Abstract
Background
Dietary glycemic index (GI) and glycemic load (GL) typically have a positive relationship with obesity and diabetes, which are risk factors for liver cancer. However, studies on their association with liver cancer have yielded inconsistent results. Therefore, we assessed the association of GI, GL, and carbohydrates with liver cancer risk.
Patients and methods
A total of 72 966 women and 60 207 men from the Shanghai Women's Health Study (SWHS) and the Shanghai Men's Health Study (SMHS) were included for analysis. Food frequency questionnaire (FFQ) data were used to calculate daily dietary GI, GL, and carbohydrate intake. These values were energy adjusted and categorized into quintiles. The hazard ratios (HRs) and 95% confidence intervals (CI) were calculated with adjustment for potential confounders.
Results
After a median follow-up time of 11.2 years for the SWHS and 5.3 years for the SMHS, 139 and 208 incident liver cancer cases were identified in the SWHS and SMHS, respectively. In multivariable Cox regression models, no statistically significant trends by quintile of GI, GL, or carbohydrate intake were observed. Stratification by chronic liver disease/hepatitis, diabetes, or body mass index (BMI) did not alter the findings.
Conclusions
There is little evidence that dietary GI, GL, or carbohydrates affect the incidence of liver cancer in this Asian population.
Keywords: Chinese men and women, cohort study, diet, glycemic index, glycemic load, primary liver cancer
introduction
Primary liver cancer is a serious global cancer burden as it is the fifth most commonly diagnosed cancer among men and seventh among women. This cancer also has a high fatality rate and is the second most common cause of cancer death among men and sixth among women [1]. China represents >50% of both incident cases and cancer deaths of liver cancer worldwide [1]. Given the high incidence and poor prognosis, primary prevention of liver cancer is extremely important. The established risk factors for liver cancer include the hepatitis B virus (HBV), hepatitis C virus (HCV), cirrhosis, heavy alcohol consumption, dietary aflatoxins, and tobacco smoking [2–6]. Overweight, obesity, diabetes mellitus and a number of other factors affiliated with metabolic syndrome are related risk factors that have also been associated with liver cancer [7–11].
The glycemic index (GI) and glycemic load (GL) were developed to quantify the effect of specific carbohydrates on blood glucose and a number of studies have reported that higher levels of dietary GI or GL increase the risk of diabetes mellitus or metabolic syndrome [12–15]. Since diabetes has been found to be associated with liver cancer [7, 8, 10, 16], it has been hypothesized that dietary GI, GL, and carbohydrates may also be associated with liver cancer. The association of GI, GL, and carbohydrates with cancer has been previously evaluated for a number of cancers, including breast, colorectal, and pancreatic cancers [17–26].
Only a few studies have been conducted to assess the association between GI or GL and liver cancer with inconsistent findings [27–29]. Therefore, we aimed to assess the association of dietary GI, GL, and carbohydrates with liver cancer using data from two prospective cohort studies, the Shanghai Women's Health Study (SWHS) and the Shanghai Men's Health Study (SMHS).
methods
study population
The SWHS and SMHS are both prospective, population-based cohort studies in Shanghai, China, with details of their designs published previously [30, 31]. In brief, for the SWHS, 74 941 women aged 40 to 70 years in Shanghai were recruited from 1996 to 2000, with an overall study participation rate of 92.7%. For the SMHS, 61 491 men aged 40 to 74 years with no previous history of cancer were recruited in Shanghai from 2002 to 2006, with an overall study participation rate of 74.1%. Trained interviewers administered the baseline surveys and obtained anthropometric measurements of the participants. The two cohorts were approved by all relevant Institutional Review Boards and an informed consent was obtained from all participants.
For this analysis, we excluded women with a previous diagnosis of cancer (N = 1 579) from the SWHS, but the SMHS had a previous diagnosis of cancer as an exclusion criterion so no additional men were excluded for this reason. We excluded men and women with extreme total energy intake (<500 or >4200 kcal per day for men or <500 or >3500 kcal per day for women), participants lost to follow-up shortly after enrollment, and those diagnosed with cancers of unknown origin (N = 233 for SMHS and N = 326 for SWHS). We also excluded participants with missing data for any of the covariates of interest (N = 1065 for SMHS and N = 78 for SWHS). After the application of these exclusion criteria, 60 207 men and 72 966 women were available for analysis.
measurement of GI, GL, and total carbohydrates
Typical dietary intake was assessed using a previously validated food frequency questionnaire (FFQ) [32, 33]. The FFQ in the SWHS included 77 food groups and items and accounted for 85.6% of the typical foods consumed by this population [32]. The FFQ in the SMHS included 81 food groups and items and captured 88.8% of the typically consumed foods in the population [33]. During the in-person interview, the participants were asked about the frequency and amount of consumption of each food item included in the FFQs over the past 12 months. The daily nutrient intake was calculated by multiplying the daily intake of each food by the nutrient content per gram of that food as derived from the Chinese Food Composition Tables [34] and summed over all food items.
