If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
To provide a quantitative assessment of the association between excess body weight (EBW) and the risk of primary liver cancer (PLC), we performed an updated meta-analysis of prospective observational studies.
Methods
We searched PUBMED and EMBASE for studies of body mass index and the risk of PLC published through 15 th September 2011. Summary relative risks (SRRs) with their corresponding 95% confidence intervals (CIs) were calculated using a random-effects model. The meta-regression and stratified methods were used to examine heterogeneity across studies.
Results
A total of 26 prospective studies, including 25,337 PLC cases, were included in this analysis. Overall, excess body weight (EBW: body mass index (BMI) ⩾ 25 kg/m2) and obesity (BMI ⩾ 30 kg/m2) were associated with an increased risk of PLC, with significant heterogeneity (EBW: SRRs 1.48, 95%CIs 1.31–1.67, Ph < 0.001, I2 = 83.6%; Obesity: SRRs 1.83, 95% CIs 1.59–2.11, Ph < 0.001, I2 = 75.0%). Subgroup analyses revealed that the positive associations were independent of geographic locations, alcohol consumption, history of diabetes or infections with hepatitis B (HBV) and/or hepatitis C virus (HCV). Obese males had a higher risk of PLC than obese females did (P = 0.027). A stronger risk of PLC with EBW was observed for patients with HCV (but not HBV) infection or cirrhosis compared with the general population.
Conclusions
Findings from this meta-analysis strongly support that EBW or obesity is associated with an increased risk of PLC in both males and females.
According to the definition from World Health Organization (WHO), there are an estimated 1.6 billion adults to be overweight, and at least 300 million of them to be obese worldwide in the year 2006.
Epidemiological evidence has linked excess body weight (EBW) with an increased risk for several cancers including cancers of the colon, breast (in premenopausal women), endometrium, pancreas, oesophagus and liver.
Histologically, hepatocellular carcinoma (HCC) accounts for between 85% and 90% of all PLC. It has been established that chronic infection with hepatitis B (HBV) or hepatitis C virus (HCV) is the leading cause of HCC worldwide.
Additionally, alcohol abuse, cigarette smoking, a history of diabetes mellitus (DM) and environmental exposure to aflatoxin B1 also increase the risk of HCC development.
Changes in HCC incidence and prevalence of overweight and obesity, as measured by body mass index (BMI), have both followed a similar upwards pattern, suggesting that overweight and obesity may account for a significant proportion of HCC cases. A recent meta-analysis of 11 cohort studies showed increased PLC risks of 17% for overweight people and of 90% for obese ones compared to those in normal weight.
Since then, lots of prospective studies, particularly from Asia (the endemic area for HBV/HCV), on the association between BMI and PLC risk have been published.
Metabolic factors and subsequent risk of hepatocellular carcinoma by hepatitis virus infection status: a large-scale population-based cohort study of Japanese men and women (JPHC Study Cohort II).
The aim of this meta-analysis was to update and expand the previous meta-analysis including all prospective studies on this issue published through 15th September 2011. Furthermore, we evaluated whether the risk of PLC varied between genders and sources of population.
2. Materials and methods
2.1 Search strategies
Two of us (C.Y. and W.X.L.) identified eligible studies by a comprehensive search in PubMed and EMBASE (up to 15th September 2011). Research papers were selected using the following keywords or Medical Subject Heading (MeSH) terms: ‘body mass index’, ‘BMI’, or ‘obesity’ or ‘excess body weight’; ‘liver cancer’ or ‘hepatocellular carcinoma’ or ‘HCC’; ‘risk’ or ‘incidence’ or ‘mortality’. No language restrictions were imposed. Furthermore, the reference lists of the retrieved articles were examined for additional relevant studies.
