Abstract
Insulin resistance likely increases the risk of chronic liver disease (CLD) and liver cancer, but long-term prospective studies with measured fasting glucose and insulin are lacking. We evaluated the associations of pre-diagnostic fasting glucose, insulin and the homeostasis model assessment of insulin resistance (HOMA-IR) with liver cancer and CLD mortality in a prospective study of Finnish male smokers with extended follow-up time (≤22 years) and information on known risk factors using data from 138 incident primary liver cancer cases 216 CLD deaths and 681 matched controls. Fasting glucose and insulin were measured in baseline serum. We used unconditional logistic regression to estimate odds ratios (ORs) and 95% confidence intervals (CI) adjusted for age, alcohol, education, smoking, body mass index, and hepatitis B and C viral status. Among those without self-reported diabetes, glucose was positively associated with liver cancer (Quartile 3 vs. Quartile 1 (Q3/Q1): OR=1.88, 1.03–3.49; Q4/Q1: OR=2.40, 1.33–4.35, P-trend=.002), and undiagnosed, biochemically defined, diabetes was associated with higher risk of liver cancer (OR=2.95, 1.46–5.96) and CLD mortality (OR=1.88, 1.00–3.56). Serum insulin and HOMA-IR were also positively associated with liver cancer (Q4/Q1: OR=3.41, 1.74–6.66, P-trend<.0001; OR=3.72, 1.89–7.32, P-trend<.0001, respectively) and CLD (OR=2.51, 1.44–4.37, P-trend=.0002; OR=2.31, 1.34–3.97, P-trend=.001, respectively), with stronger associations observed for liver cancer diagnosed >10 years after baseline. In conclusion, elevated fasting glucose and insulin, and insulin resistance were independently associated with risk of liver cancer and CLD mortality, suggesting a potentially important etiologic role for insulin and glucose dysregulation even in the absence of diagnosed diabetes.
Keywords: diabetes, insulin resistance, fasting glucose, fasting insulin, liver cancer, chronic liver disease mortality
Chronic liver disease (CLD), including cirrhosis, is a major cause of death in the United States, especially among men (1). Liver cancer is the sixth most commonly occurring cancer and the second-leading cause of cancer-related mortality worldwide (2), and most cases of liver cancer are preceded by advanced liver disease (3). Although typically more common in developing countries, rates of liver cancer have increased rapidly in developed countries, including the United States and countries in Europe (4–7). Hepatitis B (HBV) and C (HCV) viruses, excessive alcohol intake, and aflatoxin exposure are strong risk factors for CLD (8) and liver cancer (4, 9). Much of the increase in liver cancer in Western countries has been ascribed to HCV (10, 11). However, 30% to 40% of liver cancers occur in patients without established risk factors (12).
In addition to HCV, obesity and diabetes may contribute to increasing liver cancer rates. Researchers have shown that diabetes is associated with a 2-fold increased risk of CLD (8), and a large body of evidence supports a positive association between diabetes and liver cancer (13–15). Several mechanisms are possible (16, 17) including that high insulin may have mitogenic effects (18). Patients with diabetes are also more likely to have hepatic steatosis (19, 20), either as simple nonalcoholic fatty liver disease or the more extreme form of nonalcoholic steatohepatitis (NASH). Fatty liver may increase liver cancer risk through excess inflammation, oxidative stress, and other mechanisms (20–23). Recent studies estimate that one third or more of US adults have fatty liver (24).
Although associations between diabetes and liver cancer have been widely reported (13, 14), some prior studies had limitations. Previous studies, many of which relied on self-reported diabetes, likely underestimated the prevalence of diabetes, and studies of undiagnosed diabetes or higher glucose in the absence of diabetes are lacking. In addition, studies have often not had complete ascertainment of possible confounding factors such as alcohol intake, HBV, HCV, and obesity (25). Most studies with information on HBV and HCV were conducted in populations with high prevalence, limiting statistical power for examination of the diabetes association in HBV and HCV negative participants. Many previous studies employed a cross-sectional design or had only limited follow-up between diabetes assessment and cancer incidence, precluding evaluation of temporality (13, 26). This may be of particular concern since the liver plays a critical role in glucose and insulin metabolism and cirrhosis can cause insulin resistance and diabetes (27). Finally, few studies examined associations between prediagnostic insulin concentrations and subsequent risk of liver cancer (28).
Because most cases of liver cancer develop in those with advanced CLD, it is possible to gain further insight into disease etiology by studying both endpoints. Thus, we examined the associations of pre-diagnostic fasting glucose, insulin and the homeostasis model assessment of insulin resistance (HOMA-IR)(29, 30) with primary liver cancer incidence or chronic liver disease (CLD) mortality during up to 22-years of follow-up in the Alpha-Tocopherol, Beta-Carotene Cancer Prevention (ATBC) Study, a large prospective cohort with low HBV and HCV prevalence (31).
Materials and Methods
Participants
The ATBC Study was a randomized, double-blind placebo-controlled, primary prevention trial designed to determine whether daily supplementation with alpha-tocopherol (50mg/day), beta-carotene (20mg/day), or both, would reduce the incidence of lung and other cancers in male smokers (31).
The ATBC cohort includes 29 133 Finnish male smokers, aged 50–69 years old, who were enrolled between 1985 and 1988. Individuals with a history of cirrhosis or chronic alcoholism were excluded from the study. Although supplementation ended in 1993, participants have been under follow-up since that time. ATBC was approved by the institutional review boards of both the National Institutes of Health in the United States and the National Public Health Institutes in Finland. All participants provided written informed consent.
Follow-up, outcome ascertainment, and control selection
Men diagnosed with primary incident liver cancer (ICD-9=155) were identified through the Finnish Cancer Registry, which provided close to 100% case ascertainment (32). Men who died from CLD (ICD-9=571) were identified through the Finnish Register of Causes of Death. For the present study, men who developed liver cancer and died of CLD were only included in the liver cancer analysis. Controls were alive and cancer-free at the time of case diagnosis or death and were matched to cases (2:1) on age at randomization (± 1 year), date of blood draw (± 30 days), and sample availability.
With follow-up through December 31, 2007, 144 incident liver cancer cases, 218 CLD deaths, and 723 matched controls were identified in ATBC. For the present study, 138 incident liver cancer cases, 216 participants who died from CLD and 681 matched controls had adequate baseline serum to measure insulin, and glucose and test for HBV and HCV, markers.
Data collection and laboratory analysis
Prior to randomization, at baseline, participants completed questionnaires detailing demographic information, lifestyle, and medical history including whether they had been diagnosed with diabetes. Participant’s height and weight were measured by trained study staff. Participants completed a food frequency questionnaire which queried intake of alcohol and 275 other items. All participants donated a fasting (overnight) blood sample at baseline which was stored at −70°C.
The SAIC NCI-Frederick National Laboratory tested for HBV surface antigen (HBsAg), an indication of current HBV infection, antibody to hepatitis B core antigen (anti-HBc), an indication of whether a person has ever been infected, and for antibody to HCV (anti-HCV) an indication of current infection with HCV. HBsAg was tested using an enzyme immunoassay (Bio-Rad Laboratories, Redmond, WA). Anti-HBc and anti-HCV were tested using enzyme-linked immunosorbent assays (Ortho-Clinical Diagnostics, Raritan, NJ). We included a panel of samples with known HBV and HCV positivity and concordance with known status was perfect.