The details of the reliability and validity of the FFQs utilized in the two cohorts have been reported previously [32, 33]. The FFQs from both the SWHS and SMHS had reasonable validity and reproducibility with correlation coefficients between the FFQ and 24 h dietary recalls of 0.64 for carbohydrate and 0.63 for rice estimates in the SMHS [33] and 0.66 for both carbohydrates and rice in the SWHS [32]. The GI values for the major carbohydrate-containing foods (54 foods for SWHS and 62 foods for SMHS), using glucose as the reference, were obtained from the Chinese Food Composition Tables, the International Table of GI and GL Values and the Harvard Database [34–36]. The GL for each food was calculated by multiplying the food's GI value by the carbohydrate content for the food and the average amount of this food consumed daily. The products were summed over all of the food items to calculate the dietary GL. The dietary GI was calculated as the total dietary GL divided by the average carbohydrate intake for that individual. Table 1 presents the top 10 foods associated with total dietary GL in this population. For both the SWHS and SMHS, rice and noodles, steamed bread, dumplings, and other wheat foodstuffs made up the majority of the daily carbohydrate intake. The calculated values of dietary GI, GL, and total carbohydrates were adjusted for total energy intake using the residuals method [37] and standardized to 1700 kcal for women and 2000 kcal for men, the approximate mean daily caloric intake calculated from the FFQ validation studies [32, 33]. We then categorized the participants into quintiles of total energy-adjusted GI, GL, and carbohydrates separately by cohort using the baseline data.
Table 1.
Top 10 food contributors to glycemic load (GL) values
| Rank | Daily % of GL | Glycemic index (GI) value | |
|---|---|---|---|
| Women (Shanghai Women's Health Study, SWHS) | |||
| Rice | 1 | 80.9 | 83.1 |
| Noodles, steamed bread, and other wheat foodstuffs | 2 | 6.8 | 55.9 |
| All kinds of desserts | 3 | 3.1 | 60.6 |
| Bread | 4 | 2.2 | 68.0 |
| Watermelon | 5 | 1.3 | 72.0 |
| Apples | 6 | 0.7 | 36.0 |
| Candy and preserved fruit | 7 | 0.6 | 58.3 |
| Potatoes | 8 | 0.6 | 66.4 |
| Pears | 9 | 0.4 | 36.0 |
| Bananas | 10 | 0.4 | 41.0 |
| Men (Shanghai Men's Health Study, SMHS) | |||
| Rice | 1 | 81.4 | 83.1 |
| Noodles, steamed bread, dumplings, and other wheat foodstuffs | 2 | 11.1 | 55.9 |
| Bread | 3 | 1.8 | 68.0 |
| All kinds of desserts | 4 | 1.7 | 60.6 |
| Watermelon | 5 | 0.8 | 72.0 |
| Potatoes | 6 | 0.6 | 66.4 |
| Apples | 7 | 0.3 | 36.0 |
| Bananas | 8 | 0.2 | 41.0 |
| Tangerines, oranges, and grapefruits | 9 | 0.2 | 43.0 |
| Preserved fruits | 10 | 0.2 | 58.3 |
Note: The GI values used glucose as the reference.
outcome
The outcome of this analysis was incident liver cancer and defined as a primary tumor with an ICD-9 code of 155 (malignant neoplasm of liver and intrahepatic bile ducts). In brief, the participants were followed up with in-person interviews every 2 to 3 years and through annual record linkage with the population-based Shanghai Cancer Registry for incident cancer diagnoses and the Vital Statistics Unit in the Shanghai Center of Disease Prevention and Control for death confirmation. Participants identified as incident cancer cases within the Shanghai Cancer Registry were verified through home visits and medical abstracts were obtained to verify the cancer diagnosis and to document detailed diagnostic information like ICD-0-3 site/histology information. For the SWHS, the first follow-up was conducted from 2000 to 2002 with a response rate of 99.8%. For the SMHS, the first follow-up interviews were conducted from 2004 to 2008 and had a response rate of 97.6%. Follow-up has been conducted for both the studies through December 31, 2009.
potential confounding variables
Demographic variables of interest for this analysis were age, sex, education level, and family income. In the SWHS, annual family income was assessed, whereas in the SMHS, annual per capita family income was requested. The body mass index (BMI) was calculated from the measured height and weight of each participant from the baseline visit (kg/m2). For stratification analyses, this variable was dichotomized with the overweight/obese category with BMI ≥ 25.0 kg/m2 and the underweight/normal weight category with BMI < 25.0. Behavioral characteristics included in this analysis were cigarette smoking, alcohol consumption, menopausal status and hormone therapy use (women only), and amount of exercise per week (MET hours/week). Dietary characteristics included carbohydrate, total fat, and total energy intake (grams per day) obtained from the FFQ. We determined the history of diabetes and history of hepatitis or chronic liver disease from self-reports by participants.
statistical analysis
We conducted the analyses separately for each cohort. We calculated descriptive statistics based on baseline GL quintile using analysis of variance for continuous variables and the Pearson chi-square test for categorical variables. We determined the hazard ratios (HRs) for incident liver cancer using the Cox proportional hazards regression model where the entry time was the date at which the participant enrolled in the SMHS or SWHS and the exit time was the date when the participant developed incident liver cancer or was censored due to death, loss to follow-up, or end of study follow-up on December 31, 2009, using whichever censoring date occurred first. We calculated age-adjusted HRs (model 1), multivariable HRs adjusting for all potential confounders (age, education, income, smoking status, alcohol consumption, menopausal status for women, family history of liver cancer, BMI, physical activity, total energy intake, and history of diabetes and hepatitis/chronic liver disease; model 2) and also created multivariable Cox models stratified by having a history of diabetes, history of hepatitis or chronic liver disease, and BMI category. To evaluate linear trends, we entered the median level of dietary GI, GL, or carbohydrates by quintile into the model as a continuous variable. We evaluated the proportional hazards assumption by including an interaction between dietary GI, GL, or carbohydrates with the logarithm of time. The interaction was not statistically significant; therefore we assumed that the proportional hazards assumption was not violated.
