Int. J . Cancer: 48, 350-354 (1991) 0 1991 Wiley-Liss, Inc.
Publication of the International Union Against Cancer Publication de I‘Union lnternationale Contre le Cancer
VEGETABLE AND FRUIT CONSUMPTION AND CANCER RISK Eva NEGRI’,~, Car10 LA VECCHIA’,~, Silvia FRANCEsCH13, Barbara D’AVANZO’and Fabio PARAZZINI’ ‘Istituto di Ricerche Farmacologiche “Mario Negri”, Via Eritrea 62, 20157 Milan, Italy; 21nstitute of Social and Preventive Medicine, University of Lausanne, 1005 Lausanne, Switzerland; and 3Aviano Cancer Centre, 33081 Aviano, Pordenone, Italy. The relationship between cancer r i s k and frequency of consumption of green vegetables and fruit has been analyzed using data from an integrated series of case-control studies conducted in northern Italy between 1983 and 1990. The overall dataset included the following histologically confirmed cancers: oral cavity and pharynx, I19; oesophagus, 294; stomach, 564; colon, 673; rectum, 406; liver, 258; gall-bladder, 41; pancreas, 303; larynx, 149; breast, 2,860; endometrium, 567; ovary, 742; prostate, 107; bladder, 365; kidney, 147; thyroid, 120; Hodgkin’s disease, 72; non-Hodgkin lymphomas, 173; myelomas, I 17; and a total of 6, I47 controls admitted t o hospital for acute non-neoplastic conditions, unrelated t o long-term dietary modifications. Multivariate relative risks (RR) for subsequent tertiles of vegetable and fruit consumption were derived after allowance for age, sex, area of residence, education and smoking. For vegetables, there was a consistent pattern of protection for all epithelial cancers, with R k in the upper tertile ranging from 0.2 for oesophagus, liver and larynx t o 0.7 for breast. All the trends in risk were in the same direction and significant for all carcinomas except gallbladder. In contrast, no protection was afforded by high vegetable consumption against non-epithelial lymphoid neoplasms. With reference t o fruit, strong inverse relationships were observed for cancers of the upper digestive and respiratory tract, with RRs in the upper tertile between 0.2 and 0.3 for oral cavity and pharynx, oesophagus and larynx relative t o the lowest tertile. The lower the location of the tumour in the digestive tract, the weaker appeared t o be the protection afforded. Significant inverse relationships were observed for liver, pancreas, prostate and urinary sites, but not for rectum, breast and female genital cancers or thyroid. No relationship emerged for lymphomas and myelomas. Even in the absence of a clear biological interpretation, the consistency and strength of the patterns observed indicate that, in this population, frequent green vegetable intake is associated with a substantial reduction of risk for several common epithelial cancers, and that fruit intake has a favourable effect, especially on upper digestive cancers and, probably, also on urinary tract neoplasms.
There are several indications that a diet rich in green vegetables and/or fresh fruit can provide some protection against a number of cancers, not only of the digestive tract but also in respiratory, genital and various other sites. This possible effect has been related to specific nutrients contained in fruit and vegetable, and particularly to vitamins A and C (Doll and Peto, 1981; Byers, 1988). It is not clear; however, whether this apparent protection is specifically linked with any of these micronutrients, whether other substances or metabolites are involved, or whether a diet rich in fruit and vegetables represents a more general, nonspecific favourable indicator. Still, ‘quantification of this relationship is important on a public health scale. Even in the absence of a precise understanding of the underlying mechanism(s) and hence focused indications for supplementation of specific (micro)nutrients, this can open practical perspectives for prevention and health education (La Vecchia et al., 1988a). We have therefore systematically examined the relationship between vegetables and fruit intake and cancer risk for a number of sites, using data from a network of case-control studies conducted in northern Italy. This area represents a favourable situation for epidemiological studies of dietary indicators of cancer risk, since it has gone through a considerable process of immigration from other regions throughout the current century,
and therefore it now shows wide heterogeneity in the dietary habits and other lifestyle correlates of cancer risk (La Vecchia et al., 1990b). SUBJECTS AND METHODS
The data were obtained from a series of hospital-based casecontrol studies, whose general design has been described (La Vecchia et al., 1988a, 1990b). Recruitment of cases of various cancers and the corresponding controls started between 1983 and 1985, and the present analysis is based on data collected before June, 1990. Trained interviewers identified and questioned patients with selected cancers and controls with acute, non-neoplastic conditions admitted to a network of teaching and general hospitals in the Greater Milan area. All interviews were conducted in hospital, and, on the average, refusal rate of eligible subjects (cases and controls) was below 3%. The same study scheme, criteria for inclusion of cases and controls and interview setting was utilized for all of the diseases studied. All the questionnaires comprised a basic structured section including socio-demographic factors and general characteristics and habits (smoking, alcohol, coffee and