2.2 Inclusion and exclusion criteria
Three authors (C.Y, W.X.L., and W.J.H.) independently evaluated all of the studies retrieved according to the pre-specified selection criteria. Discrepancies between the three reviewers were solved by discussion. To be included, the study had to meet the following criteria:(1) published as an original article; (2) with a prospective design and (3) reporting relative risk (RR) estimates with corresponding 95% confidence intervals (CIs) for BMI and PLC (or HCC); (4) the RR and corresponding 95% CIs were at least adjusted for age. If results based on the same study population were reported in more than one study, we included the one with the largest number of cases. We excluded studies that did not provide risk estimates, only provided an RR with corresponding 95% CIs per unit increase in BMI or duplicate publications.
2.3 Data extraction
The following information from each included study was extracted using a standardized data-collection protocol: the first author’s last name, country of origin, publication year, sample size, definition of the study population, ascertainment of exposure and outcome, duration of follow-up and variables adjusted for in the analysis. When several risk estimates were presented, we used the ones adjusted for the largest number of potential confounders. Data abstraction was performed independently by two investigators (C.Y. and W.X.L.) and then cross-checked.
2.4 Statistical analysis
We used body mass categories according to the definition from WHO: normal weight (BMI ⩾18.5 and <25 kg/m2), overweight (BMI ⩾25 and <30 kg/m2) and obese (BMI ⩾ 30 kg/m2). Combined overweight and obesity may be expressed as EBW (BMI ⩾ 25 kg/m2). For studies that reported RRs for several categories of BMI that fell into the category representing overweight, obesity or EBW, we pooled the relative risks and used the pooled estimates in the meta-analysis. If studies reported results separately for men and women, we combined the sex-specific estimates to generate an estimate for both genders combined. Summary RR (SRR) estimates with their corresponding 95% CIs were derived with the method of DerSimonian and Laird using the assumptions of a random-effects model, which incorporates between-study variability.
A two-tailed P < 0.05 was considered statistically significant.
Heterogeneity among studies was assessed using Q and I2 statistics, which test total variation across studies that was attributable to heterogeneity rather than to chance.
Potential sources of heterogeneity were investigated by using subgroup analyses and restricted maximum likelihood (REML)-based random-effects meta-regression analysis gender (males and females), geographic locations (Asia and non-Asia), ascertainment of exposure (self-reported and measured) ascertainment of outcome (cancer/death registry and histological finding or non-invasive diagnosis), the number of cases(<120 and ⩾120 cases of PLC), duration of follow-up (<10 ys and ⩾10 ys) for categories of EBW and obesity, respectively, because these two categories covered categories of overweight. We also performed subgroup meta-analyses by studies that did or did not adjust for the following confounders: alcohol use, infection with HBV and/or HCV and a history of diabetes.
Univariate meta-regression analyses were performed first. Variables that were significant at the 0.1 level were entered into the multivariable model. Sensitivity analysis was also conducted to estimate the influence of each individual study on the summary results by repeating the random-effects meta-analysis after omitting one study at a time.
To test evidence of publication bias, funnel plots and statistical test for funnel plot asymmetry were performed by using Begg’s adjusted rank correlation test.
All statistical analyses were performed using STATA, version 11.0 (STATA, College Station, TX, United States of America (USA)) and R-package statistical software (Version 2.11.0 beta).
3. Results
3.1 Search results and study characteristics
A total of 26 prospective studies, which included 9,053,369 of study population and 25,337 PLC (or HCC) cases, were found to match our inclusion criteria (11–19, 26–42). Of these 26 studies, 14 studies were conducted in Asia, four in the United States, six in Europe and two in Asia Pacific region. Characteristics of studies included in the meta-analysis are presented in Table 1.
Table 1Characteristics of prospective studies of body mass index and risk of primary liver cancer.
Effects of excess weight on cancer incidences depending on cancer sites and histologic findings among men: Korea National Health Insurance Corporation Study.
Metabolic factors and subsequent risk of hepatocellular carcinoma by hepatitis virus infection status: a large-scale population-based cohort study of Japanese men and women (JPHC Study Cohort II).
