br Conclusion We observed an inverse
Conclusion: We observed an inverse association between BMI and HNC risk, which is consistent with previous studies in other geographic regions. Being underweight at age 20 was also associated with a higher risk of HNC, suggesting that reverse causality was not the main source of the association.
Corresponding author at: Division of Public Health, Department of Family & Preventive Medicine, Huntsman Cancer Institute, University of Utah School of Medicine, 2000 Circle of Hope, Salt Lake City, UT, 84112, USA. E-mail address: [email protected] (M. Hashibe).
Tobacco smoking and alcohol consumption are major risk factors for HNC, and human papilloma 863-61-6 plays a major role in oropharyngeal cancer risk [3–5]. More than half (52.9%) of men smoke cigarettes in China according to the Global Adult Tobacco Survey (GATS) in China in 2010 , but the incidence of HNC among men in China is 4.9/ 100,000, which is relatively low compared to the incidence of HNC (12.8/100,000) among men in the world [1,2]. Meanwhile, other fac-tors associated with HNC risk are of concern since some patients are non-smokers and non-drinkers [3–5].
Most of the previous large-scale studies investigating BMI and HNC risk focused on European-origin populations. The aim of this study was to investigate the role of BMI on HNC risk in an East Asian population, adjusting for potential confounders, including cigarettes smoking, al-cohol drinking, and education. In addition, we aimed to explore the role of BMI at a young age and of BMI change throughout lifetime.
2. Materials and methods
2.1. Study design and population
We conducted a multicenter case-control study in East Asia. Between December 2010 and February 2015, 921 incident HNC cases, including oral cavity, oropharynx, hypopharynx and larynx, and 806 controls were recruited in eight centers (Beijing, Fujian, Henan, Jiangsu, Liaoning, Shanghai, Sichuan, and Taiwan). The participation rates were 85% for cases and 97% for controls. Face-to-face interviews of both cases and controls were structured to obtain information on current and previous alcohol consumption, dietary habits, tobacco consumption including cigarette, pipe, betel quid and tobacco and other lifestyle factors. Written consents for participation were obtained from all study participants. Ethical approval for human subject research was obtained at the University of Utah (University of Utah IRB no. 00041033 11/10/2010), Fujian (Fujian IRB 3/10/2011), Henan (Henan Cancer Hospital 2011), Shanghai (Fudan University IRB no. Cancer Epidemiology 60 (2019) 208–215
The inclusion criteria for cases were 1) age 18–85 years, 2) incident
case of HNC, including the following categories (ICD-0-2): (i) oral cavity (including lip, tongue, gum, floor of mouth, and hard palate):
oropharynx (including base of tongue, lingual tonsil, soft palate, uvula,
performed within six months of cancer diagnosis. Hospital controls
were randomly chosen from subjects admitted as in-patients or out-
patients in the same hospital as the cases, and they were in the hospital
for less than one month when recruited. Controls were frequency-
matched by sex, 5-year age group, ethnicity, and residence area from
hospitals at each of the centers; they were selected among patients with a defined list of non-chronic diseases unrelated to alcohol, tobacco, or
dietary practices. The proportion of controls within a particular diag-
nostic group did not exceed 33%; These groups were 1) benign tumor
2.2. Statistical analyses
Height and weight variables were self-reported and collected in questionnaire-based interviews. There were only three individuals with height missing and three individuals with weight missing. We calcu-lated BMI as weight (kg) divided by height squared (m2). In addition to weight at the time individuals were interviewed, we also collected in-formation on weight 2 years and 5 years before interview and at age 20. BMI change was calculated as the diﬀerence between BMIs from two diﬀerent periods divided by the BMI at the earlier period. Height was categorized into quartiles of the distribution of the study population (160 cm, > 160-167 cm, > 167-172 cm and > 172 cm). Height was also modeled as continuous variables. Trend tests were conducted with height and weight as both categorical variables and continues variables.
Characteristics of cases and controls.
Trend tests for ordered variables were performed by assigning the score j to the j-th exposure level of a categorical variable (where j_1, 2…) and treating it as a continuous predictor in unconditional logistic regres-sion. Trend tests for continuous variables were conducted by treating the variable as a continuous predictor in unconditional logistic re-gression.