As a secondary analysis, we excluded liver cancer cases diagnosed within 2 years of baseline to assess the effect of potential prevalent liver cancer cases. We also used both baseline and follow-up dietary GI, GL, and carbohydrate values as a time varying covariate in an additional secondary analysis. Additionally, we calculated multivariable Cox models with hepatocellular carcinoma (HCC) as the outcome (ICD-0-3: 817). Statistical analyses were conducted using SAS 9.2 (SAS Institute, Cary, NC) and a two-sided P value of 0.05 was considered statistically significant.
results
After a median follow-up time of 11.2 years for the SWHS and 5.3 years for the SMHS, 139 and 208 incident liver cancer cases were identified in the SWHS and SMHS, respectively. Of these cases, 84.7% in the SWHS and 93.8% in the SMHS were HCC. Table 2 presents the baseline characteristics for the two cohorts by quintiles of GL. Women with higher GL intake were more likely to be older and post-menopausal, have less education, and have lower annual family incomes. They were also more likely to smoke and have a higher BMI and daily total energy consumption, but were less likely to exercise, consume alcohol, or to have a history of diabetes, hepatitis, or chronic liver disease. A generally similar pattern was observed for men with higher GL intake.
Table 2.
Characteristics of study participants by baseline energy adjusted quintiles of glycemic load (GL)
| Quintile of GL |
||||||
|---|---|---|---|---|---|---|
| 1 | 2 | 3 | 4 | 5 | P value | |
| Women (Shanghai Women's Health Study, SWHS) | ||||||
| Number of subjects | 14593 | 14593 | 14593 | 14593 | 14594 | |
| Age (years) | 50.7 ± 8.5 | 51.4 ± 8.7 | 52.1 ± 9.0 | 53.2 ± 9.2 | 55.2 ± 9.2 | <0.0001 |
| Educational level (%) | ||||||
| ≤Elementary school | 9.6% | 13.5% | 18.9% | 24.9% | 39.8% | <0.0001 |
| Middle school | 35.1% | 36.0% | 38.1% | 39.5% | 37.2% | |
| High school | 35.4% | 33.7% | 28.9% | 24.8% | 17.1% | |
| ≥College | 19.9% | 16.8% | 14.1% | 10.8% | 5.9% | |
| Annual family income (%) | ||||||
| <10 000 yuan | 12.5% | 13.2% | 15.1% | 17.5% | 22.0% | <0.0001 |
| 10 000–19 999 yuan | 33.9% | 36.5% | 37.6% | 40.3% | 42.9% | |
| 20 000–29 999 yuan | 29.7% | 29.9% | 29.5% | 27.2% | 24.4% | |
| ≥30 000 yuan | 23.9% | 20.4% | 17.8% | 15.0% | 10.8% | |
| Cigarette smoking (%) | 2.6% | 2.1% | 2.6% | 2.6% | 3.9% | <0.0001 |
| Alcohol consumption (%) | 3.7% | 2.1% | 1.9% | 1.6% | 1.8% | <0.0001 |
| Post-menopausal (%) | 40.4% | 43.8% | 47.2% | 51.9% | 61.2% | <0.0001 |
| Body mass index (BMI) (kg/m2) | 23.7 ± 3.2 | 23.6 ± 3.3 | 23.8 ± 3.4 | 24.1 ± 3.4 | 24.9 ± 3.7 | <0.0001 |
| Exercise (MET hours/week) | 0.8 ± 1.8 | 0.7 ± 1.6 | 0.6 ± 1.5 | 0.6 ± 1.5 | 0.6 ± 1.5 | <0.0001 |
| Total energy intake (kcal/day) | 1764.4 ± 453.0 | 1635.7 ± 368.7 | 1600.5 ± 341.7 | 1593.3 ± 347.1 | 1775.7 ± 407.0 | <0.0001 |
| History of diabetes (%) | 5.3% | 4.4% | 4.4% | 4.3% | 3.1% | <0.0001 |
| History of hepatitis/chronic liver disease (%) | 3.6% | 3.5% | 3.3% | 3.3% | 2.8% | 0.0039 |
| Family history of liver cancer (%) | 3.2% | 3.4% | 3.3% | 3.3% | 3.2% | 0.7815 |
| Men (Shanghai Men's Health Study, SMHS) | ||||||
| Number of subjects | 12041 | 12041 | 12042 | 12041 | 12042 | |
| Age (years) | 54.8 ± 9.6 | 55.5 ± 9.8 | 55.6 ± 9.8 | 55.6 ± 9.9 | 55.2 ± 9.6 | <0.0001 |
| Educational level (%) | ||||||
| ≤Elementary school | 4.1% | 4.8% | 6.5% | 7.9% | 10.2% | <0.0001 |
| Middle school | 29.6% | 29.8% | 32.2% | 35.3% | 40.8% | |
| High school | 37.2% | 36.8% | 36.3% | 35.9% | 34.4% | |
| ≥College | 29.1% | 28.6% | 25.0% | 20.8% | 14.7% | |
| Annual per capita family income (%) | ||||||
| <6000 yuan | 12.1% | 10.3% | 10.9% | 12.4% | 17.7% | <0.0001 |
| 6000–11 999 yuan | 37.8% | 39.6% | 41.9% | 45.2% | 48.9% | |
| 12 000–23 999 yuan | 36.7% | 38.3% | 37.4% | 34.5% | 28.3% | |
| ≥24 000 yuan | 13.5% | 11.9% | 9.9% | 7.9% | 5.2% | |
| Cigarette smoking (%) | 71.7% | 68.7% | 68.5% | 69.7% | 70.3% | <0.0001 |
| Alcohol consumption (%) | 45.9% | 37.3% | 33.2% | 28.6% | 23.5% | <0.0001 |