Insulin and glucose were measured in baseline serum by the Immunochemical Core Laboratory at the Mayo Clinic (Rochester, MN). Insulin was measured using a two-site immunoenzymatic assay on the Dxl automated immunoassay system from Beckman Instruments (Chaska, MN). The inter-assay coefficient of variation (CV) for a pooled quality control sample included in each batch (8% of the overall samples) was 3.2% with a range across batches of 1.5% to 5.7%. Serum glucose was measured on the Roche Cobas c311 (Roche Diagnostics, Indianapolis, IN 46250) utilizing a hexokinase reagent from Boehringer Mannheim (Indianapolis, IN). The inter-assay CV was 0.6% with a range across batches of 0.1% to 2.1%. HOMA-IR (fasting insulin × fasting glucose/22.5, with fasting insulin expressed in µU/mL and fasting glucose expressed in mmol/L) was calculated as previously described (29).
A subset of serum samples from cohort participants (n=50) were tested for the presence of aflatoxin-albumin adducts at the University of Leeds (England). As expected in the Finnish population, we found no evidence for exposure in this subset (data not shown); therefore, we did not measure aflatoxin exposure in our larger case-control set.
Statistical analysis
Diabetes was defined by either self-report or having fasting glucose ≥126 mg/dL (33). For glucose, insulin, and HOMA-IR, we used quartiles with cut-points based on the distribution of the controls who did not report diabetes at baseline. We tested for differences in the distribution of potential risk factors between cases and controls using the Chi-square and the Wilcoxon rank tests for categorical and continuous variables, respectively. Among controls who did not report a diagnosis of diabetes at baseline, we also examined baseline characteristics by median glucose, insulin and HOMA-IR using the Mantel-Haenszel Chi-Square or Fisher’s Exact test for categorical variables and the Jonckheere-Terpstra test for continuous variables.
Odds ratios (OR) and 95% confidence intervals (95% CI) were calculated using unconditional logistic regression models. Results were similar using conditional logistic regression models (data not shown). We present ORs from age (years) adjusted models and from models that were additionally adjusted for alcohol intake (≤2.8, >2.8 to ≤11, >11 to ≤26, >26 to ≤44, >44 g/day), body mass index (BMI, <18.5, 18.5 to <25, 25 to <30, 30 to <35, ≥35 kg/m2), anti-HBc, HBsAg, anti-HCV, education (elementary education or less), cigarettes per day, and duration of smoking (years). Tests for trend were conducted by treating quartiles as an ordinal variable in the model; statistical significance was then determined by the Wald test. Follow-up time began at the date of randomization and continued until the date of cancer diagnosis, death, or December 31, 2007, whichever came first. We also conducted time-stratified analyses by follow-up for the first 10 years or more than 10 years.
We performed stratified analyses by median alcohol intake (≤11.3 versus >11.3 g/day), BMI (≤26 versus >26 kg/m2), cigarettes/day (≤ 20 versus >20), and years smoked (≤ 35 versus >35). We used dichotomous cut-points for HOMA-IR and insulin concentration, comparing participants in the fourth quartile (Q4) versus those in the first through third quartiles (Q1–3). Interactions were tested by comparing models with and without cross product terms using likelihood ratio tests. We conducted sensitivity analysis excluding HBV and HCV positive participants.
Finally, among those who did not report a diagnosis of diabetes at baseline, we examined the joint effects of glucose-defined diabetes with insulin concentration and daily alcohol intake as these factors may modify associations of diabetes with liver cancer and liver disease mortality. For insulin, we used a referent group of participants who did not have diabetes and had an insulin level <6.7 µU/mL, the 75% percentile in controls. For alcohol, the referent group included participants without diabetes and with below median alcohol intake.
All analyses were conducted with SAS version 9.3. All statistical tests were two-sided.
Results
Baseline characteristics are shown in Table 1 for participants who developed liver cancer or who died from CLD and their matched controls. The age ranges of cases and controls were similar for both endpoints. Both case groups drank more alcohol than controls, although this difference reached statistical significance only for CLD mortality where above, as compared with below, median alcohol intake was associated with more than a 4-fold increased odds of CLD mortality. Relative to controls, those who developed liver cancer had a longer smoking duration, whereas those who died from CLD tended to smoke more cigarettes/day but had similar smoking duration. A very low proportion of study participants tested positive for HBV or HCV; nevertheless, the prevalence of anti-HBc but not HBsAg was higher in liver cancer cases, and the prevalence of anti-HCV was higher in both liver cancer and liver disease mortality cases as compared with controls.
Table 1.
Baseline characteristics of cases and controls in ATBC
| Liver cancer | Chronic liver disease deaths | |||||
|---|---|---|---|---|---|---|
| Baseline characteristic | Cases a (n=138) |
Controls a (n=253) |
Odds ratio (95% CI) d |
Cases a (n=216) |
Controls a (n=428) |
Odds ratio (95% CI) d |
| Entry age (years) | 58 (55–62) | 57 (54–61) | 1.03 (0.99–1.07) | 55 (52–58) | 55 (51–58) | 1.00 (0.96–1.04) |
| Cigarettes/day | 20 (15–25) | 20 (13–20) | 1.03 (1.01–1.06) | 23 (15–25) | 20 (15–25) | 1.03 (1.01–1.05) |
| Smoking duration | 40 (34–43) | 37 (30–42) | 1.02 (1.00–1.05) | 35 (30–40) | 35 (30–40) | 1.00 (0.98–1.03) |
| Alcohol, >11.3 g/day b | 71 (54.6%) | 115 (47.7%) | 1.36 (0.89–2.10) | 157 (81.8%) | 205 (51.6%) | 4.25 (2.80–6.46) |
| HBV, antibody to core antigen, yes |
22 (15.9%) | 17 (6.7%) | 2.52 (1.38–4.96) | 14 (6.5%) | 27 (6.3%) | 1.04 (0.53–2.05) |
| HBV, surface antigen, yes | 2 (1.5%) | 3 (1.2%) | 1.26 (0.21–7.67) | 1 (0.5%) | 1 (0.2%) | 2.00 (0.12–32.07) |
| HCV, antibody, yes | 6 (4.4%) | 2 (0.8%) | 5.44 (1.08–27.39) | 6 (2.8%) | 2 (0.5%) | 6.22 (1.24–31.20) |
| BMI | ||||||
| <18.5 kg/m2 | 2 (1.5%) | 1 (0.4%) | 4.62 (0.40–53.08) | 2 (0.9%) | 1 (0.2%) | 3.99 (0.36–44.6) |
| 18.5 to <25 kg/m2 | 38 (27.5%) | 93 (36.8%) | 1.00 | 77 (35.7%) | 154 (36.0%) | 1.00 |
| 25 to <30 kg/m2 | 63 (45.7%) | 120 (47.4%) | 1.30 (0.80–2.12) | 100 (46.3%) | 203 (47.4%) | 0.98 (0.68–1.42) |
| 30 to <35 kg/m2 | 28 (20.3%) | 37 (14.6%) | 1.85 (1.00–3.45) | 31 (14.4%) | 62 (14.5%) | 1.00 (0.60–1.67) |
| ≥35 kg/m2 | 7 (5.1%) | 2 (0.8%) | 8.39 (1.67–42.29) | 6 (2.7%) | 8 (1.9%) | 1.50 (0.50–4.47) |
| Elementary education or less | 103 (74.6%) | 192 (75.9%) | 0.94 (0.58–1.52) | 154 (71.3%) | 339 (79.2%) | 0.65 (0.45–0.95) |
| Self-reported diabetes, yes | 14 (10.1%) | 8 (3.2%) | 3.35 (1.37–8.23) | 9 (4.2%) | 18 (4.2%) | 0.99 (0.44–2.25) |
| Glucose ≥126 mg/dL c | 15 (12.1%) | 10 (4.1%) | 3.44 (1.49–7.96) | 22 (10.6%) | 22 (5.4%) | 2.10 (1.13–3.88) |
| Either self-reported or glucose ≥126 mg/dL |
29 (21.0%) | 18 (7.1%) | 3.53 (1.88–6.65) | 31 (14.4%) | 40 (9.4%) | 1.63 (0.99–2.68) |
| Glucose (mg/dL) c | 103 (95–113) | 98 (92–106) | 1.03 (1.02–1.04) | 101 (93–111) | 99 (93–106) | 1.00 (1.00–1.01) |
| Insulin (µU/mL) c | 7.0 (3.8–11.2) | 4.3 (2.8–6.5) | 1.16 (1.10–1.22) | 4.3 (2.9–6.6) | 5.4 (3.2–9.5) | 1.06 (1.03–1.09) |
| HOMA-IR c | 1.9 (0.9–3.3) | 1.1 (0.7–1.6) | 1.65 (1.38–1.98) | 1.1 (0.7–1.7) | 1.3 (0.8–2.5) | 1.18 (1.07–1.30) |
Median (interquartile range) for continuous variables; n (column %) for categorical variables
Median level of alcohol consumption was 11.3 g/day
Restricted to participants who did not report a diagnosis of diabetes at baseline.