The combined results based on all studies showed that there was a statistically significant link between EBW, overweight or obesity and PLC risk, with significant heterogeneity among studies (EBW, SRR = 1.49, 95% CIs = 1.31–1.68, Q = 122.23, Ph < 0.001, I2 = 83.6, n = 21; overweight, SRR = 1.18, 95% CIs = 1.06–1.31, Q = 33.71, Ph = 0.001, I2 = 61.4, n = 14; obesity, SRR = 1.83, 95% CIs = 1.59–2.11, Q = 56.14, Ph < 0.001, I2 = 75.0, n = 19) (Fig. 1A–C).
Fig. 1Forest plots of risk of primary liver cancer associated with excess body weight, overweight and obesity. Squares represent the study-specific relative risk. Diamonds represent the summary relative risks (SRRs). Horizontal lines represent 95% confidence intervals (CIs). (A) Forest plots of risk of primary liver cancer associated with excess body weight (EBW, body mass index (BMI) ⩾25 kg/m2). (B) Forest plots of risk of primary liver cancer associated with overweight (BMI ⩾25 and <30 kg/m2). (C) Forest plots of risk of primary liver cancer associated with obesity (BMI ⩾ 30 kg/m2).
We then conducted subgroup meta-analyses by gender, geographic locations, ascertainment of exposure and outcome, the number of cases, duration of follow-up and confounders. The links between EBW and obesity and PLC risk were positive in all strata (Table 2).
Table 2Stratified meta-analyses of excess body weight (BMI ⩾25 kg/m2) and obesity (BMI ⩾30 kg/m2) and risk of primary liver cancer.
The duration of follow-up in one study29 was not available.
<10
9
4756
1.45(1.17–1.80)
<0.001
87.3
8
4269
1.81(1.42–2.30)
0.019
58.2
⩾10
12
14585
1.47(1.25–1.73)
<0.001
0.846
80.3
10
18,109
1.88(1.54–2.30)
<0.001
0.812
75.1
Adjustment for confounders
Alcohol use
No
7
11198
1.53(1.13–2.07)
<0.001
88.2
10
16,996
1.98(1.60–2.46)
<0.001
77.5
Yes
14
8143
1.46(1.25–1.69)
<0.001
0.817
81.7
9
5382
1.69(1.41–2.04)
0.078
0.273
43.5
HBV and/or HCV
No
9
14742
1.28(1.14–1.43)
<0.001
73.1
12
20,182
1.81(1.52–2.14)
<0.001
71.1
Yes
12
4599
1.74(1.35–2.25)
<0.001
0.03
86.5
7
2196
1.89(1.43–2.50)
0.015
0.796
61.8
Diabetes
No
15
18172
1.40(1.23–1.59)
<0.001
82.5
15
21,319
1.77(1.51–2.07)
<0.001
68.4
Yes
6
1169
1.70(1.27–2.27)
<0.001
0.255
76.9
4
1059
2.06(1.53–2.78)
0.077
0.379
56.2
Bold values indicate statistical significance at the p value equal to 0.10 level.
Abbreviations: RR, relative risk; CI, confidence interval; HBV, hepatitis B virus; HCV, hepatitis C virus; BMI, body mass index; Ph, p value for heterogeneity; Pd, p value for difference.
Metabolic factors and subsequent risk of hepatocellular carcinoma by hepatitis virus infection status: a large-scale population-based cohort study of Japanese men and women (JPHC Study Cohort II).
Effects of excess weight on cancer incidences depending on cancer sites and histologic findings among men: Korea National Health Insurance Corporation Study.
there was significant difference in RR between the strata of genders and confounder (infection with HBV and/or HCV). When stratifying for gender, we found that males with EBW had a higher risk of PLC development than females with EBW (SRR = 1.42, 95% CI 1.22–1.65, n = 11 for males versus SRR = 1.18, 95% CI 1.08–1.30, n = 5 for females; Pd = 0.041). The SRRs of PLC with EBW for studies adjusted for infection with HBV and/or HCV were significantly stronger than those for studies not adjusted for it (SRRs, 1.74 versus 1.28; Pd = 0.03). There was no difference in PLC risk with EBW between strata in ascertainment of exposure and outcome, the number of cases, duration of follow-up and confounders (alcohol consumption, history of DM) (Table 2).
