The EPIC link between diet & cancer
EPIC-Norfolk is part of the Europe-wide EPIC study, the largest study of diet and health ever undertaken. It involves over half a million people in ten countries (figure 1, below). By studying large numbers of people in different countries with differing diets, using carefully designed and tested questionnaires, EPIC aims to produce much more specific information about the effect of diet on long-term health compared with previous studies that have tried to answer this question.
Different groups of people (cohorts) are being recruited in each centre. The elements common to all centres are: detailed information on diet and lifestyle obtained by questionnaire, measurements such as height, weight, waist and hip, and blood samples which are stored in liquid nitrogen.
With the help of over 30,000 people living in Norfolk (figure 2, above ), Dr. Sheila Bingham's team at the MRC-Dunn Nutrition Unit, University of Cambridge, are finding out about the factors that are most often present when people stay healthy and those more likely to be there when they develop cancer or other diseases.
EPIC-Norfolk participants are men and women who were aged between 45 and 74 when they joined the study, and who lived in Norwich and the surrounding towns and rural areas. They have been contributing information about their diet, lifestyle and health through questionnaires, and through health checks carried out by EPIC nurses.
Background Norfolk was chosen because very few people in the age group of interest move out of the county, making it easier to follow participants. Also, it is served mainly by one district general hospital, the Norfolk and Norwich University Hospital NHS Trust.
If a study is carried out on people who already have cancer, this may affect, or bias, the way they report what they have eaten in the past. Regardless of this, it is always difficult to describe what you were eating several years ago. A more accurate way of examining the effect of diet is to gather a great deal of lifestyle and nutritional information from a very large number of people. As time goes by, more and more information about this group is collected. If anybody in the group develops one of the diseases being studied, this is recorded. This type of investigation is called a 'cohort study'.
Using this study design, the team should be able to make precise comparisons of how food intake can increase the risk of disease or - more interestingly - reduce it.
The original aim was to identify a cohort of 25,000 men and women from the general population of Norfolk. This cohort size was a compromise between the large numbers needed to get enough observations on people who developed cancer or other diseases, and the expense of making detailed measurements on everybody in the cohort. In fact, the team recruited over 30,000 people to EPIC-Norfolk.
Aims It has been estimated that more than 30% of cancers are related to diet and are hence potentially preventable, although the estimated contribution of diet varies from around 10% for lung cancer to 80% for cancer of the prostate and large bowel. Many dietary items have been suggested as promoting or preventing cancer, but evidence for only a few specific relationships has been found. The primary aim of the international EPIC study is to examine the relationships between what people have reported as their dietary intake, and incident cancers; that is, cancers which have developed after they joined the study.
A secondary aim is to study the relationship between dietary intake and other diseases and disease risk factors. In EPIC-Norfolk, these include heart attacks and strokes, rheumatoid arthritis, diabetes, thyroid disease, osteoporosis, Alzheimer's disease and many others. The team are also studying the link between disease and other factors, such as psychosocial health. Some of this work on different diseases and factors is carried out with colleagues from other institutions (see acknowledgements section).
Methods: Baseline Recruitment to EPIC Norfolk began in March 1993 and was completed at the end of 1997. Invitations were sent to all 45-74 year olds on the list of collaborating general medical practitioners (GPs). The GP surgeries approached were in rural areas of Norfolk and market towns as well as the city of Norwich. The information collected when people first joined the study is known as baseline information.
Over 30,000 of the people approached returned a consent form indicating that they wished to join the study. They also completed a health and lifestyle questionnaire, which asked about their medical history, lifestyle and other known risk factors, such as smoking and alcohol intake. In addition, people who agreed to enter the study completed a form recalling what they had eaten in the last twenty-four hours.
|Figure 3: Food frequency questionnaire|
After returning the consent form and first set of questionnaires, participants were sent an appointment for a health check and a food frequency questionnaire (figure 3), which asked how often they had eaten different foods over the last year.
|Figure 4 : Laboratory technician preparing samples|
At the health check, the participant was examined by one of seven nurses who have been carefully trained to adhere to the EPIC protocol. The nurse recorded people's height and weight, their waist, hip and chest measurements and took readings of their blood pressure and lung capacity. The participants also gave blood and urine samples. Over 25,000 people attended for health checks.
At the end of the health check, the nurse gave each participant guidance on how to complete the remaining questionnaire, the seven-day food diary. The diary required participants to record in detail what they ate and drank over a period of a week. The nurses received special training in dietary survey methods to ensure that participants recorded their diet accurately, and that nurses were not 'judgmental' about food intake. As a result, 93% of participants completed the food diary. The diaries were completed to a very high standard, some people sending recipes and packaging from foods. This helped the team to identify accurately the nutritional composition of their diet. All the results from the health check, including blood pressure and cholesterol levels, were sent to the participant's GP, who was responsible for any further follow-up and treatment.
