Introduction Dietary docosahexaenoic acid (DHA) and eicosapentanaenoic acid (EPA) are both n-3 polyunsaturated fatty acids (n3 PUFAs) which have both anti-inflammatory and anti-carcinogenic actions, including inhibiting angiogenesis and tumour cell proliferation, and promoting apoptosis. The aim of this study was to conduct the first epidemiological investigation to determine if there is an inverse association between DHA and EPA intake and the risk of developing both Barrett's oesophagus (BO) and oesophageal adenocarcinoma (OAC).

Methods A total of 25 639 men and women aged 40–75 years, were recruited between the years 1993 and 1997 into the European Prospective Investigation into Cancer (EPIC)-Norfolk cohort Study. At baseline, participants completed detailed 7-day food diaries which were then coded by nutritionists. Subjects were then followed-up over subsequent years for the development of BO and OAC. A review of case notes confirmed these diagnoses. Dietary intake of n3 PUFAs were compared between cases and a random sample of 3 797 controls in a case-cohort analysis. Intakes were divided into quintiles and both biological gradients and threshold effects were calculated. Cox regression estimated the HR for both DHA and EPA, adjusted for age, gender, BMI, smoking, alcohol, and arachidonic acid intake.

Results During follow-up, 104 patients were diagnosed with BO (80% men, median age 67.0 yrs [IQR 61.1–73.1] at diagnosis) after a median follow-up of 6.2 yrs (IQR 4.1–8.1). A further 63 patients developed OAC (83% men, median age 73.0 yrs [IQR 67.0–78.0] at diagnosis) after a median follow-up of 6.4 yrs (IQR 4.4–8.9). For OAC, there was a large significant inverse association, in a threshold manner, comparing the lowest quintile with a summation of the top four quintiles of dietary intake, for DHA (HR 0.47, 95% CI 0.25 to 0.88, p=0.02). EPA (HR 0.60, 95% CI 0.31 to 1.15, p=0.12) intake was not significantly associated with OAC. For BO, no significant associations were found with either dietary DHA (HR 0.69, 95% CI 0.40 to 1.20, p=0.19) or EPA (HR 1.03, 95% CI 0.56 to 1.90, p=0.93). No gradient effects across quintiles were seen for either BO or OAC.

Conclusion Dietary DHA was associated with an approximate 50% risk reduction for OAC, although there were no significant effects for BO. The data support a role for dietary DHA in preventing the malignant transformation of BO to OAC, and therefore should be measured in future aetiological studies of OAC.

Competing interests None declared.