Models, other than the models for entry age, are age-adjusted.
Participants who developed liver cancer were more likely than controls to be obese at baseline; yet, no difference was observed for CLD. The prevalence of diabetes, defined as either having a self-report or a glucose ≥126 mg/dL, was higher in liver cancer (21.0%) and CLD cases (14.4%) than in their matched controls (7.1% and 9.4%, respectively). Among those who did not report a diabetes diagnosis at baseline, a higher percentage of liver cancer (12.1%) and CLD cases (10.6%) relative to controls (4.1% and 5.4%, respectively) had glucose ≥126 mg/dL. Finally, fasting insulin concentrations were higher in both case groups relative to controls, as were HOMA-IR scores (Table 1).
Distributions of baseline characteristics by median glucose, insulin, and HOMA-IR among controls who did not report a previous diagnosis of diabetes are shown in Table 2. The prevalence of overweight and obesity were associated with higher glucose and insulin concentrations and higher HOMA-IR scores. As expected, insulin concentration tended to be higher among those with higher glucose concentration.
Table 2.
Baseline characteristics of controls (n=655) by median glucose, insulin, and HOMA-IR among participants who did not report a previous diagnosis of diabetes.
| Glucose a | Insulin a | HOMA-IR a | |||||||
|---|---|---|---|---|---|---|---|---|---|
| Baseline characteristic |
≤99 mg/dL (n=346) |
>99 mg/dL (n=309) |
P b | ≤4.3 µU/mL (n=337) |
>4.3 µU/mL (n=318) |
P b | ≤1.05 (n=327) |
>1.05 (n=328) |
P b |
| Entry age (years) | 56 (52–59) | 56 (52–59) | 0.36 | 56 (52–59) | 56 (52–60) | 0.16 | 56 (51–58) | 56 (52–60) | 0.04 |
| Cigarettes/day | 20 (15–25) | 20 (15–25) | 0.43 | 20 (15–24) | 20 (15–25) | 0.36 | 20 (15–25) | 20 (15–25) | 0.71 |
| Smoking duration | 35 (30–40) | 35 (30–40) | 0.43 | 35 (30–40) | 35 (30–40) | 0.56 | 35 (30–40) | 35 (30–40) | 0.23 |
| Alcohol, >11.3 g/day |
156 (47.7%) | 156 (54.2%) | 0.11 | 166 (52.4%) | 146 (49.0%) | 0.40 | 165 (53.6%) | 147 (47.9%) | 0.16 |
| HBV, antibody to core antigen, yes |
24 (6.9%) | 19 (6.2%) | 0.68 | 18 (5.3%) | 25 (7.9%) | 0.19 | 18 (5.5%) | 25 (7.6%) | 0.27 |
| HBV, surface antigen, yes |
4 (1.2%) | 0 (0%) | 0.13 | 2 (0.6%) | 2 (0.6%) | 1.00 | 2 (0.3%) | 2 (0.3%) | 1.00 |
| HCV, antibody, yes | 3 (0.9%) | 1 (0.3%) | 0.63 | 1 (0.3%) | 3 (0.9%) | 0.36 | 1 (0.3%) | 3 (0.9%) | 0.62 |
| BMI | <0.0001 | <0.0001 | <0.0001 | ||||||
| <18.5 kg/m2 | 0 (0%) | 2 (0.7%) | 2 (0.6%) | 0 (0%) | 2 (0.6%) | 0 (0%) | |||
| 18.5 to <25 kg/m2 | 153 (44.2%) | 84 (27.2%) | 182 (54.0%) | 55 (17.3%) | 177 (54.1%) | 60 (18.3%) | |||
| 25 to <30 kg/m2 | 156 (45.1) | 157 (50.8%) | 138 (41.0%) | 175 (55.0%) | 135 (41.3%) | 178 (54.3%) | |||
| 30 to <35 kg/m2 | 34 (9.8%) | 60 (19.4%) | 15 (4.4%) | 79 (24.8%) | 13 (4.0%) | 81 (24.7%) | |||
| ≥35 kg/m2 | 3 (0.9%) | 6 (1.9%) | 0 (0%) | 9 (2.8%) | 0 (0%) | 9 (2.7%) | |||
| ≤ Elementary education |
264 (76.3%) | 244 (79.0%) | 0.42 | 264 (78.3%) | 244 (76.7%) | 0.62 | 258 (78.9%) | 250 (76.2%) | 0.41 |
| Glucose (mg/dL) | 93 (89–97) | 106 (103–114) | nd | 96 (90–102) | 102 (97–111) | <0.0001 | 95 (90–100) | 103 (98–112) | nd |
| Insulin (µU/mL) | 3.4 (2.4–5.2) | 5.5 (3.7–8.3) | <0.0001 | 2.9 (2.3–3.5) | 6.7 (5.3–8.8) | nd | 2.9 (2.3–3.5) | 6.5 (5.2–11.9) | nd |
| HOMA-IR | 0.8 (0.5–1.2) | 1.5 (1.0–2.3) | nd | 0.7 (0.5–0.9) | 1.7 (1.3–2.4) | nd | 0.7 (0.5–0.9) | 1.7 (1.3–2.3) | nd |
Restricted to participants who did not report a diagnosis of diabetes at baseline.
P-value for Mantel-Haenszel Chi-Square test or Fisher’s Exact test for categorical variables and Jonckheere-Terpstra test for continuous variables
nd: not determined
Tables 3 and 4 show the associations for prediagnostic diabetes, fasting concentration of glucose or insulin, and HOMA-IR with liver cancer or CLD. Diabetes, defined either by self-report or fasting glucose, was associated with both liver cancer (Table 3: OR= 2.79, 95% CI=1.65–4.75) and CLD mortality (Table 4: OR=1.83, 95% CI=1.09–3.10) in multivariable models. ORs for self-reported diabetes and glucose-defined diabetes were of similar magnitude and direction for liver cancer, but for CLD mortality, the OR for self-reported diabetes, although positive, was not statistically significant.
Table 3.
