccoli (Brassica oleracea Var. italica) sprouts and determination of isothiocyanates, dithiocarbamates, carbon their effect on the induction of mammalian phase 2 enzymes. disulfide, and related thiocarbonyl compounds by cyclocondensa- J Agric Food Chem 50, 6239–6244. tion with 1,2-benzenedithiol. Anal Biochem 239, 160–167.

European Journal of Clinical Nutrition