Objectives: To determine whether low or high extremes of body mass index (BMI) in otherwise healthy individuals affect mortality only after they develop cancer or affect the likelihood that cancer will occur.
Methods: We conducted a cohort analysis on the possible association between BMI and the risk of total cancer incidence and mortality among a middle-aged Japanese population consisting of a population-based cohort of 88,927 subjects (42,093 men and 46,834 women) with a 10-year follow-up.
Results: In men, a U-shaped association between BMI and cancer occurrence was observed, with men with a BMI of 23.0-24.9 having the lowest risk of cancer occurrence (BMI 14.0-18.9: HR = 1.29, 95% CI = 1.08-1.54; BMI 30.0-39.9: HR = 1.22, 95% CI = 0.92-1.61). This tendency did not change substantially after excluding cases diagnosed early during the follow-up period; cancer mortality showed a similar trend but with higher risk values. When analyzed according to smoking category, a low BMI affected cancer occurrence more strongly among current smokers than in never-smokers. Unlike men, no marked fluctuation in risk was observed in women.
Conclusions: A very low BMI seems to have an impact on the total cancer risk in populations with a low average BMI. Therefore, while much attention has been given to the effects of obesity, the health effects of both extreme ends of BMI should be taken into consideration in populations with a low average BMI.