Although evidence has been mounting that obesity may be related to the increased incidence of oesophageal and gastric cardia malignancies, these reports (mainly case-control studies) have relied on imperfect measures of obesity such as body mass index (BMI), and generally have not clearly distinguished between anatomical subsites within the oesophagus and stomach. In a prospective study of people aged 27-75 years, we directly measured fat mass and fat-free mass (using bioelectrical impedance analysis), height, weight and waist and hip circumferences. Among 41,295 people followed on average for 11.3 years, 30 cases with cancers in the gastric cardia or lower third of the oesophagus and 68 cases with noncardia gastric adenocarcinomas were ascertained via the population cancer registry. The risk of adenocarcinoma of the lower oesophagus and gastric cardia was positively associated with BMI with a hazards ratio (HR) and (95% confidence interval) for people with BMI ≥ 30 kg/m 2 compared with those <25 kg/m 2 , of 3.7 (1.1-12.4), an HR per 10 cm increase in waist circumference of 1.46 (1.05-2.04), and a HR per 10 kg increase on fat-free mass of 2.06 (1.15-3.69). Noncardia gastric adenocarcinoma showed little relationship with body size. We observed an increased risk of adenocarcinoma of the lower oesophagus and gastric cardia associated with increased BMI, central adiposity and the nonfat component of weight, but found no association with noncardia gastric adenocarcinoma. An increasing prevalence of obesity may be associated with the increasing incidence of gastro-oesophageal cancer observed in many populations. ' 2005 Wiley-Liss, Inc.Key words: gastro-oesophageal adenocarcinoma; body size; body composition; bioelectrical impedance analysis; cohort study Gastric cancer has been declining in Western populations for many years but the incidence of cancer of the gastric cardia and lower oesophagus has gone against this trend.1 Rising incidence has been linked with the increasing prevalence of Barrett's oesophagus.2 Although not entirely consistent, evidence has been mounting that obesity may be related to the increased incidence of oesophageal and gastric cardia malignancies. [3][4][5][6][7][8][9][10][11][12][13][14][15][16][17] However, most studies have been case-control studies, which are prone to several sources of bias, including recall bias in regard to self-reporting of body weight after diagnosis. Further, they have considered only body mass index (BMI) that does not distinguish between fat and nonfat mass. They have also not distinguished between anatomical and histopathological subtypes within the oesophagus and stomach, particularly malignancies arising around the gastro-oesophageal junction.Using a prospective cohort study, we examined whether direct measurements of body size (including waist circumference, fat-free mass and fat mass) were associated with the incidence of adenocarcinoma of the lower oesophagus and gastric cardia (gastro-oesophageal adenocarcinoma) and with noncardia gastric adenocarcinoma.
M...