consumption of other methylxanthine-containing beverages). Further, patients were asked to indicate the frequency of consumption per week of selected indicator foods, including between 14 and 37 items for digestive sites. All the questionnaires also included a summary question on total vegetable and fruit consumption, thus making possible a combination of data from various studies in relation to these issues and major covariates of interest. Cases The cases considered for the present review were patients below the age of 75 years with histologically confirmed cancers of the oral cavity and pharynx (n = 119), oesophagus (n = 294), stomach (n = 564), colon (n = 673), rectum (n = 406), liver (n = 258), gall-bladder (n = 41), pancreas (n = 303), larynx (n = 149), breast (n = 2,860), endometrium (n = 567), ovary (n = 742), prostate (n = 107), bladder (n = 3651, kidney (n = 147), thyroid (n = 120), Hodgkin’s disease (n = 72), non-Hodgkin lymphomas (n = 173), and multiple myelomas (n = 117). The patients were admitted to the National Cancer Institute, to several university clinics, and to the Ospedale Maggiore, which includes the 4 largest teaching and general hospitals in Milan. Controls The control group comprised patients below age 75 admitted for a wide spectrum of acute conditions to the same network of hospitals in which cases had been identified. For admission diagnoses, specific exclusions were made for malignant tumours, digestive tract diseases or any disorder related to alco-
4T0whom correspondence and reprint requests should be sent, at the Istituto “Mario Negri”. Received: November 1 1 , 1990 and in revised form January 31, 1991.
VEGETABLES, FRUIT hol or tobacco consumption or which might have induced a long-term modification of the diet. A total of 6,147 controls (2,522 males, 3,625 females) was included in the present analysis. Of these, 26% were admitted for traumatic conditions (mostly fractures and sprains), 24% had non-traumatic orthopaedic disorders (mostly low back pain and disc disorders), 34% were admitted for acute surgical conditions (including plastic surgery), and 16% had other miscellaneous illnesses, such as ear, nose and throat, skin or dental disorders. The median age of the comparison group was 55 years, and the distribution of cases and controls according to sex and broad age groups is given in Table I. Data analysis Frequencies of fruit and vegetable consumption were divided into 3 levels, including, as far as possible, comparable numbers of cases and controls combined. These cut-off points, however, were defined on the basis of discrete frequencies (7 portions per week for vegetables;
Publication of the International Union Against Cancer Publication de I‘Union lnternationale Contre le Cancer
VEGETABLE AND FRUIT CONSUMPTION AND CANCER RISK Eva NEGRI’,~, Car10 LA VECCHIA’,~, Silvia FRANCEsCH13, Barbara D’AVANZO’and Fabio PARAZZINI’ ‘Istituto di Ricerche Farmacologiche “Mario Negri”, Via Eritrea 62, 20157 Milan, Italy; 21nstitute of Social and Preventive Medicine, University of Lausanne, 1005 Lausanne, Switzerland; and 3Aviano Cancer Centre, 33081 Aviano, Pordenone, Italy. The relationship between cancer r i s k and frequency of consumption of green vegetables and fruit has been analyzed using data from an integrated series of case-control studies conducted in northern Italy between 1983 and 1990. The overall dataset included the following histologically confirmed cancers: oral cavity and pharynx, I19; oesophagus, 294; stomach, 564; colon, 673; rectum, 406; liver, 258; gall-bladder, 41; pancreas, 303; larynx, 149; breast, 2,860; endometrium, 567; ovary, 742; prostate, 107; bladder, 365; kidney, 147; thyroid, 120; Hodgkin’s disease, 72; non-Hodgkin lymphomas, 173; myelomas, I 17; and a total of 6, I47 controls admitted t o hospital for acute non-neoplastic conditions, unrelated t o long-term dietary modifications. Multivariate relative risks (RR) for subsequent tertiles of vegetable and fruit consumption were derived after allowance for age, sex, area of residence, education and smoking. For vegetables, there was a consistent pattern of protection for all epithelial cancers, with R k in the upper tertile ranging from 0.2 for oesophagus, liver and larynx t o 0.7 for breast. All the trends in risk were in the same direction and significant for all carcinomas except gallbladder. In contrast, no protection was afforded by high vegetable consumption against non-epithelial lymphoid neoplasms. With reference t o fruit, strong inverse relationships were observed for cancers of the upper digestive and respiratory tract, with RRs in the upper tertile between 0.2 and 0.3 for oral cavity and pharynx, oesophagus and larynx relative t o the lowest tertile. The lower the location of the tumour in the digestive tract, the weaker appeared t o be the protection afforded. Significant inverse relationships were observed for liver, pancreas, prostate and urinary sites, but not for rectum, breast and female genital cancers or thyroid. No relationship emerged for lymphomas and myelomas. Even in the absence of a clear biological interpretation, the consistency and strength of the patterns observed indicate that, in this population, frequent green vegetable intake is associated with a substantial reduction of risk for several common epithelial cancers, and that fruit intake has a favourable effect, especially on upper digestive cancers and, probably, also on urinary tract neoplasms.