Odds ratios and 95% confidence intervals for diabetes, glucose, insulin, and insulin resistance with primary incident liver cancer over 22 years of follow-up in a nested case-control study from the ATBC cohort
| Diabetes | Self-reported diabetes only | Fasting glucose ≥126 mg/dL only b | Self-report or fasting glucose ≥126 mg/dL |
||||||
|---|---|---|---|---|---|---|---|---|---|
| No (ref) | Yes | No (ref) | Yes | No (ref) | Yes | ||||
| Controls/ cases |
Controls/ cases |
Odds ratio (95%CI) |
Controls/ cases |
Controls/ cases |
Odds ratio (95%CI) |
Controls/ cases |
Controls/ cases |
Odds ratio (95%CI) |
|
| Age adjusted | 655/124 | 26/14 | 2.59 (1.29–5.18) | 623/109 | 32/15 | 3.11 (1.61–6.03) | 623/109 | 58/29 | 2.97 (1.80–4.90) |
| Multivariable model a | 655/124 | 26/14 | 2.48 (1.20–5.12) | 623/109 | 32/15 | 2.95 (1.46–5.96) | 623/109 | 58/29 | 2.79 (1.65–4.75) |
| >0 to 10 years of follow-up a | 655/57 | 26/8 | 3.45 (1.38–8.63) | 623/51 | 32/6 | 2.69 (0.93–7.80) | 623/51 | 58/14 | 3.14 (1.51–6.54) |
| >10 years of follow-up a | 655/67 | 26/6 | 1.95 (0.73–5.21) | 623/58 | 32/9 | 3.07 (1.34–7.05) | 623/58 | 58/15 | 2.61 (1.34–5.09) |
| Glucose (mg/dL) b |
Q1 (ref) (89, 86–92) c |
Q2 (97, 95–98) c |
Q3 (103, 101–104) c |
Q4 (114, 110–124) c |
|||||
| Controls/ cases |
Controls/ cases |
Odds ratio (95%CI) |
Controls/ cases |
Odds ratio (95%CI) |
Controls/ cases |
Odds ratio (95%CI) |
P-trend | ||
| Age adjusted | 193/25 | 153/20 | 1.05 (0.56–1.98) | 155/34 | 1.69 (0.96–2.98) | 154/45 | 2.30 (1.34–3.96) | 0.001 | |
| Multivariable model a | 193/25 | 153/20 | 1.28 (0.66–2.49) | 155/34 | 1.88 (1.03–3.49) | 154/45 | 2.40 (1.33–4.35) | 0.002 | |
| >0 to 10 years of follow-up a | 193/18 | 153/11 | 1.26 (0.53–3.01) | 155/12 | 1.07 (0.45–2.50) | 154/16 | 1.40 (0.61–3.21) | 0.511 | |
| >10 years of follow-up a | 193/7 | 153/9 | 1.74 (0.62–4.90) | 155/22 | 3.83 (1.55–9.45) | 154/29 | 4.82 (1.99–11.70) | <0.0001 | |
| Insulin (µU/mL) b |
Q1 (ref) 2.4 (1.9–2.6)c |
Q2 3.6 (3.3–4.0)c |
Q3 5.3 (4.8–5.8)c |
Q4 8.7 (7.5–11.2)c |
|||||
| Controls/ cases |
Controls/ cases |
Odds ratio (95%CI) |
Controls/ cases |
Odds ratio (95%CI) |
Controls/ cases |
Odds ratio (95%CI) |
P-trend | ||
| Age adjusted | 175/20 | 162/17 | 0.91 (0.46–1.82) | 155/21 | 1.15 (0.60–2.22) | 163/66 | 3.42 (1.97–5.94) | <0.0001 | |
| Multivariable model a | 175/20 | 162/17 | 0.96 (0.47–1.96) | 155/21 | 1.14 (0.55–2.33) | 163/66 | 3.41 (1.74–6.66) | <0.0001 | |
| >0 to 10 years of follow-up a | 175/14 | 162/11 | 0.87 (0.36–2.12) | 155/9 | 0.52 (0.19–1.45) | 163/23 | 1.22 (0.47–3.19) | 0.816 | |
| >10 years of follow-up a | 175/6 | 162/6 | 1.05 (0.32–3.39) | 155/12 | 2.20 (0.77–6.29) | 163/43 | 7.42 (2.79–19.76) | <0.0001 | |
| HOMA-IR b |
Q1 (ref) 0.51 (0.42–60) c |
Q2 0.85 (0.76–0.94) c |
Q3 1.32 (1.20–1.51) c |
Q4 2.39 (2.07–3.35) c |
|||||
| Controls/ cases |
Controls/ cases |
Odds ratio (95%CI) |
Controls/ cases |
Odds ratio (95%CI) |
Controls/ cases |
Odds ratio (95%CI) |
P-trend | ||
| Age adjusted | 166/18 | 163/18 | 1.00 (0.50–2.01) | 163/20 | 1.05 (0.53–2.08) | 163/68 | 3.70 (2.09–6.55) | <0.0001 | |
| Multivariable model a | 166/18 | 163/18 | 1.04 (0.50–2.13) | 163/20 | 0.97 (0.46–2.04) | 163/68 | 3.72 (1.89–7.32) | <0.0001 | |
| >0 to 10 years of follow-up a | 166/12 | 163/11 | 0.94 (0.38–2.35) | 163/10 | 0.58 (0.21–1.63) | 163/24 | 1.60 (0.62–4.14) | 0.390 | |
| >10 years of follow-up a | 166/6 | 166/7 | 1.13 (0.36–3.54) | 163/10 | 1.50 (0.51–4.46) | 163/44 | 6.87 (2.60–18.17) | <0.0001 | |
Multivariable models are adjusted for age (years), cigarettes per day, and duration of smoking (years), alcohol intake (≤2.8, >2.8 to ≤11, >11 to ≤26, >26 to ≤44, >44 g/day), anti-HBc, HBsAg, anti-HCV, body mass index (BMI, <18.5, 18.5 to <25, 25 to <30, 30 to <35, ≥35 kg/m2), and education (elementary education or less, more than elementary education).
Restricted to participants who did not report a diagnosis of diabetes at baseline.
Median and interquartile range (IQR) among controls.
Table 4.