| BMI (kg/m2) | 23.7 ± 3.1 | 23.7 ± 3.1 | 23.6 ± 3.0 | 23.7 ± 3.1 | 23.9 ± 3.1 | <0.0001 |
| Exercise (MET hours/week) | 1.2 ± 2.3 | 1.0 ± 2.1 | 0.9 ± 1.9 | 0.9 ± 2.0 | 0.8 ± 1.9 | <0.0001 |
| Total energy intake (kcal/day) | 2020.0 ± 524.2 | 1823.0 ± 435.0 | 1789.3 ± 439.0 | 1830.0 ± 461.0 | 2073.1 ± 455.8 | <0.0001 |
| History of diabetes (%) | 9.0% | 7.8% | 6.5% | 5.0% | 3.1% | <0.0001 |
| History of hepatitis/chronic liver disease (%) | 6.2% | 5.7% | 5.3% | 5.7% | 4.6% | <0.0001 |
| Family history of liver cancer (%) | 3.8% | 3.8% | 3.6% | 3.5% | 3.2% | 0.1139 |
Note: Continuous variables are presented as the mean ± the standard deviation.
As seen in Table 3, among women, after adjustment for age, compared with the lowest quintile of GL, the HRs for incident liver cancer were 0.94, 1.04, 1.19, and 1.13 for the second through the fifth quintile, respectively (Ptrend = 0.4622) with all 95% confidence intervals (CIs) including the null value of 1.0. Similarly, all of the 95% CIs for the age-adjusted HRs for quintiles of carbohydrate intake included 1.0 (Ptrend = 0.6659), but the association estimates were <1.0. The results were similar for the GL and carbohydrates after adjustment. For the GI, compared with the first quintile, the HRs for incident liver cancer were 2.01 (95% CI 0.98–4.12), 3.01 (95% CI 1.53–5.90), 2.22 (95% CI 1.11–4.44), and 2.41 (95% CI 1.23–4.74) for the second through the fifth quintile, respectively, after adjustment for age (Ptrend = 0.0202). After additional adjustment, the Ptrend was no longer significant (Ptrend = 0.0621), but many of the individual association estimates were still significant. For men, the trends were not statistically significant for the quintiles of GL (Model 1 Ptrend = 0.7496), GI (Model 1 Ptrend = 0.4896), or carbohydrates (Model 1 Ptrend = 0.8694) and the association estimates ranged around 1.0 (Table 3).
Table 3.
Hazard ratios (HRs) for liver cancer incidence by quintiles of glycemic load (GL), glycemic index (GI), and carbohydrates
| Quintile of intake |
||||||
|---|---|---|---|---|---|---|
| 1 | 2 | 3 | 4 | 5 | P for trend | |
| GL | ||||||
| Women | ||||||
| Median intake | 166.3 | 190.0 | 205.1 | 219.3 | 241.9 | |
| Incident liver cancer cases | 22 | 22 | 26 | 33 | 36 | |
| Model 1 [95% confidence interval (CI)] | 1.00 | 0.94 (0.52, 1.69) | 1.04 (0.59, 1.84) | 1.19 (0.69, 2.05) | 1.13 (0.66, 1.93) | 0.4622 |
| Model 2 (95% CI) | 1.00 | 0.90 (0.50, 1.63) | 0.97 (0.54, 1.72) | 1.09 (0.62, 1.90) | 1.02 (0.59, 1.79) | 0.7477 |
| Men | ||||||
| Median intake | 194.4 | 222.8 | 240.7 | 258.3 | 286.0 | |
| Incident liver cancer cases | 40 | 50 | 42 | 34 | 42 | |
| Model 1 (95% CI) | 1.00 | 1.23 (0.81, 1.87) | 1.04 (0.68, 1.61) | 0.85 (0.54, 1.34) | 1.07 (0.70, 1.66) | 0.7496 |
| Model 2 (95% CI) | 1.00 | 1.28 (0.84, 1.96) | 1.07 (0.68, 1.66) | 0.82 (0.51, 1.30) | 1.07 (0.68, 1.67) | 0.6514 |
| GI | ||||||
| Women | ||||||
| Median intake | 63.9 | 68.5 | 71.2 | 73.6 | 76.8 | |
| Incident liver cancer cases | 11 | 23 | 37 | 30 | 38 | |
| Model 1 (95% CI) | 1.00 | 2.01 (0.98, 4.12) | 3.01 (1.53, 5.90) | 2.22 (1.11, 4.44) | 2.41 (1.23, 4.74) | 0.0202 |
| Model 2 (95% CI) | 1.00 | 1.93 (0.94, 3.96) | 2.85 (1.45, 5.61) | 2.06 (1.02, 4.15) | 2.17 (1.08, 4.35) | 0.0621 |
| Men | ||||||
| Median intake | 64.4 | 69.0 | 71.7 | 74.2 | 77.2 | |
| Incident liver cancer cases | 47 | 44 | 40 | 34 | 43 | |
| Model 1 (95% CI) | 1.00 | 0.96 (0.64, 1.44) | 0.89 (0.58, 1.35) | 0.76 (0.49, 1.18) | 0.95 (0.63, 1.43) | 0.4896 |
| Model 2 (95% CI) | 1.00 | 0.99 (0.66, 1.50) | 0.90 (0.59, 1.37) | 0.72 (0.46, 1.12) | 0.89 (0.58, 1.37) | 0.2943 |
| Carbohydrates | ||||||
| Women | ||||||
| Median intake | 253.6 | 276.8 | 290.6 | 303.5 | 323.4 | |
| Incident liver cancer cases | 29 | 23 | 28 | 22 | 37 | |
| Model 1 (95% CI) | 1.00 | 0.74 (0.43, 1.27) | 0.86 (0.51, 1.44) | 0.62 (0.36, 1.08) | 0.92 (0.56, 1.50) | 0.6659 |
| Model 2 (95% CI) | 1.00 | 0.72 (0.41, 1.25) | 0.80 (0.47, 1.36) | 0.58 (0.33, 1.03) | 0.85 (0.51, 1.41) | 0.4520 |
| Men | ||||||
