Search

AbstractThe European Prospective Investigation into Cancer and Nutrition (EPIC) is an ongoing multi-centre prospective cohort study designed to investigate the relationship between nutrition and cancer, with the potential for studying other diseases as well. The study currently includes 519 978 participants (366 521 women and 153 457 men, mostly aged 35?70 years) in 23 centres located in 10 European countries, to be followed for cancer incidence and cause-specific mortality for several decades. At enrolment, which took place between 1992 and 2000 at each of the different centres, information was collected through a non-dietary questionnaire on lifestyle variables and through a dietary questionnaire addressing usual diet. Anthropometric measurements were performed and blood samples taken, from which plasma, serum, red cells and buffy coat fractions were separated and aliquoted for long-term storage, mostly in liquid nitrogen. To calibrate dietary measurements, a standardised, computer-assisted 24-hour dietary recall was implemented at each centre on stratified random samples of the participants, for a total of 36 900 subjects. EPIC represents the largest single resource available today world-wide for prospective investigations on the aetiology of cancers (and other diseases) that can integrate questionnaire data on lifestyle and diet, biomarkers of diet and of endogenous metabolism (e.g. hormones and growth factors) and genetic polymorphisms. First results of case?control studies nested within the cohort are expected early in 2003. The present paper provides a description of the EPIC study, with the aim of simplifying reference to it in future papers reporting substantive or methodological studies carried out in the EPIC cohort.

Background Validation studies of dietary instruments developed for epidemiological studies have typically used some form of diet record as the standard for comparison. Recent work suggests that comparison with diet record may overestimate the ability of the epidemiological instrument to measure habitual dietary intake, due to lack of independence of the measurement errors. The degree of regression dilution in estimating diet-disease association may therefore have been correspondingly underestimated. Use of biochemical measures of intake may mitigate the problem. In this paper, we report on the use of urinary measures of intakes of nitrogen, potassium and sodium to compare the performance of a semi-quantitative food frequency questionnaire (FFQ) and a 7-day diet diary (7DD) to estimate average intake of these nutrients over one year.Methods In all, 179 individuals were asked to complete an FFQ and a 7DD on two occasions separated by approximately 12 months. The individuals were also asked to provide 24-hour urine samples on six occasions over a 6–9-month period, covering the time at which the record FFQ and 7DD were completed. The urine was assayed for nitrogen, potassium and sodium. The protocol was completed by 123 individuals. The data from these individuals were analysed to estimate the covariance structure of the measurement errors of the FFQ, the 7DD and a single 24-hour urine measurement, and to estimate the degree of regression dilution associated with the FFQ and 7DD.Results The results demonstrated that: (1) the error variances for each of the three nutrients was more than twice as great with the FFQ than the 7DD; (2) there was substantial correlation (0.46–0.58) between the error of both the FFQ and the 7DD completed on different occasions; (3) there was moderate correlation (0.24– 0.29) between the error in the FFQ and the error in the 7DD for each nutrient; (4) the correlation between errors in different nutrients was higher for the FFQ (0.77–0.80) than for the 7DD (0.52–0.70).Conclusions The regression dilution with the FFQ is considerably greater than with the 7DD and also, for the nutrients considered, greater than would be inferred if validation studies were based solely on record or diary type instruments.

Aims:to investigate whether connective tissue laxity is associated with abdominal aortic aneurysms (AAA). Methods:a nested case control study in a population-based screening programme. The presence of pes planus, scoliosis, pectus deformities, flexible auricular cartilages and Gorling's sign were combined with the Beighton joint mobility score to form a connective tissue laxity score. The association between connective tissue laxity and the risk of AAA was investigated through a logistic regression model. Type III collagen turnover was assessed using a serum radio-immunoassay for type III procollagen (PIIINP). Results:data from 231 controls (aortic diameter <2.5 cm) and 190 cases (AAA >2.9 cm) were analysed. Odds ratios (OR), adjusted for known confounders were 3.1 (95% CI: 1.1–8.6) for the highest group of connective tissue scores and 2.4 (95% CI: 1.0–5.4) for the middle group, compared with those with no signs of abnormal connective tissue function. There was no difference in mean collagen turnover between cases and controls, nor between those with a stable AAA >4 cm and those with an expanding AAA. Conclusion:connective tissue laxity is associated with a higher risk of having an AAA. The collagen turnover is similar in subjects with an AAA and controls. Aneurysms may be associated with abnormal connective tissue rather than an increased breakdown of normal collagen.