After the baseline measurements, a calibration study was carried out on a sample of 600 EPIC-Norfolk participants. This used a standardised, computerised 24-hour diet recall method (EPIC-SOFT) to enable dietary comparisons to be made between the 10 countries taking part in EPIC. Weight and height measurements were taken and 24-hour urine collections made.
Methods: Follow up Once the collection of baseline data is complete, it is important to establish what happens to the participant at a later date. The follow-up of participants employed several different methods :
- Self reported questionnaires (at eighteen months) Everyone who consented to join the study was approached eighteen months later. They were sent an information sheet containing a description of EPIC, Europe wide, and a progress report. They were asked to complete a second seven-day food diary and a follow-up health questionnaire. The eighteen-month gap between diaries was selected to ensure that seasonal variations in diet were recorded. If participants responded to this, they were also sent a health and life experiences questionnaire.
- Second health check (at three years) Three years after the first health check, all participants were invited back for a second health check in which most measurements performed at the baseline health check were repeated. Also, the participants completed a second follow-up health questionnaire, a twenty-four hour diet recall, a food frequency questionnaire and a seven-day food diary. Some measures obtained were not carried out at the baseline health check: heel ultrasound measurements of each foot were performed and percentage body fat was estimated. Participants also completed a questionnaire on their physical activity.
Methods: Health Events The team needs to find out whether people in the EPIC cohort have developed cancer or other diseases, or have died. It is important to have this information for all participants, so that the team can make sure that they are making valid comparisons when they look at diet and other factors in people who have developed a particular disease and people who have not. Several organisations are helping them to collect this information:
- The East Anglian Cancer Registry records information on people in East Anglia diagnosed with cancer. The team find out about newly diagnosed cancers in people in the EPIC cohort through the questionnaires they return, and from the Cancer Registry.
- ENCORE is the database of hospital admissions kept by the East Norfolk Health Commission. At regular intervals, details of hospital admissions are sent to EPIC. ENCORE provides excellent data on admissions to hospital for all but those people admitted to private hospitals.
- The Family Health Service Authorities - some individuals leave the area after they have joined EPIC. The Family Health Service Authorities (FHSAs) helps EPIC to track these individuals. If a participant has registered with a new GP within the area, the FHSA informs EPIC of the name and address of the new GP. The participant is then tracked with the co-operation of the new GP. The same method is used if someone leaves the area, but in this case the FHSA notifies EPIC of the new FHSA, which notifies EPIC of the new GP.
- The Office for National Statistics has a list of EPIC participants. They notify the team when people on the list have died or have been diagnosed with cancer. Working with the Office in addition to the East Anglian Cancer Registry is necessary to get this information for people who have moved out of Norfolk since they joined EPIC.
Key Findings :
- Imprecise methods may be obscuring a relation between fat and breast cancer - Published in the Lancet, July 2003 The team studied more than 13 000 women participants and found that those who ate the most saturated fat were almost twice as likely to develop breast cancer as those who ate the least. Saturated fats are found mainly in full-fat milk, meat and products such as biscuits and cakes. In the past many large studies have failed to find a link between fat intake and breast cancer, possibly due to imprecise methods. EPIC participants were asked to complete a detailed food diary over the course of seven days. Even the brand of food was recorded so that the nutritional content could be worked out more precisely.
The team found that women who ate more than 90 grams of fat per day have twice the risk of developing breast cancer as women who ate less than 40 grams of fat per day.
Incidentally, two-thirds of a pint of full-fat milk contains 16 grams of fat. The same volume of semi-skimmed milk contains only 7 grams of fat.
- Eating fruits and vegetables reduces the risk of an early death - Published in the Lancet, March 2001 The blood levels of vitamin C in almost 20,000 EPIC participants were also examined. It was found that a rise in the level of vitamin C equivalent to a 50g per day increase in fruit and vegetable consumption could cut the risk of dying early from any cause by 20 per cent. This roughly corresponds to eating an extra apple a day - adding two more daily portions of fruit and vegetables could reduce the risk by as much as half. These findings hold regardless of people's age, blood pressure or whether they smoked. The team also found that men and women with the highest levels of vitamin C were 60 to 70 per cent less likely to die from heart attacks or strokes than those with the lowest levels.
These are some of the largest effects of a small change in diet ever detected. They show that there could be significant health gains from even a small increase in fruit and vegetable consumption.