| Median intake | 292.5 | 321.3 | 337.9 | 353.8 | 377.7 | |
| Incident liver cancer cases | 37 | 56 | 36 | 36 | 43 | |
| Model 1 (95% CI) | 1.00 | 1.47 (0.97, 2.23) | 0.94 (0.60, 1.49) | 0.94 (0.60, 1.49) | 1.16 (0.75, 1.81) | 0.8694 |
| Model 2 (95% CI) | 1.00 | 1.51 (0.99, 2.30) | 0.96 (0.60, 1.54) | 0.94 (0.58, 1.50) | 1.15 (0.73, 1.81) | 0.8167 |
Model 1: Adjusted for age.
Model 2: Adjusted for age, education, income, smoking status, alcohol consumption, menopausal status (women only), family history of liver cancer, BMI, physical activity, total energy intake, and history of diabetes and hepatitis/chronic liver disease.
Note: For the Shanghai Women's Health Study (SWHS), the cut-points for quintiles of GL were 180.4, 197.8, 211.9, and 228.2, for GI were 66.6, 69.9, 72.4, and 75.0, and for carbohydrates were 267.8, 284.1, 296.8, and 311.5. For the Shanghai Men's Health Study (SMHS), the cut-points for quintiles of GL were 211.5, 232.2, 249.1, and 269.6, for GI were 67.2, 70.4, 72.9, and 75.5, and for carbohydrates were 310.1, 330.1, 345.6, and 363.6.
When stratified by chronic liver disease and hepatitis status, no statistically significant trends were observed for men or women with a history of chronic liver disease or hepatitis. For women without a history of chronic liver disease or hepatitis, however, a significant trend (Ptrend = 0.0107) of increasing liver cancer risk with increasing GI (HR 5th quintile versus 1st quintile 2.57; 95% CI 1.23–5.36) was observed in the age-adjusted model, which remained after adjustment for all potential confounders (Ptrend = 0.0413). No other trends were observed for men or women without a history of chronic liver disease or hepatitis (supplementary Table S1a and b, available at Annals of Oncology online). Similarly, when stratified by history of diabetes, no statistically significant trends were observed for men or women, except for the association between GI and liver cancer among women without a history of diabetes, but only in age-adjusted analyses (Model 1 Ptrend = 0.0257; Model 2 Ptrend = 0.0724) (supplementary Table S2a and b, available at Annals of Oncology online). When the data were stratified by BMI category, no statistically significant trends were observed for men or women (supplementary Table S3a and b, available at Annals of Oncology online).
After removal of participants who developed liver cancer within 2 years of baseline (N = 29 for SWHS and N = 82 for SMHS), similar associations were observed for GL, GI, and carbohydrates (supplementary Table S4, available at Annals of Oncology online). When quintiles of GI, GL, and carbohydrates were treated as time-varying covariates, none of the associations between GL, GI, and carbohydrates were statistically significant for either sex with the exception of the second quintile of carbohydrates versus the first quintile among men (supplementary Table S5, available at Annals of Oncology online). When incident HCC was used as the outcome, some of the associations were strengthened, but no statistically significant trends were observed (supplementary Table S6, available at Annals of Oncology online).
discussion
To our knowledge, this is the first study considering the association of GI, GL, and carbohydrates with liver cancer risk in the Chinese population. However, there is little evidence that dietary GI, GL, or carbohydrates affect the incidence of liver cancer in this population. Most of the associations with GI, GL, or carbohydrates were not statistically significant, with the exception of the association between GI and liver cancer among women. However, when GI, GL, and carbohydrates were entered as time-varying covariates, nearly all of the observed associations either remained non-significant or were closer to the null and no longer statistically significant. Stratification analyses did not give a strong indication for effect modification by chronic liver disease, hepatitis status, history of diabetes, or BMI. In addition, dose–response relationships were generally not observed for any of the associations.
