Alcohol drinking and risks of liver cancer and non-neoplastic chronic liver diseases in China: a 10-year prospective study of 0.5 million adults

Background Alcohol consumption is an important risk factor for hepatic neoplastic and non-neoplastic diseases. Questions remain, however, about the relevance to disease risk of drinking patterns and alcohol tolerability, which differ appreciably between Chinese and Western populations. Methods The prospective China Kadoorie Biobank included 512,715 adults (41% men) aged 30–79 years recruited from ten areas during 2004–2008, recording alcohol intake, drinking patterns, and other characteristics. After median 10 years’ follow-up, 2531 incident liver cancer, 2040 liver cirrhosis, 260 alcoholic liver disease (ALD), and 1262 non-alcoholic fatty liver disease (NAFLD) cases were recorded among 492,643 participants without prior cancer or chronic liver disease at baseline. Cox regression was used to estimate adjusted hazard ratios (HR) relating alcohol intake and drinking patterns to each disease. Results Overall, 33% of men and 2% of women drank alcohol regularly (i.e. at least weekly) at baseline. Among male current regular drinkers, alcohol consumption showed positive dose-response associations with risks of several major chronic liver diseases, with HRs per 280 g/week (i.e. around four drinks/day) higher usual alcohol intake of 1.44 (95% CI 1.23–1.69) for liver cancer (n = 547), 1.83 (1.60–2.09) for liver cirrhosis (n = 388), 2.01 (1.77–2.28) for ALD (n = 200), 1.71 (1.35–2.16) for NAFLD (n = 198), and 1.52 (1.40–1.64) for total liver disease (n = 1775). The association with ALD appeared stronger among men reporting flushing (i.e., with low alcohol tolerance). After adjustment for the total amount of weekly alcohol consumption, daily drinkers had significantly increased risk of ALD (2.15, 1.40–3.31) compared with non-daily drinkers, and drinking without meals was associated with significantly greater risks of liver cancer (1.32, 1.01–1.72), liver cirrhosis (1.37, 1.02–1.85), and ALD (1.60, 1.09–2.33) compared with drinking with meals. Female current regular drinkers had significantly higher risk of ALD, but not other liver diseases, than female abstainers. Conclusions In Chinese men, alcohol intake was associated with significantly increased risks of several major chronic liver diseases, and certain drinking patterns (e.g. drinking daily, drinking without meals) may further exacerbate the disease risks. Supplementary Information The online version contains supplementary material available at 10.1186/s12916-021-02079-1.


Additional File 1 Alcohol drinking and risks of liver cancer and non-neoplastic chronic liver diseases in China: a 10-year prospective study of 0.5 million adults
Pek

Assessment of alcohol consumption
Detailed questionnaire assessment of alcohol consumption has been described previously. 1-4 At the baseline questionnaire, participants were asked how often they had drunk alcohol during the past 12 months (never or almost never, occasionally, only at certain seasons, every month but less than weekly, usually at least once a week). Those who had not drunk alcohol at least weekly in the past 12 months were asked if there was a period of at least a year prior to that when they had drunk some alcohol at least once a week. For this report, participants were classified into four main drinking categories based on the frequency of alcohol drinking during the past 12 months and prior to the past year: (1) abstainers -who had no alcohol use in the past 12 months and had never drunk in most weeks in the past; (2) ex-regular drinkers -who had none or occasional (i.e., occasionally, only at certain seasons, every month but less than weekly) alcohol use in the past 12 months but had previously drunk in most weeks; (3) occasional drinkers -who had occasional alcohol use in the past 12 months but had never drunk in most weeks; and (4) current regular drinkers -who had some alcohol use in most weeks (i.e., ≥weekly) in the past 12 months.  Table S1.

Main analytic models
Cox regression models were used to estimate the associations of alcohol drinking status (reference group: abstainers) with incident liver diseases in all men and women separately, and of alcohol For analyses involving more than two exposure categories, floating SEs were used for the log HRs of all categories including the reference group, enabling comparison between any two categories. 6

Adjustment for regression dilution bias
A gradual regression to the mean over time was observed for alcohol consumption in the subset of participants involved in all three CKB surveys (Table S2). Within-person variation of self-reported alcohol intake was addressed using the regression dilution approach, 7 whereby the usual alcohol intake was taken to be the average intake of the two resurveys in 2008 and 2013-2014, assuming that occasional drinkers consumed 5 g/week. To assess the shape of associations between amount of alcohol intake and liver diseases, the HRs of baseline consumption categories were plotted against their corresponding mean usual alcohol intake. The regression dilution ratio (RDR) was calculated using the assumption-free, non-parametric McMahon-Peto method, 8   Ex-regular drinkers Past 12 months: Never drank alcohol; or had drunk alcohol occasionally, at certain seasons, or monthly but less than weekly. In the past: Had drunk at least weekly.
Occasional drinkers Past 12 months: Had drunk alcohol occasionally, at certain seasons, or monthly but less than weekly. In the past: Had not drunk at least weekly.
Current regular drinkers Past 12 months: At least weekly (i.e., drank alcohol in most weeks). In the past: N/A.

Amount of alcohol consumed and drinking patterns among current regular drinkers (i.e., drank ≥weekly in the past 12 months)
Weekly intake (g per typical drinking week) Calculated based on the beverage type a and amount drunk a on a typical drinking day, and drinking frequency in the past year. In categories: Men: <140; 140-279; 280-419; ≥420 g/week. Women: <70; 70-139; ≥140 g/week. As a continuous variable: per 280 g/week higher intake.

