br Since some of the interviews took
Since some of the interviews took place 6 months after diagnosis and we had concerns on reverse causality, we restricted the analysis to individuals who were diagnosed within 1 month, and the results were consistent with our overall results. We observed similar levels of risk, for example, being underweight at age 20 was associated with an in-creased risk of HNC (OR = 1.51, 95%CI 1.01, 2.24 overall and OR = 1.63, 95% CI 1.06–2.51 for patients diagnosed within 1 month). Approximately 63.9% of HNC patients were diagnosed within 1 month.
In this study, participants were recruited from eight diﬀerent cen-ters in one of the largest case-control studies on HNC that has been conducted in East Asia. Similar to previous studies [3,7,8,10–14], being
obese at interview was associated with a decreased risk of head and neck cancer, and being underweight at interview was associated with an increased risk of head and neck cancer. The increased risks of HNC associated with a low BMI at interview may be due to reverse causality . One possible explanation of the association is that right before diagnosis, individuals may get dysphagia (diﬃculty in swallowing) or odynophagia (painful swallowing) or may have a change in taste and appetite because of undiagnosed lesions in the head and neck, which may decrease the overall caloric intake and lead to weight loss . In addition, we did not find this association for BMI 2 or 5 years before interview in our study. The INHANCE pooled study of HNC also did not find this association between BMI at 2–5 years before diagnosis and HNC risk . We believe that BMI 2 and 5 years prior to interview may be less impacted by the disability or loss of appetite to eat caused by HNC or early symptoms of HNC.
However, being underweight at age 20 was also associated with a higher HNC risk, and early symptoms would not explain such ob-servations. The association was detected among never cigarette smo-kers and never alcohol drinkers at age 20 in our analysis. A possible explanation for this association is that people who are underweight may also have lower levels of vitamins and other micronutrients, which are associated with increased risks of HNC [15–17]. However, we did not detect any associations between BMI and dietary vegetable and Vitamin K2 intake in this study. The dietary vegetable and fruit intake was assessed for individuals at the time of interview in our study, thus we were unable to assess if they changed their dietary habits during their life-time. Another possible explanation to this was that there has been major socioeconomic development in Asia over the last several decades. The majority of our study population were living on very low incomes at age 20. The GDP per capita in China was approximately 90 US dollars in 1960 compared to about 450 US dollars worldwide, 113 US dollars in China in 1970 compared to about 802 US dollars worldwide and 195 US dollars in China in 1980 compared to about 2,518 US dollars worldwide . The percentage of being underweight among age 7–18 in 1985 was 31.3% in China . Perhaps being underweight was a marker for low socioeconomic status, which has been associated with HNC in previous studies .
Similar to the association observed among never smokers, the higher risk of HNC among individuals who were underweight at age 20 was only observed among never drinkers. These results support the role of low BMI at age 20 in the HNC development via the SES pathway. In the INHANCE pooled analysis, it has been shown that the association with SES was stronger among never tobacco smokers and never alcohol drinkers . Our results suggest that BMI decreasing from age 20 to 5 years before interview was associated with a higher HNC risk. However,
BMI, height, BMI change (as continuous variables) and the risk of HNC by cigarette smoking and drinking status.
Never smoker Ever smoker
Never drinker Ever drinker
a Adjusted for age, sex, ethnicity, education, center, daily cigarettes per day, cigarette years, alcohol drinks per week, alcohol years, betel quid chewed per day and betel years.
b Adjusted for age, sex, ethnicity, education, center, alcohol drinks per week, alcohol years, betel quid chewed per day and betel years.
c Adjusted for age, sex, ethnicity, education, center, daily cigarettes per day, cigarette years, betel quid chewed per day and betel years.
BMI gain showed no association with HNC. When the analysis was stratified by smoking status, BMI loss from age 20 to 5 years before interview was associated with a higher risk of HNC only among never smokers. Reverse causality is a possible explanation for this association, although for the HNC patients diagnosed at older ages, for example at age 60, we would not expect reverse causality to start at age 20.