There are several indications that a diet rich in green vegetables and/or fresh fruit can provide some protection against a number of cancers, not only of the digestive tract but also in respiratory, genital and various other sites. This possible effect has been related to specific nutrients contained in fruit and vegetable, and particularly to vitamins A and C (Doll and Peto, 1981; Byers, 1988). It is not clear; however, whether this apparent protection is specifically linked with any of these micronutrients, whether other substances or metabolites are involved, or whether a diet rich in fruit and vegetables represents a more general, nonspecific favourable indicator. Still, ‘quantification of this relationship is important on a public health scale. Even in the absence of a precise understanding of the underlying mechanism(s) and hence focused indications for supplementation of specific (micro)nutrients, this can open practical perspectives for prevention and health education (La Vecchia et al., 1988a). We have therefore systematically examined the relationship between vegetables and fruit intake and cancer risk for a number of sites, using data from a network of case-control studies conducted in northern Italy. This area represents a favourable situation for epidemiological studies of dietary indicators of cancer risk, since it has gone through a considerable process of immigration from other regions throughout the current century,
and therefore it now shows wide heterogeneity in the dietary habits and other lifestyle correlates of cancer risk (La Vecchia et al., 1990b). SUBJECTS AND METHODS
The data were obtained from a series of hospital-based casecontrol studies, whose general design has been described (La Vecchia et al., 1988a, 1990b). Recruitment of cases of various cancers and the corresponding controls started between 1983 and 1985, and the present analysis is based on data collected before June, 1990. Trained interviewers identified and questioned patients with selected cancers and controls with acute, non-neoplastic conditions admitted to a network of teaching and general hospitals in the Greater Milan area. All interviews were conducted in hospital, and, on the average, refusal rate of eligible subjects (cases and controls) was below 3%. The same study scheme, criteria for inclusion of cases and controls and interview setting was utilized for all of the diseases studied. All the questionnaires comprised a basic structured section including socio-demographic factors and general characteristics and habits (smoking, alcohol, coffee and consumption of other methylxanthine-containing beverages). Further, patients were asked to indicate the frequency of consumption per week of selected indicator foods, including between 14 and 37 items for digestive sites. All the questionnaires also included a summary question on total vegetable and fruit consumption, thus making possible a combination of data from various studies in relation to these issues and major covariates of interest. Cases The cases considered for the present review were patients below the age of 75 years with histologically confirmed cancers of the oral cavity and pharynx (n = 119), oesophagus (n = 294), stomach (n = 564), colon (n = 673), rectum (n = 406), liver (n = 258), gall-bladder (n = 41), pancreas (n = 303), larynx (n = 149), breast (n = 2,860), endometrium (n = 567), ovary (n = 742), prostate (n = 107), bladder (n = 3651, kidney (n = 147), thyroid (n = 120), Hodgkin’s disease (n = 72), non-Hodgkin lymphomas (n = 173), and multiple myelomas (n = 117). The patients were admitted to the National Cancer Institute, to several university clinics, and to the Ospedale Maggiore, which includes the 4 largest teaching and general hospitals in Milan. Controls The control group comprised patients below age 75 admitted for a wide spectrum of acute conditions to the same network of hospitals in which cases had been identified. For admission diagnoses, specific exclusions were made for malignant tumours, digestive tract diseases or any disorder related to alco-
4T0whom correspondence and reprint requests should be sent, at the Istituto “Mario Negri”. Received: November 1 1 , 1990 and in revised form January 31, 1991.
VEGETABLES, FRUIT hol or tobacco consumption or which might have induced a long-term modification of the diet. A total of 6,147 controls (2,522 males, 3,625 females) was included in the present analysis. Of these, 26% were admitted for traumatic conditions (mostly fractures and sprains), 24% had non-traumatic orthopaedic disorders (mostly low back pain and disc disorders), 34% were admitted for acute surgical conditions (including plastic surgery), and 16% had other miscellaneous illnesses, such as ear, nose and throat, skin or dental disorders. The median age of the comparison group was 55 years, and the distribution of cases and controls according to sex and broad age groups is given in Table I. Data analysis Frequencies of fruit and vegetable consumption were divided into 3 levels, including, as far as possible, comparable numbers of cases and controls combined. These cut-off points, however, were defined on the basis of discrete frequencies (7 portions per week for vegetables;