Drinking frequency (in the past year)
During the past 12 months, on how many days did you drink alcohol in a typical week? Daily: 6-7 days/week (i.e., "daily or almost every day"). Non-daily: "1-2 days/week" or "3-5 days/week".
Types of alcohol (on a typical drinking day) a Derived from "On days when you drink, how much alcohol do you usually drink in a day?", for which participants had to choose one beverage type to report the amount usually drunk on a typical drinking day. Spirits: "Strong spirit (≥40% alcohol)" or "weak spirit (<40% alcohol)". Non-spirits: "Rice wine", "grape wine", or "beer".
Heavy episodic drinking (on a typical drinking day) a Derived based on the amount of alcohol consumed on a typical drinking day. Men: Consumption of >60 g (i.e., 7.5 UK units or 4.3 US standard drinks) of alcohol on a typical drinking day. Women: Consumption of >40 g (i.e., 5 UK units or 2.9 US standard drinks) of alcohol on a typical drinking day. 11 Drinking with/outside of meals (on a typical drinking day) On a typical day when you drink alcohol, when do you usually take the drink? With meals: "Usually drank with meals". Outside of meals: "Usually drank between or after meals" or "no regular patterns".

Duration of regular drinking
Number of years of regular drinking calculated by the difference between baseline age (years) and age started regular drinking (years).

Flushing response
After drinking alcohol, do you usually experience hot flushes or dizziness? No: "Yes, but only after drinking a large amount of alcohol" or "No". Yes: "Yes, soon after the first mouthful" or "Yes, after drinking a small amount of alcohol". a Data was available on a typical drinking day, on special occasions, and the last time the participants drank. For this study, alcohol data reported on a typical drinking day was used in the analyses to reflect the usual drinking habits of the participants. Usual consumption is the average alcohol intake of the two resurveys. a Regression dilution ratio is estimated among baseline current regular drinkers using baseline and usual alcohol intakes. The regression dilution ratios calculated using the MacMahon-Peto method based on baseline survey and first resurvey are 0.57 for men and 0.64 for women, and those based on baseline and second resurvey are 0.52 for men and 0.56 for women.     Figure S1. Spline curves of adjusted HRs for total and major chronic liver diseases associated with alcohol intake over the range of 0-1500 g/week, in male current regular drinkers Cox models are stratified by age-at-risk, study area, and HBsAg, and adjusted for education, household income, smoking, physical activity, and body mass index.  Figure S2. Association of alcohol consumption with total liver disease, in female current regular drinkers Cox models are stratified by age-at-risk, study area, and HBsAg, and adjusted for education, household income, smoking, physical activity, and body mass index. Each solid square represents an HR. The size of each box is inversely proportional to the "floated" variance of the log hazard in each group and the error bars indicate the group-specific 95% CIs. Group-specific 95% CIs are plotted using floating standard errors to allow for comparison between any two categories. The numbers above the error bars are point estimates for HRs, and the numbers below are number of events. Usual alcohol intake is calculated by the average of the self-reported alcohol intake at the two resurveys in each group. HR per 280 g/week increment in usual alcohol intake = 1.34 (95% CI 0.73-2.46). HR, hazard ratio; CI, confidence interval; HBsAg, hepatitis B surface antigen.

Figure S3. Adjusted HRs per 280g/week higher usual alcohol intake for liver cancer, liver cirrhosis and non-alcoholic fatty liver disease, by population subgroups in male current regular drinkers
Cox models are stratified by age-at-risk, study area, and HBsAg, and adjusted for education, household income, smoking, physical activity, and body mass index, where appropriate. Each solid square represents an HR. The size of each box is inversely proportional to the variance of the log HR and the error bars indicate the 95% CIs. HR, hazard ratio; CI, confidence interval; HBsAg, hepatitis B surface antigen.

Figure S4. Associations of the joint effects of alcohol consumption and body mass index with total and major chronic liver diseases, in male current regular drinkers
Conventions are as in Figure S2.

Usual alcohol intake (g/w eek) (D) Total liver disease (N=1775)
HBsAg -HBsAg + Figure S6. Adjusted HRs per 280 g/week higher usual alcohol intake for total and major chronic liver diseases, by HBsAg seroprevalence in male current regular drinkers Non-alcoholic fatty liver disease is not shown here due to low number of events in HBsAg positive participants (n=4). Conventions are as in Figure S3.   Participants with unclear or missing HBsAg test result, or with self-reported prior cancer, liver cirrhosis, or chronic hepatitis were excluded. a Cox models are stratified by age-at-risk, study area, and HBsAg, and adjusted for education, household income, smoking, body mass index, and physical activity.

Figure S7. Associations of the joint effects of alcohol consumption and mealtime drinking habits with total and major chronic liver diseases, in male current regular drinkers
Non-alcoholic fatty liver disease is not shown here due to the low number of events among men usually drinking without meals (n=12). Conventions are as in Figure S2. With meals Without meals Figure S8. Adjusted HRs per 280 g/week higher usual alcohol intake for total and major chronic liver diseases, by beverage type in male current regular drinkers Conventions are as in Figure S3.

Figure S9. Associations of alcohol consumption with total and major chronic liver diseases, in all men
Conventions are as in Figure S2.    HR, hazard ratio; CI, confidence interval; ICD-10, international classification of diseases, 10 th revision. Participants with unclear or missing HBsAg test result, or with self-reported prior cancer, liver cirrhosis, or chronic hepatitis were excluded. Cox models are stratified by age-at-risk, study area, and HBsAg, and adjusted for education, household income, smoking, physical activity, and body mass index. Attributable fraction is calculated as P(HR-1)/HR, where P is the prevalence of ever-regular alcohol consumption (i.e., current and ex-regular drinking) among those who developed the relevant liver disease during follow-up. a Included all non-neoplastic liver diseases (ICD-10: B18-B19, B94.2, K70-K77, Z22.5).