Odds ratios and 95% confidence intervals for diabetes, glucose, insulin, and insulin resistance with chronic liver disease mortality over 22 years of follow-up in a nested case-control study from the ATBC cohort
| Diabetes | Self-reported diabetes only | Fasting glucose ≥126 mg/dL only b | Self-report or fasting glucose ≥126 m/dL |
||||||
|---|---|---|---|---|---|---|---|---|---|
| No (ref) | Yes | No (ref) | Yes | No (ref) | Yes | ||||
| Controls/ cases |
Controls/ cases |
Odds ratio (95%CI) |
Controls/ cases |
Controls/ cases |
Odds ratio (95%CI) |
Controls/cases | Controls/ cases |
Odds ratio (95%CI) |
|
| Age adjusted | 655/207 | 26/9 | 1.12 (0.51–2.43) | 623/185 | 32/22 | 2.31 (1.31–4.08) | 623/185 | 58/31 | 1.81 (1.13–2.89) |
| Multivariable model a | 655/207 | 26/9 | 1.67 (0.71–3.94) | 623/185 | 32/22 | 1.88 (1.00–3.56) | 623/185 | 58/31 | 1.83 (1.09–3.10) |
| >0 to 10 years of follow-up a | 655/109 | 26/7 | 2.42 (0.93–6.30) | 623/92 | 32/17 | 3.33 (1.62–6.85) | 623/92 | 58/24 | 3.08 (1.70–5.58) |
| >10 years of follow-up a | 655/98 | 26/2 | 0.70 (0.15–3.40) | 623/58 | 32/9 | 0.71 (0.26–1.99) | 623/93 | 58/7 | 0.72 (0.30–1.72) |
| Glucose (mg/dL) b |
Q1 (ref) (89, 86–92) c |
Q2 (97, 95–98) c |
Q3 (103, 101–104) c |
Q4 (114, 110–124) c |
|||||
| Controls/ cases |
Controls/ cases |
Odds ratio (95%CI) |
Controls/ cases |
Odds ratio (95%CI) |
Controls/ cases |
Odds ratio (95%CI) |
P-trend | ||
| Age adjusted | 193/58 | 153/32 | 0.68 (0.42–1.10) | 155/47 | 1.00 (0.64–1.55) | 154/70 | 1.54 (1.02–2.31) | 0.019 | |
| Multivariable model a | 193/58 | 153/32 | 0.76 (0.45–1.29) | 155/47 | 1.00 (0.62–1.64) | 154/70 | 1.49 (0.93–2.37) | 0.064 | |
| >0 to 10 years of follow-up a | 193/27 | 153/14 | 0.74 (0.35–1.56) | 155/25 | 1.24 (0.65–2.36) | 154/43 | 2.08 (1.14–3.80) | 0.007 | |
| >10 years of follow-up a | 193/31 | 153/18 | 0.74 (0.38–1.44) | 155/22 | 0.83 (0.44–1.58) | 154/27 | 1.02 (0.54–1.90) | 0.929 | |
| Insulin (µU/mL) b |
Q1 (ref) 2.4 (1.9–2.6) c |
Q2 3.6 (3.3–4.0) c |
Q3 5.3 (4.8–5.8) c |
Q4 8.7 (7.5–11.2) c |
|||||
| Controls/ cases |
Controls/ cases |
Odds ratio (95%CI) |
Controls/ cases |
Odds ratio (95%CI) |
Controls/ cases |
Odds ratio (95%CI) |
P-trend | ||
| Age adjusted | 175/45 | 162/33 | 0.79 (0.48–1.29) | 155/45 | 1.15 (0.72–1.83) | 163/84 | 2.09 (1.37–3.20) | <.0001 | |
| Multivariable model a | 175/45 | 162/17 | 0.88 (0.50–1.52) | 155/45 | 1.37 (0.80–2.37) | 163/84 | 2.51 (1.44–4.37) | 0.0002 | |
| >0 to 10 years of follow-up a | 175/20 | 162/19 | 1.04 (0.51–2.12) | 155/24 | 1.53 (0.75–3.10) | 163/46 | 2.85 (1.41–5.75) | 0.002 | |
| >10 years of follow-up a | 175/25 | 162/14 | 0.75 (0.36–1.60) | 155/21 | 1.24 (0.60–2.56) | 163/38 | 2.17 (1.06–4.42) | 0.016 | |
| HOMA-IR b |
Q1 (ref) 0.51 (0.42–0.60) c |
Q2 0.85 (0.76–0.94) c |
Q3 1.32 (1.20–1.51) c |
Q4 2.39 (2.07–3.35) c |
|||||
| Controls/ cases |
Controls/ cases |
Odds ratio (95%CI) |
Controls/ cases |
Odds ratio (95%CI) |
Controls/ cases |
Odds ratio (95%CI) |
P-trend | ||
| Age adjusted | 166/46 | 163/33 | 0.72 (0.44–1.18) | 163/42 | 0.97 (0.60–1.55) | 163/86 | 1.98 (1.30–3.02) | 0.0002 | |
| Multivariable model a | 166/46 | 163/33 | 0.87 (0.50–1.53) | 163/42 | 1.24 (0.71–2.17) | 163/86 | 2.31 (1.34–3.97) | 0.001 | |
| >0 to 10 years of follow-up a | 166/23 | 163/16 | 0.70 (0.33–1.48) | 163/23 | 1.23 (0.61–2.49) | 163/47 | 2.24 (1.13–4.43) | 0.005 | |
| >10 years of follow-up a | 166/23 | 163/17 | 1.04 (0.50–2.17) | 163/19 | 1.23 (0.58–2.62) | 163/39 | 2.30 (1.12–4.71) | 0.016 | |
Multivariable models are adjusted for age (years), cigarettes per day, and duration of smoking (years), alcohol intake (≤2.8, >2.8 to ≤11, >11 to ≤26, >26 to ≤44, >44 g/day), anti-HBc, HBsAg, anti-HCV, body mass index (BMI, <18.5, 18.5 to <25, 25 to <30, 30 to <35, ≥35 kg/m2), and education (elementary education or less, more than elementary education).
Restricted to participants who did not report a diagnosis of diabetes at baseline.
Median and interquartile range (IQR) among controls
Among participants who did not report a previous diagnosis of diabetes, both liver cancer (Table 3) and CLD mortality (Table 4) were positively associated with insulin concentration and HOMA-IR. Relative to Q1, risk estimates for insulin and HOMA-IR were elevated in Q4 for liver cancer (OR=3.41, 95% CI=1.74–6.66, P-trend<0.0001; and OR=3.72, 95% CI=1.89–7.3, P-trend<0.0001, respectively) and CLD mortality (OR=2.51, 95% CI=1.44–4.37, P-trend=0.0002; and OR=2.31, 95% CI=1.34–3.97, P-trend=0.001, respectively). For glucose, participants in Q3 (median glucose = 103 mg/dL; IQR=101–104) as well as those in Q4 (median glucose = 114 mg/dL; IQR=110–124) were at higher risk of liver cancer relative to Q1 (OR=1.88, 95% CI=1.03–3.49 and OR=2.40, 95% CI=1.33–4.35, respectively). However, no association was observed for glucose and CLD mortality (P-trend=0.064).
In analyses stratified by follow-up time, the association between diabetes and liver cancer appeared similar in each follow-up period (Table 3). The associations for glucose, insulin concentration, and HOMA-IR appeared stronger in liver cancer cases that occurred more than 10 years after baseline than in cases that occurred in the first 10 years of follow-up. The pattern was different for CLD where associations with glucose appeared stronger for deaths in the first 10 years (than for deaths more than 10 years after baseline. In contrast, the ORs for CLD with insulin and HOMA-IR appeared similar in each follow-up period (Table 4).
We observed similar associations for diabetes, insulin, and HOMA-IR with each endpoint after excluding HBV and HCV positive participants. We also observed similar associations for diabetes, insulin, and HOMA-IR with each endpoint across strata defined by baseline alcohol use, BMI, and smoking history (Supplementary Table). Of all the examined stratifications, four deviations from homogeneity were observed; owing to relatively small sample sizes and multiple comparisons, these results should, however, be interpreted with caution.
Finally, we examined the joint effects of biochemically defined diabetes with insulin concentration and daily alcohol intake (Table 5). In these analyses, we observed little evidence for an association with diabetes among participants with lower insulin concentration, although there were few cases in this group. Among participants without diabetes, we observed some evidence for an association of insulin concentration with liver cancer and CLD. The highest ORs were among participants who had both higher insulin concentration and diabetes. For alcohol, we observed similar ORs for diabetes and liver cancer among those with high and low alcohol intake. In contrast, participants with higher alcohol intake and diabetes had more than two-fold higher odds of CLD mortality than those with lower alcohol intake and diabetes. These observed differences should, however, be interpreted with caution as multiple comparisons were made and P-values were greater than 0.05 for statistical tests of multiplicative and additive interactions for diabetes with insulin or alcohol intake.