Previously, a case–control study in Athens, Greece, found an overall positive association between quintiles of GL and liver cancer, although the findings were not statistically significant [27]. Similarly, a case–control study in Italy found a positive association between quintiles of GL and liver cancer with statistically significant results [28]. Both the studies found a stronger association between GL and liver cancer when restricted to cases with evidence of chronic HBV or HCV infection [27, 28]. However, our findings do not support these case–control study results as we did not find strong evidence for an overall association between GL and liver cancer nor did we find evidence for an association specifically among patients with chronic hepatitis or liver disease. Due to the case–control nature of both of the studies, recall bias in reporting dietary intake may have occurred, especially for cases, since they may associate diet with their disease status. Similarly, cases may have recently changed their diet due to symptoms or treatment and therefore the FFQ results would not represent habitual intake. However, it is also possible that general dietary differences between the populations, European versus Asian diet, may have led to the different findings. In our study, over 80% of the daily dietary GL was contributed by rice in both the men and women's study compared with only 17.0% in men and 12.9% in women observed in a study of people living in Hawaii or southern California [38].
One prospective cohort study considering the association of GI and GL with multiple cancers, conducted in the USA, did not find any statistically significant association between GI and liver cancer for women, but a positive association was observed for men with a non-significant trend (P = 0.185). A statistically significant protective relationship was observed between GL and liver cancer for both women and men. However, the authors explain that the site-specific associations should be interpreted with caution due to the multiple testing and that the association between GL and liver cancer in women disappeared after restriction to never smokers, which suggests possible residual confounding [29]. The authors conclude that there is no strong evidence for a GI, GL and cancer relationship, which is in agreement with the conclusions from this current study.
Our study is not without limitations. First, GI, GL, and carbohydrate intakes were all assessed using an FFQ which may not be accurate at estimating the actual amount of dietary intake. However, the correlation between the estimated carbohydrate intake from an FFQ and a 24-h recall instrument was high in both study populations (SWHS = 0.66; SMHS = 0.64) [32, 33]. In addition, we excluded participants who had extreme energy intake to remove participants who may not have been accurately reporting nutritional intake. A second limitation of the FFQ was that data on only adult nutritional intake were available for analysis. It is possible that the effect of diet on the risk of cancer occurs earlier in life, but these data were not available, so we were unable to test that hypothesis. Similarly, data from only two FFQs were available for analysis so changes over the entire follow-up were not able to be considered. Third, data on acute hepatitis infection were unavailable, so we were only able to adjust for chronic hepatitis infection and chronic liver disease. Additionally, data on these chronic conditions were only available by self-report, so under-reporting of chronic hepatitis or liver disease was possible. Finally, the SMHS had a shorter follow-up than the SWHS, but this population still had a sufficiently large number of incident liver cancer cases to conduct the analyses.
This study, however, has a number of important strengths. First, the SWHS and SMHS are both rigorously designed cohort studies with high participation and retention rates. Second, all of the covariates used in analysis were assessed before the development of any cancer which decreases potential misclassification bias. We also determined that prevalent liver cancer was unlikely to have affected the results by excluding participants diagnosed with liver cancer within 2 years of baseline. Third, we had a relatively large number of incident liver cancer cases to base our analysis. Finally, the results from the various secondary analyses we conducted yielded similar results throughout which suggest that our findings are fairly robust.
In conclusion, there is very little evidence from this prospective study that GI, GL, or carbohydrate intake affects the risk of liver cancer in this population, where rice is the principal component of total carbohydrate intake. Further research to find other modifiable risk factors for liver cancer may be warranted.
funding
This work was supported by funds from the State Key Project Specialized for Infectious Diseases of China [No. 2008ZX10002-015 and 2012ZX10002008-002]; the United States National Institutes of Health [R37 CA070867, R01 CA82729, and R01 HL095931]; the Fogarty International Clinical Research Scholars and Fellows Program at Vanderbilt University [R24 TW007988-5 to EV and HLL]; and the Cancer Prevention and Control Training Program at the University of Alabama at Birmingham funded through the National Institutes of Health [5R25 CA047888 to EV]. The funding organizations had no role in the design and conduct of the study; the collection, analysis, and interpretation of the data; or the preparation, review, or approval of the manuscript.
disclosure
The authors have declared no conflicts of interest.