Similarly, for studies on the association between obesity and PLC,
Effects of excess weight on cancer incidences depending on cancer sites and histologic findings among men: Korea National Health Insurance Corporation Study.
obese males had a stronger risk of PLC development than obese females (males: SRR = 1.91, 95% CI 1.51–2.41, n = 12; females: SRR = 1.55, 95% CI 1.30–1.85, n = 6; Pd = 0.027). In addition, geographic locations were also found to significantly modify the association between obesity and PLC risk. The SRRs were significantly higher in non-Asian studies (Europen/North American) than that in Asian studies (non-Asia: SRRs 2.09, 95% CI 1.69–2.59; Asia: SRRs 1.54, 95% CI: 1.33–1.78; Pd = 0.02). However, no differences in the association between obesity and the risk of PLC were found between strata in ascertainment of exposure and outcome, the number of cases, duration of follow-up and confounders (alcohol use, history of DM and infection with HBV and/or HCV) (Table 2).
We then conducted meta-regression analyses to investigate the sources of heterogeneity among studies according to the above subgroups. For studies of EBW, in univariate meta-regression analyses, only confounder (adjustment for infection with HBV and/or HCV; P = 0.077) was found to be a significant factor. The between-study variance was reduced from 0.0894 to 0.0802 based on REML estimate, and the heterogeneity explained by this confounder was 10.2%. For studies of obesity, geographic location was the only significant factor (P = 0.06) and the heterogeneity explained by geographic location was 13.2%.
We investigated the effect of EBW and obesity on PLC risk in population of specific liver diseases (patients with HBV and/or HCV infection or cirrhosis) (Fig. 2A and B). For studies assessing EBW and PLC risk (Fig. 2A), the SRRs of this association for participants with specific liver diseases (3222 cases of PLC) were comparable to those for general population (specific liver diseases: SRR 1.73, 95% CI 1.28–2.35, n = 9 versus general population: SRR 1.36, 95% CI 1.20–1.53, n = 17; Pd = 0.149). Further analysis indicated that patients with HCV infection (SRR 2.15, 95% CI 1.50–3.09, n = 4, 2337 cases of PLC) had a significantly stronger risk of PLC associated with EBW than general population with EBW had (Pd = 0.059). Similarly, the SRRs of this association for cirrhotic patients were 2.41 (95% CI 1.95–2.98, n = 2320 cases of PLC), which were also higher than those for the counterparts of general population (Pd = 0.002). However, the SRRs of this association for patients with HBV infection were not the case (SRR 1.27, 95% CI 0.93–1.73, n = 4, 565 cases of PLC; Pd = 0.687). For studies assessing the association between obesity and PLC risk (Fig. 2B), the SRRs of this association were also similar for participants with specific liver diseases (1693 cases of PLC) with those for general population (specific liver diseases: SRR 2.05, 95% CI 1.50–2.80, n = 5 versus general population: SRR 1.78, 95% CI 1.50–2.08, n = 14; Pd = 0.432). Further analysis found that patients with HCV infection (392 cases of PLC) had a stronger risk of PLC associated with obesity than general population with obesity had (Pd = 0.082), however, patients with HBV infection (322 cases of PLC) or with cirrhosis (979 cases of PLC) had a similar risk compared with counterparts of general population (Pd = 0.163 for HBV and Pd = 0.319 for cirrhosis, respectively).
Fig. 2The pooled estimates of the association between excess body weight (EBW, body mass index (BMI) ⩾25 kg/m2 (A) and obesity (BMI ⩾30 kg/m2, (B) and primary liver cancer risk in population of specific liver diseases (patients with HBV or HCV infection or cirrhosis) and counterparts of the general population.
We also conducted a sensitivity analysis by omitting one study at a time and calculating the pooled RRs for the remainder of studies, and found that there were no changes in the direction of effect when any one study was excluded. For example, when study by Jee et al.
was excluded (which seemed to have a strong influence on the meta-analysis estimate of effect), the pooled RR remained the same (SRR 1.68, 95% CI 1.46–1.90).