Table 5.
Odds ratios and 95% confidence intervals for the joint effects analysis of diabetes with fasting insulin levels c and alcohol consumption d with liver cancer incidence and chronic liver disease mortality among those without a self-reported history of diabetes at baseline
| No diabetes (Glucose < 126 mg/dL) | Diabetes (Glucose ≥126 mg/dL) | P-interaction | ||||
|---|---|---|---|---|---|---|
| Controls/cases | Odds ratio (95%CI) |
Controls/cases | Odds ratio (95%CI) |
Multiplicative | Additive | |
| Liver cancer a | 0.606 | 0.421 | ||||
| Insulin ≤6.7 µU/mL | 478/56 | 1.00 (ref) | 14/2 | 1.54 (0.32–7.48) | ||
| Insulin >6.7 µU/mL | 145/53 | 2.91 (1.77–4.80) | 18/13 | 7.17 (3.10–16.55) | ||
| Chronic liver disease a | 0.252 | 0.322 | ||||
| Insulin ≤6.7 µU/mL | 478/120 | 1.00 (ref) | 14/3 | 0.73 (0.19–2.90) | ||
| Insulin >6.7 µU/mL | 145/65 | 2.05 (1.32–3.18) | 18/19 | 3.79 (1.75–8.22) | ||
| Liver cancer b | 0.694 | 0.944 | ||||
| Alcohol ≤11.3 g/day | 291/47 | 1.00 (ref) | 12/6 | 3.56 (1.21–10.50) | ||
| Alcohol >11.3 g/day | 296/55 | 1.24 (0.79–1.97) | 16/8 | 3.32 (1.25–8.87) | ||
| Chronic liver disease b | 0.295 | 0.930 | ||||
| Alcohol ≤11.3 g/day | 291/26 | 1.00 (ref) | 12/5 | 4.52 (1.38–14.78) | ||
| Alcohol >11.3 g/day | 296/136 | 4.96 (3.08–7.99) | 16/16 | 10.68 (4.63–24.65) | ||
Models are adjusted for age (years), cigarettes per day, and duration of smoking (years), alcohol intake (≤2.8, >2.8 to ≤11, >11 to ≤26, >26 to ≤44, >44 g/day), anti-HBc, HBsAg, anti-HCV, body mass index (BMI, <18.5, 18.5 to <25, 25 to <30, 30 to <35, ≥35 kg/m2), education (elementary education or less, more than elementary education).
Models are adjusted for age (years), cigarettes per day, and duration of smoking (years), anti-HBc, HBsAg, anti-HCV, body mass index (BMI, <18.5, 18.5 to <25, 25 to <30, 30 to <35, ≥35 kg/m2), education (elementary education or less, more than elementary education).
Above versus below the 75% of insulin in controls, 6.7 µU/mL.
Above versus below the median level of alcohol consumption 11.3 g/day.
Discussion
In our study, among Finnish male smokers without a prior diabetes diagnosis, higher glucose concentration was associated with increased risk of incident liver cancer, and higher insulin concentration or higher HOMA-IR was associated with increased risk of incident liver cancer and CLD mortality during 22 years of follow-up. These associations were independent of other CLD and liver cancer risk factors, including HBV and HCV status, alcohol intake, BMI, and smoking history.
Many previous studies have observed associations between diabetes and liver cancer, using a number of different study designs, including case-control, record linkage, and prospective cohorts (13). Similar to our estimate for self-reported diabetes (OR=2.48, 95% CI=1.20–5.12), a recent meta-analysis reported summary relative risk estimates for hepatocellular carcinoma (HCC) and primary liver cancer of 2.06 (95% CI=1.64–2.60) and 1.75 (95% CI=1.25–2.47), respectively (13). In the current study, the OR for diabetes either by self-report or serum glucose testing was 2.79 (95% CI=1.65–4.75) indicating that people with undiagnosed diabetes are similarly at increased risk for liver cancer. Moreover, a trend (P-trend =0.002) was observed for liver cancer across increasing quartiles of serum glucose, indicating that higher glucose concentrations, including those below 126 mg/dL (i.e. Q3: median glucose = 103 mg/dL; IQR=101–104), are associated with higher odds of liver cancer. Another recent study observed a positive, albeit not statistically significant association, between categories of serum glucose and liver cancer noting a limited sample size and imprecise risk estimates in the higher categories of serum glucose (34).
Previous findings for insulin and liver cancer, although more limited than for diabetes, are also consistent with our results. C-peptide, a marker of hyperinsulinemia, has been positively associated with liver cancer in a large European prospective cohort (35). High insulin concentrations have been associated with poorer prognosis in patients with liver cancer (36, 37), as well as liver disease progression (38) and poorer prognosis after liver transplant (39) among HCV(+) patients. Higher insulin concentrations have also been associated with liver cancer in a cohort of HBV carriers (28). Our observation of an association with higher fasting insulin among participants without diabetes is intriguing and suggests that insulin may promote carcinogenesis in the absence of diabetes. Previous results of an association between higher insulin concentrations and more rapid liver tumor growth (36) are also consistent with our findings, as are pharmacoepidemiologic studies (40). For example, a recent meta-analysis found that among patients with diabetes, prescribed insulin was associated with increased liver cancer risk, whereas metformin and thiazolidinedione were associated with decreased liver cancer risk (40). Such findings could, however, reflect confounding by indication.
Although numerous studies have explored the interrelationship between diabetes and nonalcoholic fatty liver disease (NAFLD) (8, 24), associations of insulin and glucose with subsequent mortality from chronic liver disease, particularly among those without a prior diabetes diagnosis, are poorly understood. In the current study, we found stronger findings for diabetes, insulin, and HOMA-IR with liver cancer than with CLD mortality. It is possible that higher insulin and glucose levels may be more strongly related to subsequent liver cancer than fatal non-cancer liver disease endpoints. In support, some studies have suggested that high glucose and insulin levels may promote liver tumor growth (36). However, the observed differences could also be due to chance.
Alternatively, associations with glucose and insulin concentrations could reflect reverse-causality. Cirrhosis can cause diabetes (27) and previous findings suggest that blood insulin concentration and insulin resistance are affected by diminished insulin clearance from fatty liver (41). Although our study excluded participants manifesting cirrhotic symptoms at baseline, we lacked information on asymptomatic underlying liver disease at baseline. However, we consistently observed weaker associations for glucose, insulin, and HOMA-IR with liver cancers in the first 10 years of follow-up, when participants developing an endpoint would be more likely to have cirrhosis at baseline, than with liver cancers in years 11–22 of follow-up. If insulin resistance and diabetes at baseline were solely a reflection of advanced liver disease, then a higher risk of liver cancer would have been expected in the first decade of observation. These data, coupled with previous findings of associations with liver disease progression in the context of HCV (38, 39, 42), suggest that our results for liver cancer do not simply reflect the metabolic alterations of undiagnosed cirrhosis. In contrast, associations for self-reported diabetes and glucose were only apparent for CLD mortality that occurred within 10 years of baseline, suggesting more of a concern for reverse causality.