The backcalculation method has been extensively used in AIDS modelling and forecasting. Knowledge of reported AIDS cases, information on the time between HIV infection and onset of AIDS, and assumptions on the rate at which infections occurs, can be used to reconstruct the past history of the HIV epidemic, as well as to provide short term predictions of AIDS incidence. Uncertainty in the three components of the backcalculation method and the increasingly available information on HIV prevalence must be taken into account in order to provide realistic projections. In this paper we discuss ways of acknowledging uncertainty and suggest a Bayesian formulation of the backcalculation idea as a means of combining into a single model both random and systematic variation as well as prior information.

In a previous study in France, we reported that the relative risk of breast cancer associated with a family history of breast cancer was higher in those subjects with a history of abortions. The present study was undertaken to check the existence of this interaction in other studies and to investigate whether the interaction is modified by the time at which abortions occur. Data were obtained from six case-control studies in France, Australia and Russia, with information on family history of breast cancer and abortion for 2693 breast cancer cases and 3493 controls. The interaction effect was estimated in each study separately, then combined using a multivariate weighted average. The relative risk conferred by a family history of breast cancer increased with the number of abortions (1.8 for no abortion, 1.9 for one abortion, 2.8 for two or more). There was a significant interaction between total number of abortions and family history (P = 0.04), but this was no longer significant when adjusted for other risk factors. The familial risk was highest for those who had had an abortion before first childbirth (1.9 for abortion after first childbirth, 2.7 for abortion before first childbirth). The adjusted risk associated with family history was significantly higher in those with an abortion before first childbirth (P = 0.04). Our findings suggest a synergism between familial factors and abortion. The interaction was not substantially modified by the type of abortion (spontaneous or induced) but was modified by the time at which it occurred in relation to first childbirth. This suggests an effect of abortion itself rather than predisposition to abortion. Further studies of breast cancer cases, particularly among BRCA1 gene carriers and their families, could improve our understanding of this effect.

The main risk factors for oesophageal cancer previously identified in western Europe are tobacco smoking and alcohol drinking. However, a study of the time trend from 1951 to 1985 of the mortality from oesophageal cancer in 17 European countries shows that, except among the younger age groups in men, oesophageal cancer had either decreased or increased only slightly in most countries. This trend differed from that of lung cancer, cirrhosis and alcohol consumption which had in general increased substantially during the period. The results strongly suggest that population-wide changes in certain undetermined risk/protective factor(s), one possibility of which is the consumption of fruit, had overridden the effect of tobacco and alcohol and resulted in a reduction of oesophageal cancer risk. Apart from further efforts to reduce smoking and drinking, studies to identify the factor(s) will be of great public health importance to the prevention of oesophageal cancer.

It is important that the introduction of breast screening is closely monitored. The anticipated effect on breast cancer mortality will take 10 years or more fully to emerge, and will only occur if a succession of more short-term end points are met. Data from the Swedish two-county randomised trial provide targets that should be achieved, following a logical progression of compliance with the initial invitation, prevalence and stage distribution at the prevalence screen, the rate of interval cancers after the initial screen, the pick-up rate and stage distribution at later screening tests, the rate of interval cancers after later tests, the absolute rate of advanced cancer and finally the breast cancer mortality rate. For evaluation purposes, historical data on stage at diagnosis is desirable; it is suggested that tumour size is probably the most relevant variable available in most cases.

A case-control study of cancer of the gingiva was carried out in Kerala, Southern India, using 187 cases and 895 hospital-based controls. We investigated the effects on risk in males of pan (betel)-tobacco chewing, bidi and cigarette smoking, drinking alcohol and taking snuff. In females only pan-tobacco chewing was investigated as very few females indulged in the other habits. Among males, significant positive associations with risk were observed for pan-tobacco chewing (P less than 0.001), bidi smoking (P less than 0.001) alcohol drinking (P less than 0.001) and snuff use (P less than 0.05). In females, pan-tobacco chewing had a similar predisposing effect (P less than 0.001). Daily frequency of pan-tobacco chewing was the strongest predictor of risk in males, with a relative risk of 15.07 associated with chewing ten or more quids per day. The corresponding relative risk among females was 13.69. In males a relative risk of 3.20 was associated with smoking more than 20 bidis per day, and relative risks of 2.62 and 3.90 were associated with regular use of alcohol and snuff respectively. Surprisingly high relative risks were observed in association with occasional use of pan-tobacco, bidi, cigarettes, alcohol and snuff. A stepwise logistic regression analysis yielded four predictors: pan-tobacco daily frequency, duration of bidi use, and alcohol and snuff use (regular versus never). There were also significantly elevated risks associated with occasional indulgence in these four habits. Total lifetime exposure was no better at predicting risk than daily frequency or duration of habits.