Supplementary Material
acknowledgements
We would like to thank the participants and the staff from the SWHS and SMHS for their contribution to this research.
references
- 1.Ferlay J, Shin HR, Bray F, et al. GLOBOCAN 2008 v1.2, Cancer Incidence and Mortality Worldwide: IARC CancerBase No. 10 [Internet]. In. Lyon, France: International Agency for Research on Cancer; 2010. [Google Scholar]
- 2.Chen JG, Zhang SW. Liver cancer epidemic in China: past, present and future. Semin Cancer Biol. 2011;21:59–69. doi: 10.1016/j.semcancer.2010.11.002. [DOI] [PubMed] [Google Scholar]
- 3.Bouvard V, Baan R, Straif K, et al. A review of human carcinogens––Part B: biological agents. Lancet Oncol. 2009;10:321–322. doi: 10.1016/s1470-2045(09)70096-8. [DOI] [PubMed] [Google Scholar]
- 4.Baan R, Straif K, Grosse Y, et al. Carcinogenicity of alcoholic beverages. Lancet Oncol. 2007;8:292–293. doi: 10.1016/s1470-2045(07)70099-2. [DOI] [PubMed] [Google Scholar]
- 5.IARC Working Group on the Evaluation of Carcinogenic Risks to Humans. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans Volume 82 Some Traditional Herbal Medicines, Some Mycotoxins, Naphthalene and Styrene. In. Lyon, France: IARC Press and World Health Organization Marketing and Dissemination; 2002. [PMC free article] [PubMed] [Google Scholar]
- 6.IARC Working Group on the Evaluation of Carcinogenic Risks to Humans. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans Volume 83 Tobacco Smoke and Involuntary Smoking. Lyon, France: IARC Press and World Health Organization Marketing and Dissemination; 2004. [PMC free article] [PubMed] [Google Scholar]
- 7.La Vecchia C, Negri E, Decarli A, et al. Diabetes mellitus and the risk of primary liver cancer. Int J Cancer. 1997;73:204–207. doi: 10.1002/(sici)1097-0215(19971009)73:2<204::aid-ijc7>3.0.co;2-#. [DOI] [PubMed] [Google Scholar]
- 8.Lagiou P, Kuper H, Stuver SO, et al. Role of diabetes mellitus in the etiology of hepatocellular carcinoma. J Natl Cancer Inst. 2000;92:1096–1099. doi: 10.1093/jnci/92.13.1096. [DOI] [PubMed] [Google Scholar]
- 9.Wolk A, Gridley G, Svensson M, et al. A prospective study of obesity and cancer risk (Sweden) Cancer Causes Control. 2001;12:13–21. doi: 10.1023/a:1008995217664. [DOI] [PubMed] [Google Scholar]
- 10.El-Serag HB, Hampel H, Javadi F. The association between diabetes and hepatocellular carcinoma: a systematic review of epidemiologic evidence. Clin Gastroenterol Hepatol. 2006;4:369–380. doi: 10.1016/j.cgh.2005.12.007. [DOI] [PubMed] [Google Scholar]
- 11.Borena W, Strohmaier S, Lukanova A, et al. Metabolic risk factors and primary liver cancer in a prospective study of 578,700 adults. Int J Cancer. 2012;131:193–200. doi: 10.1002/ijc.26338. [DOI] [PubMed] [Google Scholar]
- 12.Finley CE, Barlow CE, Halton TL, et al. Glycemic index, glycemic load, and prevalence of the metabolic syndrome in the cooper center longitudinal study. J Am Diet Assoc. 2010;110:1820–1829. doi: 10.1016/j.jada.2010.09.016. [DOI] [PubMed] [Google Scholar]
- 13.Dong JY, Zhang L, Zhang YH, et al. Dietary glycaemic index and glycaemic load in relation to the risk of type 2 diabetes: a meta-analysis of prospective cohort studies. Br J Nutr. 2011;106:1649–1654. doi: 10.1017/S000711451100540X. [DOI] [PubMed] [Google Scholar]
- 14.McKeown NM, Meigs JB, Liu S, et al. Carbohydrate nutrition, insulin resistance, and the prevalence of the metabolic syndrome in the Framingham offspring cohort. Diabetes Care. 2004;27:538–546. doi: 10.2337/diacare.27.2.538. [DOI] [PubMed] [Google Scholar]
- 15.Silva FM, Steemburgo T, de Mello VD, et al. High dietary glycemic index and low fiber content are associated with metabolic syndrome in patients with type 2 diabetes. J Am Coll Nutr. 2011;30:141–148. doi: 10.1080/07315724.2011.10719953. [DOI] [PubMed] [Google Scholar]
- 16.Inoue M, Iwasaki M, Otani T, et al. Diabetes mellitus and the risk of cancer: results from a large-scale population-based cohort study in Japan. Arch Intern Med. 2006;166:1871–1877. doi: 10.1001/archinte.166.17.1871. [DOI] [PubMed] [Google Scholar]
- 17.Gnagnarella P, Gandini S, La Vecchia C, et al. Glycemic index, glycemic load, and cancer risk: a meta-analysis. Am J Clin Nutr. 2008;87:1793–1801. doi: 10.1093/ajcn/87.6.1793. [DOI] [PubMed] [Google Scholar]
- 18.Mulholland HG, Murray LJ, Cardwell CR, et al. Glycemic index, glycemic load, and risk of digestive tract neoplasms: a systematic review and meta-analysis. Am J Clin Nutr. 2009;89:568–576. doi: 10.3945/ajcn.2008.26823. [DOI] [PubMed] [Google Scholar]