Key strengths of our study include its prospective design, 22-year follow-up, measured fasting glucose and insulin concentrations, ability to adjust for major liver cancer risk factors including HBV, HCV, alcohol use, and smoking, and the exclusion of patients with cirrhosis and chronic alcoholism at baseline. Limitations include our relatively modest sample size, a lack of histology information for liver cancer cases, though the majority were likely HCC, and a single measurement of fasting glucose and insulin, which could lead to misclassification of the exposure. Although repeat measures are ideal, national prevalence estimates generally rely on a single serum measurement to define undiagnosed diabetes (43). We were also unable to differentiate between type-1 and type-2 diabetes, although most diabetes would be type-2 in this older population. We lacked data on undiagnosed CLD at baseline, as discussed above. We also lacked data on incident CLD during follow-up, and the associations of diabetes and insulin concentrations with CLD risk may differ from those observed for CLD mortality. The ATBC study included only male smokers, which may affect the generalizability of our findings to women and to never-smokers. Although our measured levels of insulin in controls were consistent with those previously measured in the cohort (44–46), they were lower than those in US population surveys (47). Future studies in populations with higher insulin levels are needed to extend and replicate these results.
In summary, participants in the ATBC cohort with higher glucose and insulin levels as well as those with diagnosed and undiagnosed diabetes were more likely to develop liver cancer and die from CLD over 22 years of follow-up. Associations were independent of known liver cancer and CLD risk factors and suggest a potentially important role for glucose and insulin homeostasis in liver cancer and CLD mortality.
Supplementary Material
Acknowledgments
Funding/Support: This study was supported by the Intramural Research Program of the National Institutes of Health, Division of Cancer Epidemiology and Genetics, National Cancer Institute, Department of Health and Human Services.
Footnotes
Conflict of Interest: None to disclose.
References
- 1.National Center for Health Statistics. Health, United States, 2015: With Special Feature on Racial and Ethnic Health Disparities. Hyattsville, MD: 2016. [PubMed] [Google Scholar]
- 2.Ferlay J, Soerjomataram I, Ervik M, Dikshit R, Eser S, Mathers C, et al. GLOBOCAN 2012 v1.0, Cancer Incidence and Mortality Worldwide: IARC CancerBase No. 11. Lyon, France: International Agency for Research on Cancer; 2013. Available from: http://globocan.iarc.fr. [Google Scholar]
- 3.Siegel AB, Zhu AX. Metabolic syndrome and hepatocellular carcinoma: two growing epidemics with a potential link. Cancer. 2009;115:5651–5661. doi: 10.1002/cncr.24687. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.McGlynn KA, London WT. The global epidemiology of hepatocellular carcinoma: present and future. Clin Liver Dis. 2011;15:223–243. vii–x. doi: 10.1016/j.cld.2011.03.006. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.McGlynn KA, Petrick JL, London WT. Global epidemiology of hepatocellular carcinoma: an emphasis on demographic and regional variability. Clin Liver Dis. 2015;19:223–238. doi: 10.1016/j.cld.2015.01.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Center MM, Jemal A. International trends in liver cancer incidence rates. Cancer Epidemiol Biomarkers Prev. 2011;20:2362–2368. doi: 10.1158/1055-9965.EPI-11-0643. [DOI] [PubMed] [Google Scholar]
- 7.Altekruse SF, McGlynn KA, Reichman ME. Hepatocellular carcinoma incidence, mortality, and survival trends in the United States from 1975 to 2005. J Clin Oncol. 2009;27:1485–1491. doi: 10.1200/JCO.2008.20.7753. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.El-Serag HB, Tran T, Everhart JE. Diabetes increases the risk of chronic liver disease and hepatocellular carcinoma. Gastroenterology. 2004;126:460–468. doi: 10.1053/j.gastro.2003.10.065. [DOI] [PubMed] [Google Scholar]
- 9.El-Serag HB. Epidemiology of hepatocellular carcinoma in USA. Hepatol Res. 2007;37(Suppl 2):S88–S94. doi: 10.1111/j.1872-034X.2007.00168.x. [DOI] [PubMed] [Google Scholar]
- 10.Davila JA, Morgan RO, Shaib Y, McGlynn KA, El-Serag HB. Hepatitis C infection and the increasing incidence of hepatocellular carcinoma: a population-based study. Gastroenterology. 2004;127:1372–1380. doi: 10.1053/j.gastro.2004.07.020. [DOI] [PubMed] [Google Scholar]
- 11.Kanwal F, Hoang T, Kramer JR, Asch SM, Goetz MB, Zeringue A, et al. Increasing prevalence of HCC and cirrhosis in patients with chronic hepatitis C virus infection. Gastroenterology. 2011;140:1182–1188. doi: 10.1053/j.gastro.2010.12.032. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Di Bisceglie AM, Lyra AC, Schwartz M, Reddy RK, Martin P, Gores G, et al. Hepatitis C-related hepatocellular carcinoma in the United States: influence of ethnic status. Am J Gastroenterol. 2003;98:2060–2063. doi: 10.1111/j.1572-0241.2003.t01-1-07641.x. [DOI] [PubMed] [Google Scholar]
- 13.Wang C, Wang X, Gong G, Ben Q, Qiu W, Chen Y, et al. Increased risk of hepatocellular carcinoma in patients with diabetes mellitus: a systematic review and meta-analysis of cohort studies. Int J Cancer. 2012;130:1639–1648. doi: 10.1002/ijc.26165. [DOI] [PubMed] [Google Scholar]
- 14.Schlesinger S, Aleksandrova K, Pischon T, Jenab M, Fedirko V, Trepo E, et al. Diabetes mellitus, insulin treatment, diabetes duration, and risk of biliary tract cancer and hepatocellular carcinoma in a European cohort. Ann Oncol. 2013;24:2449–2455. doi: 10.1093/annonc/mdt204. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.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]
- 16.Gallagher EJ, LeRoith D. Epidemiology and molecular mechanisms tying obesity, diabetes, and the metabolic syndrome with cancer. Diabetes Care. 2013;36(Suppl 2):S233–S239. doi: 10.2337/dcS13-2001. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Khandekar MJ, Cohen P, Spiegelman BM. Molecular mechanisms of cancer development in obesity. Nat Rev Cancer. 2011;11:886–895. doi: 10.1038/nrc3174. [DOI] [PubMed] [Google Scholar]
- 18.Amaya MJ, Oliveira AG, Guimaraes ES, Casteluber MC, Carvalho SM, Andrade LM, et al. The insulin receptor translocates to the nucleus to regulate cell proliferation in liver. Hepatology. 2014;59:274–283. doi: 10.1002/hep.26609. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Page JM, Harrison SA. NASH and HCC. Clin Liver Dis. 2009;13:631–647. doi: 10.1016/j.cld.2009.07.007. [DOI] [PubMed] [Google Scholar]
- 20.Cohen JC, Horton JD, Hobbs HH. Human fatty liver disease: old questions and new insights. Science. 2011;332:1519–1523. doi: 10.1126/science.1204265. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Baffy G, Brunt EM, Caldwell SH. Hepatocellular carcinoma in non-alcoholic fatty liver disease: an emerging menace. J Hepatol. 2012;56:1384–1391. doi: 10.1016/j.jhep.2011.10.027. [DOI] [PubMed] [Google Scholar]