Further results are presented from the Swedish two-county breast cancer screening trial. The reduction in the rate of advanced cancers and of breast cancer mortality in the group allocated to screening when compared to the control group has accelerated with a further year of follow-up. Mortality due to other causes and the rate of other cancers remains similar in the two groups. Attention has been focused on the rate at which cancers start re-emerging among women with negative mammograms. Among women over 50 years of age at entry to the study, relatively few interval cancers are seen in the first two years after a screening test; in the third year the rate rises to nearly 50% of the comparable rate in the control group. Among women aged 40-49 years at entry, by contrast, the rate of interval cancers even in the first post screening year is nearly 40% of that in the controls and in the second year nearly 70%. In older women in the group allocated to screening, much of the breast cancer mortality comes from the refusers and little from the interval cancers; in younger women the picture is reversed. The implications for screening policy, including the interscreening interval are discussed.

A case-control study was conducted to determine whether the development of leukaemia was associated with chemotherapy for neoplasms of the ovary or breast, in a population where most such chemotherapy consisted of cyclophosphamide alone. Cases and controls were identified from the National Cancer Registry of the German Democratic Republic. Cases were women who had developed leukaemia as a second primary after an initial diagnosis, at least one year before, of an ovarian or breast tumour. Controls were patients with an ovarian tumour or breast cancer who had survived to the year when the case developed a leukaemia but who had not themselves developed a second malignancy. Controls were matched to cases by the site of the first primary and its year of diagnosis, as well as year of birth. The relative risk for acute leukaemia following treatment with cyclophosphamide alone was significantly elevated (P less than 0.05), at 14.6 for ovarian tumour patients and 2.7 for breast cancer patients. Among breast cancer patients the increased risk of leukaemia associated with chemotherapy was confined to women who had been under 50 years of age at the time of diagnosis of the breast cancer (for whom the relative risk was 13.1). No excess risk of leukaemia was observed in association with radiotherapy for either ovarian or breast cancer patients. The present findings strongly suggest that cyclophosphamide as a single chemotherapeutic agent is capable of inducing leukaemia in humans.

In a case-control study of second primary cancers in Moscow, there were 165 cases and 294 controls, matched for site of first primary, duration of follow-up since first primary and relapse history. Of the cases, 18 were of acute, non-lymphocytic leukaemia (ANLL), with 39 matched controls. Risk of ANLL was assessed with respect to chemotherapy for the first primary tumour. The chemotherapeutic agents investigated were nitrogen mustard, cyclophosphamide, procarbazine, doxorubicin, bleomycin, vinblastine, vincristine, prednisone and combinations. Increased risks were associated with use of nitrogen mustard (odds ratio = 9.94, not significant), doxorubicin (odds ratio = 11.25, 0.1 > P > 0.05) and vincristine (odds ratio = 26.57, P < 0.05). Despite the small number of cases and potential confounding by other agents, these findings, together with those of previous studies, suggest that some non-alkylating agents may predispose to second malignancies.

The results of the Swedish two-county study are analysed with respect to tumour size, nodal status and malignancy grade, and the relationship of these prognostic factors to screening and to survival. It is shown that these factors can account for much of the differences in survival between incidence screen detected, interval and control group cancers but to a lesser extent for cancers detected at the prevalence screen where length bias is greatest. Furthermore, examination of the relationships among the prognostic factors and mode of detection indicates that malignancy grade, as a measure of inherent malignant capacity, evolves as a tumour grows. The proportion of cancers with poor malignancy grade is several fold lower for cancers of diameter less than 15 cm than for cancers greater than 30 cm, independent of the length bias of screening. The implications of these findings for screening frequency are briefly discussed.