- 19.Mulholland HG, Murray LJ, Cardwell CR, et al. Dietary glycaemic index, glycaemic load and breast cancer risk: a systematic review and meta-analysis. Br J Cancer. 2008;99:1170–1175. doi: 10.1038/sj.bjc.6604618. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Barclay AW, Petocz P, McMillan-Price J, et al. Glycemic index, glycemic load, and chronic disease risk––a meta-analysis of observational studies. Am J Clin Nutr. 2008;87:627–637. doi: 10.1093/ajcn/87.3.627. [DOI] [PubMed] [Google Scholar]
- 21.Dong JY, Qin LQ. Dietary glycemic index, glycemic load, and risk of breast cancer: meta-analysis of prospective cohort studies. Breast Cancer Res Treat. 2011;126:287–294. doi: 10.1007/s10549-011-1343-3. [DOI] [PubMed] [Google Scholar]
- 22.Slattery ML, Benson J, Berry TD, et al. Dietary sugar and colon cancer. Cancer Epidemiol Biomarkers Prev. 1997;6:677–685. [PubMed] [Google Scholar]
- 23.Larsson SC, Giovannucci E, Wolk A. Dietary carbohydrate, glycemic index, and glycemic load in relation to risk of colorectal cancer in women. Am J Epidemiol. 2007;165:256–261. doi: 10.1093/aje/kwk012. [DOI] [PubMed] [Google Scholar]
- 24.Michaud DS, Liu S, Giovannucci E, et al. Dietary sugar, glycemic load, and pancreatic cancer risk in a prospective study. J Natl Cancer Inst. 2002;94:1293–1300. doi: 10.1093/jnci/94.17.1293. [DOI] [PubMed] [Google Scholar]
- 25.Augustin LS, Dal Maso L, La Vecchia C, et al. Dietary glycemic index and glycemic load, and breast cancer risk: a case-control study. Ann Oncol. 2001;12:1533–1538. doi: 10.1023/a:1013176129380. [DOI] [PubMed] [Google Scholar]
- 26.Wen W, Shu XO, Li H, et al. Dietary carbohydrates, fiber, and breast cancer risk in Chinese women. Am J Clin Nutr. 2009;89:283–289. doi: 10.3945/ajcn.2008.26356. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Lagiou P, Rossi M, Tzonou A, et al. Glycemic load in relation to hepatocellular carcinoma among patients with chronic hepatitis infection. Ann Oncol. 2009;20:1741–1745. doi: 10.1093/annonc/mdp059. [DOI] [PubMed] [Google Scholar]
- 28.Rossi M, Lipworth L, Maso LD, et al. Dietary glycemic load and hepatocellular carcinoma with or without chronic hepatitis infection. Ann Oncol. 2009;20:1736–1740. doi: 10.1093/annonc/mdp058. [DOI] [PubMed] [Google Scholar]
- 29.George SM, Mayne ST, Leitzmann MF, et al. Dietary glycemic index, glycemic load, and risk of cancer: a prospective cohort study. Am J Epidemiol. 2009;169:462–472. doi: 10.1093/aje/kwn347. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Zheng W, Chow WH, Yang G, et al. The Shanghai women's health study: rationale, study design, and baseline characteristics. Am J Epidemiol. 2005;162:1123–1131. doi: 10.1093/aje/kwi322. [DOI] [PubMed] [Google Scholar]
- 31.Cai H, Zheng W, Xiang YB, et al. Dietary patterns and their correlates among middle-aged and elderly Chinese men: a report from the Shanghai men's health study. Br J Nutr. 2007;98:1006–1013. doi: 10.1017/S0007114507750900. [DOI] [PubMed] [Google Scholar]
- 32.Shu XO, Yang G, Jin F, et al. Validity and reproducibility of the food frequency questionnaire used in the Shanghai women's health study. Eur J Clin Nutr. 2004;58:17–23. doi: 10.1038/sj.ejcn.1601738. [DOI] [PubMed] [Google Scholar]
- 33.Villegas R, Yang G, Liu D, et al. Validity and reproducibility of the food-frequency questionnaire used in the Shanghai men's health study. Br J Nutr. 2007;97:993–1000. doi: 10.1017/S0007114507669189. [DOI] [PubMed] [Google Scholar]
- 34.Yang YX, Wang GY, Pan XC. China Food Composition Tables 2002. Beijing, China: Beijing University Medical Press; 2002. [Google Scholar]
- 35.Foster-Powell K, Holt SH, Brand-Miller JC. International table of glycemic index and glycemic load values: 2002. Am J Clin Nutr. 2002;76:5–56. doi: 10.1093/ajcn/76.1.5. [DOI] [PubMed] [Google Scholar]
- 36.Salmeron J, Manson JE, Stampfer MJ, et al. Dietary fiber, glycemic load, and risk of non-insulin-dependent diabetes mellitus in women. JAMA. 1997;277:472–477. doi: 10.1001/jama.1997.03540300040031. [DOI] [PubMed] [Google Scholar]
- 37.Willett WC, Howe GR, Kushi LH. Adjustment for total energy intake in epidemiologic studies. Am J Clin Nutr. 1997;65:1220S–1228S. doi: 10.1093/ajcn/65.4.1220S. discussion 1229S–1231S. [DOI] [PubMed] [Google Scholar]
- 38.Howarth NC, Murphy SP, Wilkens LR, et al. The association of glycemic load and carbohydrate intake with colorectal cancer risk in the multiethnic cohort study. Am J Clin Nutr. 2008;88:1074–1082. doi: 10.1093/ajcn/88.4.1074. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.