- 22.Starley BQ, Calcagno CJ, Harrison SA. Nonalcoholic fatty liver disease and hepatocellular carcinoma: a weighty connection. Hepatology. 2010;51:1820–1832. doi: 10.1002/hep.23594. [DOI] [PubMed] [Google Scholar]
- 23.White DL, Kanwal F, El-Serag HB. Association Between Nonalcoholic Fatty Liver Disease and Risk for Hepatocellular Cancer, Based on Systematic Review. Clin Gastroenterol Hepatol. 2012;10:1342–1359. doi: 10.1016/j.cgh.2012.10.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Lazo M, Hernaez R, Eberhardt MS, Bonekamp S, Kamel I, Guallar E, et al. Prevalence of nonalcoholic fatty liver disease in the United States: the Third National Health and Nutrition Examination Survey, 1988–1994. Am J Epidemiol. 2013;178:38–45. doi: 10.1093/aje/kws448. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.White DL, Ratziu V, El-Serag HB. Hepatitis C infection and risk of diabetes: A systematic review and meta-analysis. J Hepatol. 2008;49:831–844. doi: 10.1016/j.jhep.2008.08.006. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Nordenstedt H, White DL, El-Serag HB. The changing pattern of epidemiology in hepatocellular carcinoma. Dig Liver Dis. 2010;42(Suppl 3):S206–S214. doi: 10.1016/S1590-8658(10)60507-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Hickman IJ, Macdonald GA. Impact of diabetes on the severity of liver disease. Am J Med. 2007;120:829–834. doi: 10.1016/j.amjmed.2007.03.025. [DOI] [PubMed] [Google Scholar]
- 28.Chao LT, Wu CF, Sung FY, Lin CL, Liu CJ, Huang CJ, et al. Insulin, glucose and hepatocellular carcinoma risk in male hepatitis B carriers: results from 17-year follow-up of a population-based cohort. Carcinogenesis. 2011;32:876–881. doi: 10.1093/carcin/bgr058. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Matthews DR, Hosker JP, Rudenski AS, Naylor BA, Treacher DF, Turner RC. Homeostasis model assessment: insulin resistance and beta-cell function from fasting plasma glucose and insulin concentrations in man. Diabetologia. 1985;28:412–419. doi: 10.1007/BF00280883. [DOI] [PubMed] [Google Scholar]
- 30.Wallace TM, Levy JC, Matthews DR. Use and abuse of HOMA modeling. Diabetes Care. 2004;27:1487–1495. doi: 10.2337/diacare.27.6.1487. [DOI] [PubMed] [Google Scholar]
- 31.The ATBC Cancer Prevention Study Group. The alpha-tocopherol, beta-carotene lung cancer prevention study: Design, methods, participant characteristics, and compliance. Ann Epidemiol. 1994;4:1–10. doi: 10.1016/1047-2797(94)90036-1. [DOI] [PubMed] [Google Scholar]
- 32.Korhonen P, Malila N, Pukkala E, Teppo L, Albanes D, Virtamo J. The Finnish Cancer Registry as follow-up source of a large trial cohort--accuracy and delay. Acta Oncol. 2002;41:381–388. doi: 10.1080/028418602760169442. [DOI] [PubMed] [Google Scholar]
- 33.American Diabetes Association. Standards of medical care in diabetes--2012. Diabetes Care. 2012;35(Suppl 1):S11–S63. doi: 10.2337/dc12-s011. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Petrick JL, Freedman ND, Demuth J, Yang B, Van Den Eeden SK, Engel LS, et al. Obesity, diabetes, serum glucose, and risk of primary liver cancer by birth cohort, race/ethnicity, and sex: Multiphasic health checkup study. Cancer Epidemiol. 2016;42:140–146. doi: 10.1016/j.canep.2016.04.009. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Aleksandrova K, Boeing H, Nothlings U, Jenab M, Fedirko V, Kaaks R, et al. Inflammatory and metabolic biomarkers and risk of liver and biliary tract cancer. Hepatology. 2014;60:858–871. doi: 10.1002/hep.27016. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Saito K, Inoue S, Saito T, Kiso S, Ito N, Tamura S, et al. Augmentation effect of postprandial hyperinsulinaemia on growth of human hepatocellular carcinoma. Gut. 2002;51:100–104. doi: 10.1136/gut.51.1.100. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Miuma S, Ichikawa T, Taura N, Shibata H, Takeshita S, Akiyama M, et al. The level of fasting serum insulin, but not adiponectin, is associated with the prognosis of early stage hepatocellular carcinoma. Oncol Rep. 2009;22:1415–1424. doi: 10.3892/or_00000583. [DOI] [PubMed] [Google Scholar]
- 38.Everhart JE, Lok AS, Kim HY, Morgan TR, Lindsay KL, Chung RT, et al. Weight-related effects on disease progression in the hepatitis C antiviral long-term treatment against cirrhosis trial. Gastroenterology. 2009;137:549–557. doi: 10.1053/j.gastro.2009.05.007. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Veldt BJ, Poterucha JJ, Watt KD, Wiesner RH, Hay JE, Rosen CB, et al. Insulin resistance, serum adipokines and risk of fibrosis progression in patients transplanted for hepatitis. C. Am J Transplant. 2009;9:1406–1413. doi: 10.1111/j.1600-6143.2009.02642.x. [DOI] [PubMed] [Google Scholar]
- 40.Singh S, Singh PP, Singh AG, Murad MH, Sanchez W. Anti-diabetic medications and the risk of hepatocellular cancer: a systematic review and meta-analysis. Am J Gastroenterol. 2013;108:881–891. doi: 10.1038/ajg.2013.5. quiz 92. [DOI] [PubMed] [Google Scholar]
- 41.Kotronen A, Vehkavaara S, Seppala-Lindroos A, Bergholm R, Yki-Jarvinen H. Effect of liver fat on insulin clearance. Am J Physiol Endocrinol Metab. 2007;293:E1709–E1715. doi: 10.1152/ajpendo.00444.2007. [DOI] [PubMed] [Google Scholar]
- 42.Foxton MR, Quaglia A, Muiesan P, Heneghan MA, Portmann B, Norris S, et al. The impact of diabetes mellitus on fibrosis progression in patients transplanted for hepatitis. C. Am J Transplant. 2006;6:1922–1929. doi: 10.1111/j.1600-6143.2006.01408.x. [DOI] [PubMed] [Google Scholar]
- 43.Menke A, Casagrande S, Geiss L, Cowie CC. Prevalence of and Trends in Diabetes Among Adults in the United States, 1988–2012. JJAMA. 2015;314:1021–1029. doi: 10.1001/jama.2015.10029. [DOI] [PubMed] [Google Scholar]
- 44.Limburg PJ, Stolzenberg-Solomon RZ, Vierkant RA, Roberts K, Sellers TA, Taylor PR, et al. Insulin, glucose, insulin resistance, and incident colorectal cancer in male smokers. Clin Gastroenterol Hepatol. 2006;4:1514–1521. doi: 10.1016/j.cgh.2006.09.014. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Stolzenberg-Solomon RZ, Graubard BI, Chari S, Limburg P, Taylor PR, Virtamo J, et al. Insulin, glucose, insulin resistance, and pancreatic cancer in male smokers. JAMA. 2005;294:2872–2878. doi: 10.1001/jama.294.22.2872. [DOI] [PubMed] [Google Scholar]
- 46.Albanes D, Weinstein SJ, Wright ME, Mannisto S, Limburg PJ, Snyder K, et al. Serum insulin, glucose, indices of insulin resistance, and risk of prostate cancer. J Natl Cancer Inst. 2009;101:1272–1279. doi: 10.1093/jnci/djp260. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47.Wu T, Dorn JP, Donahue RP, Sempos CT, Trevisan M. Associations of serum C-reactive protein with fasting insulin, glucose, and glycosylated hemoglobin: the Third National Health and Nutrition Examination Survey, 1988–1994. Am J Epidemiol. 2002;155:65–71. doi: 10.1093/aje/155.1.65. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.