Vegetarian and pescatarian diets have been linked to lower risks of several cancers in the largest analysis of its kind, but unexpected increases in oesophageal and colorectal cancer highlight the need for caution.
Cutting meat from the menu may reduce the risk of some cancers, but a major international study suggests the relationship between vegetarian diets and cancer is far from straightforward.
A pooled analysis of more than 1.8 million people found pescatarians had lower risks of colorectal, breast, and kidney cancer than meat eaters, while vegetarians had lower risks of multiple myeloma and pancreatic, breast, prostate, and kidney cancers.
But vegetarians also had almost twice the risk of oesophageal squamous cell carcinoma, while vegans had a 40% higher risk of colorectal cancer.
The study, published in the British Journal of Cancer, brought together nine prospective cohorts from the UK, US, Taiwan, and India and represented the largest investigation of vegetarian diets and cancer risk to date, the researchers said.
They analysed data from 1,645,555 meat eaters, 57,016 poultry eaters, 42,910 pescatarians, 63,147 vegetarians, and 8849 vegans. After a median 16 years of follow-up, the analysis included more than 220,000 incident cancers across 17 cancer sites.
Compared with meat eaters, pescatarians had a 15% lower risk of colorectal cancer (HR 0.85, 95% CI 0.77-0.93), a 7% lower risk of breast cancer (HR 0.93, 95% CI 0.88-0.98), and a 27% lower risk of kidney cancer (HR 0.73, 95% CI 0.58-0.93).
Vegetarians had a 21% lower risk of pancreatic cancer (HR 0.79, 95% CI 0.65-0.97), 9% lower risk of breast cancer (HR 0.91, 95% CI 0.86-0.97), 12% lower risk of prostate cancer (HR 0.88, 95% CI 0.79-0.97), 28% lower risk of kidney cancer (HR 0.72, 95% CI 0.57-0.92), and 31% lower risk of multiple myeloma (HR 0.69, 95% CI 0.51-0.93).
Poultry eaters had a more modest 7% reduction in prostate cancer risk compared with meat eaters.
The breast cancer results appeared to be driven by postmenopausal disease, the researchers found. Pescatarians had a 9% lower risk of postmenopausal breast cancer and vegetarians an 11% lower risk, while the associations were not statistically significant among premenopausal women.
There was no significant difference between diet groups for endometrial or ovarian cancer.
The gastrointestinal findings were less reassuring.
Vegetarians had a 93% higher risk of oesophageal squamous cell carcinoma than meat eaters (HR 1.93, 95% CI 1.30-2.87). Vegans had a 40% higher risk of colorectal cancer (HR 1.40, 95% CI 1.12-1.75), and subsite analysis showed a 78% higher risk of rectal cancer.
Meanwhile, colorectal cancer risk among vegetarians was essentially unchanged compared with meat eaters (HR 1.03, 95% CI 0.94-1.13).
The vegan result was based on just 93 colorectal cancer cases and the researchers urged particular caution in interpretation. After the first four years of follow-up were excluded to reduce the possibility of reverse causation, the association weakened and was no longer statistically significant, they said.
The finding was also unexpected because vegans reported the lowest alcohol consumption and highest wholegrain and fibre intake. The researchers noted that vegans had the lowest calcium intake across cohorts with available nutrient data, raising one possible avenue for further investigation, but the study could not establish what was responsible for the association.
Related
The researchers also tested whether their main findings held up after excluding the first four years of follow-up and when analyses were restricted to never smokers.
Of 11 initially significant associations, the most consistent were the increased risk of oesophageal squamous cell carcinoma and reduced kidney cancer risk among vegetarians.
For oesophageal squamous cell carcinoma, the risk among vegetarians remained elevated after excluding the first four years of follow-up (HR 2.09, 95% CI 1.34-3.26) and was higher again among never smokers (HR 2.93, 95% CI 1.82-4.73).
The lower kidney cancer risk among vegetarians was similarly robust, with an HR of 0.70 after excluding the first four years and 0.64 among never smokers.
Other associations were less consistent in the sensitivity analyses, including those involving breast and prostate cancer, multiple myeloma and vegan diets, and colorectal cancer.
The researchers noted that the observational findings could not establish causation. Diet groups differed in characteristics other than their consumption of animal products, with vegetarians generally having healthier lifestyle profiles, including lower levels of smoking and alcohol consumption.
While analyses adjusted for BMI, smoking, alcohol, physical activity, diabetes, education, and other potential confounders, residual confounding from unmeasured or imperfectly measured factors remained possible.
Dietary categories themselves were also broad. Vegetarian diets were defined primarily by the foods excluded rather than the quality of foods consumed, meaning they could range from diets rich in fruit, vegetables, and wholegrains to those high in refined carbohydrates and other less healthy foods.
The researchers also lacked information on food processing and cooking methods and could not determine how long all participants had followed their diets before entering the studies.
Generalisability was another concern. Most participants lived in the UK or US, and vegans accounted for only 8849 participants, or 0.5% of the study population.
The researchers said more studies were needed to investigate the possible roles of metabolic factors and nutritional deficiencies, particularly among vegans and populations outside Western Europe and North America.
“Future research should examine the possible mediating roles of both metabolic factors and nutritional deficiencies, and collect more data particularly in vegans and in populations outside Western Europe and North America,” the researchers concluded.
“The generalisability of the findings should be considered cautiously, because the diets and nutritional intakes of both vegetarians and non-vegetarians can vary substantially within and between populations.”