In this paper, a combined analysis was performed to study the interaction between familial risk and reproductive life factors. In particular, the interaction between familial risk and breast cell mitotic activity (BCMA), as assessed by duration of ovarian activity, was investigated because of the potential importance of mitotic activity on genetically susceptible cells. The present analysis included 3152 cases and 4404 controls in seven case-control studies from four countries. The interaction effect was estimated in each study separately, then combined using two different methods: a multivariate weighted average and a Bayesian random-effects model. The main effects of reproductive life factors on the risk of breast cancer were in agreement with the previous findings. In particular, an increased duration of BCMA before the first childbirth and over life was found to increase the risk of breast cancer (P < 0.001). Slightly increasing but non-significant, familial risks were observed with increasing number of children (P = 0.17), increasing age at first childbirth (P > 0.2) and increasing duration of BCMA (P > 0.2). There was no modification in familial risk with age at menarche and no clear pattern with menopause characteristics. A weak influence of reproductive and menstrual factors on the familial risk emerged from the present study.

A recent study showed an association between a single base substitution, T-->C, in the promotor region of the CYP17 gene, the risk of breast cancer and age at menarche in Asian, African-American and Latino women from California and Hawaii. The C allele was associated with increased risk of breast cancer, significantly so for patients presenting with advanced disease, whereas the TT genotype was associated with later age at menarche in control subjects. We attempted to confirm these findings in a large case-control study in East Anglia, England (835 cases and 591 control subjects). We found no evidence of an increased risk of breast cancer [odds ratio (OR) 1.10, confidence interval (CI) 0.89-1.37] or advanced breast cancer (OR 0.88, CI 0.38-2.01) in C allele carriers, nor any association between age at menarche and genotype. We conclude that these alleles do not significantly alter breast cancer risk in the English population.

Literature on seasonality of leukaemia shows conflicting results. We analysed the month of diagnosis of acute leukaemia in East Anglia, UK, for the period 1971-94, which showed a significant 40% summer excess (P < 0.001) for acute lymphocytic leukaemia both in children (P < 0.01) and adults (P = 0.01). Methodology, results and possible aetiological interpretations are presented.

Background In nutritional epidemiology, it is common to fit models in which several dietary variables are included. However, with standard instruments for dietary assessment, not only are the intakes of many nutrients often highly correlated, but the errors in the estimation of the intake of different nutrients are also correlated. The effect of this error correlation on the results of observational studies has been little investigated. This paper describes the effect on multivariate regression coefficients of different levels of correlation, both between the variables themselves and between the errors of estimation of these variables. Methods Using a simple model for the multivariate error structure, we examine the effect on the estimates of bivariate linear regression coefficients of (1) differential precision of measurement of the two independent variables, (2) differing levels of correlation between the true values of the two variables, and (3) differing levels of correlation between the errors of measurement of the two variables. As an example, the prediction of plasma vitamin C levels by dietary intake variables is considered, using data from the European Prospective Investigation of Cancer (EPIC) Norfolk study in which dietary intake was estimated using both a food frequency questionaire (FFQ) and a 7-day diary (7DD). The dietary variables considered are vitamin C, fat, and energy, with different approaches taken to energy adjustment. Results When the error correlation is zero, the estimates of the bivariate regression coefficients reflect the precision of measurement of the two variables and mutual confounding. The sum of the observed regression coefficients is biased towards the null as in univariate regression. When the error correlation is non-zero but below about 0.7, the effect is minor. However, as the error correlation increases beyond 0.8 the effect becomes large and highly dependent on the relative precision with which the two variables are measured. At the extreme, the bivariate estimates can become indefinitely large. In the example, the error correlation between fat and energy using the FFQ appears to be over 0.9, the corresponding value for the 7DD being approximately 0.85. The error correlation between vitamin C and fat, and vitamin C and energy, appears to be below 0.5 and smaller for the 7DD than for the FFQ. The impact of these error correlations on bivariate regression coefficients is large. The effect of energy adjustment differs widely between vitamin C and fat. Conclusion High levels of error correlation can have a large effect on bivariate regression estimates, varying widely depending on which two variables are considered. In particular, the effect of energy adjustment will vary widely. For vitamin C, the effect of energy adjustment appears negligible, whereas for fat the effect is large indicating that error correlation close to one can partially remove regression dilution due to measurement error. If, for fat intake, energy adjustment is performed by using energy density, the partial removal of regression dilution is achieved at the expense of substantial reduction in the true variance.