- High impact sports may preserve bone density - Published in the British Medical Journal, January 2001 Data from over 5,000 EPIC participants aged 45-74 who had attended for a second health check was examined. At the check, ultrasound measurements of the heels were made. Low heel bone ultrasound values (low attenuation) are known to be associated with low bone mineral density. The participants also completed a questionnaire on their physical activity. From this questionnaire, people's recreational activities were classified into four groups according to levels of impact. Activities with no impact included swimming and snooker. Badminton and step aerobics, for example, were classified into the highest impact group. The team found that reported time spent on high impact physical activity was strongly and positively associated with higher ultrasound levels, regardless of people's age and weight. Men who reported taking part in high impact activities for more than two hours a week had around 10% higher ultrasound levels than men who recorded no activity of this type. For women, the high impact group had levels 3% higher, an effect similar in size to an age difference of four years. Women who climbed stairs more often also had higher ultrasound measurements, and those who spent more time watching television had lower levels.
Results suggest that participation in high impact activities may help preserve bone density and reduce the risk of fracture for people in mid-life.
(However, this would not be appropriate for older people, who have thinner bones, as these activities could increase the likelihood of falls and fractures).
- Blood glucose levels could help predict risk of heart disease - Published in the British Medical Journal, January 2001 Nearly 5,000 male EPIC participants aged 45 to 79 completed the baseline health examination and had the levels of glycosylated haemoglobin (HbA1c) in their blood measured. HbA1c is an indicator of average blood glucose levels over the previous three months. It was previously known that people with diabetes have an increased risk of heart disease and that HbA1c concentrations can predict risk of heart disease in people with diabetes.
It was found that around half of the male population studied had an increased risk of fatal heart attacks and strokes because of raised blood glucose levels.
The vast majority of these men would not be classified as diabetic. Risk of death from all causes, and from heart disease in particular, was found to increase with increasing HbA1c concentration at all levels. Even mildly elevated blood glucose seemed to predict heart disease. HbA1c concentration in women was also measured. Women are less likely to develop heart disease than men, and the team did not observe enough deaths in the study period to draw any firm conclusions. The team would expect similar results to apply to women, though it will have to wait longer to confirm this.
Findings suggest that blood glucose should be considered with high blood pressure and high cholesterol as an important factor in heart disease. Reducing HbA1c concentration in the whole population by just 0.2 per cent could reduce overall mortality by 10 per cent.
|Figure 5 : An EPIC-Norfolk Group photograph taken at an open day|
ACKNOWLEDGEMENTS: Funders/Sponsors: Medical Research Council, Cancer Research UK, Food Standards Agency, British Heart Foundation, Department of Health, Europe Against Cancer Programme, MAFF, Stroke Association and the World Health Organisation
COLLABORATORS: The CRC Human Cancer Genetics Group; The European Prospective Osteoporosis Study; The European Prospective Osteoporosis Study; Dr Andrew Hart University of East Anglia (Gallstones & inflammatory bowel disease); GPs in Norfolk; The Norfolk Arthritis Register; The UK Flexible Sigmoidoscopy Screening Trial; Dr Paul Surtees, Dr Nick Wainwright of the Social Epidemiology group at Strangeways Reseach Laboratory and Dr Carol Brayne of the Department of Public Health and Primary Care and Dr Felicia Huppert of the Department of Psychiatry in the University of Cambridge (Psychosocial measures)
Ethics committee permission: EPIC are in close and regular contact with the Norwich District Ethics Committee. At each stage, ethical permission has been sought. For instance, by February 2001, 25 separate permissions have been sought and granted by ethics committees in both Norfolk and Suffolk.
The Data Protection Act and confidentiality: The Department of Public Health and Primary Care of the University of Cambridge, based in the Institute of Public Health in Cambridge, has a licence permitting it to carry out research in collaboration with general practitioners. General practitioners have a licence to permit them to conduct research in collaboration with universities. This study is therefore covered by both these licences. All practices associated with EPIC comply with the Data Protection Act 2000. In addition, all members of staff who work with EPIC sign a confidentiality declaration, stating that they will not divulge data to any other party and that specific and identifiable information will not be used in any other context than that of health care. Data, both in the form of paper records and computer records, are kept in a coded form, which makes it impossible to identify any individual without the key. When blood samples are provided, a 'double encryption' is used to preserve anonymity. All paper records are kept in locked rooms in a locked building.
This article first appeared as a feature article in TranScript (14th January 2004) - an online newsletter produced by Biotext (ICP Europe Publishing PLC)