We found no indication of publication bias for studies on the association between BMI and PLC risk in all the analyses, Begg’s test showed P = 0.216 for EBW and PLC, 0.228 for overweight, and 0.529 for obesity.
4. Discussion
In this meta-analysis, we found that overweight, obesity and EBW were associated with 18%, 83% and 48% increased risk of PLC, respectively. This increased risk of liver cancer seemed to be independent of gender, geographic locations, case size, ascertainment of exposure and outcome, duration of follow-up and confounders. Like the previous meta-analysis, a higher risk of PLC with EBW or obesity was seen in males than that in females. Importantly, patients with HCV infection or cirrhosis had a higher risk of PLC with EBW than the counterparts of general population did.
The exact biologic mechanisms underlying the association between EBW and increased risk of PLC remain unclear, but certainly involve the development of non-alcoholic fatty liver disease (NAFLD) and subsequent non-alcoholic steatohepatitis (NASH). NAFLD is the chief hepatic manifestation of obesity, diabetes mellitus and metabolic syndrome, all of which are related to insulin resistance.
Insulin resistance leads to elevated levels of the pro-inflammatory cytokine, such as tumor necrosis factor (TNF) and interleukin (IL)-6, which favour the development of hepatic steatosis and inflammation and subsequent cancer within the liver.
In addition, hyperinsulinemia may up-regulate the production of insulin-like growth factor-1 (IGF-1), which stimulates cellular proliferation and inhibits apoptosis within the liver.
Relationship of insulin-like growth factors system gene polymorphisms with the susceptibility and pathological development of hepatocellular carcinoma.
To our knowledge, the strengths of this study include as follows: (1) our meta- analysis was based on 26 prospective studies, which might minimize the possibility of recall or selection bias. (2) All the included studies evaluated multiple potential confounders and the relationships between BMI and PLC risk in each study were derived from regression after adjustment at least for age, and most of the studies were adjusted for the important risk factors for PLC, such as alcohol use, infection of HBV and/or HCV, a history of DM. (3) The large number of studies with different populations expand prior observational studies by permitting additional evaluation of subgroups (e.g. by gender and sources of population), which may permit us to more precisely evaluate risk with EBW or obesity on different subgroups, and to make some inferences with regards to the different population, especially patients with specific liver disease.
As with any meta-analysis of observational studies, our study has limitations. First, great heterogeneity was presented across studies, which would throw some doubt on the reliability of the summary RR estimates. The significant heterogeneity may exist in terms of geographic location, ascertainment of anthropometry and outcome, duration of follow-up, sources of population and confounders. We used subgroup and REML meta-regression analyses to explore the sources of heterogeneity, and confounder (infection with HBV and/or HCV) or geographic location was found to be significant factor, and accounted for only 10.2% or 13.2% of heterogeneity across studies. Second, the possibility that the observed relationship between EBW and PLC risk due to unmeasured or residual confounding should be considered, since inadequate control for confounders may bias the results towards exaggeration or underestimation of risk estimates. Obesity tends to be associated with unhealthy behaviours linked to increased risk of PLC, such as heavy alcohol consumption and a history of DM, both of which are associated with increased PLC risk. However, a positive association between BMI and PLC risk persisted when we restricted the meta-analysis to studies that controlled for both confounders, respectively. These results exclude the possibility that the association between obesity and PLC is explained by diabetic status and alcohol use. Third, several studies in this meta-analysis relied on self-reported anthropometric measures, which may have led to some underestimation of the true associations. However, the SRR estimates for the studies that had measured weight and height were similar with those for studies that relied on self-reporting. Finally, as in any meta-analysis, it is possible that an observed association is the result of publication bias, because small studies with null results tend not to be published. However, the results obtained from funnel plot analysis and formal statistical tests did not provide evidence for such bias.
In agreement with previous meta-analysis by Larrson et al.
we also found that the summary RR of PLC for obesity was statistically significantly higher for men (SRR = 1.91) than for women (SRR = 1.55). These results were based on more researches and confirm previous conclusions. The reasons for the apparent sex difference in the associations for EBW remain elusive, but might be related to differences between males and females in the association between adiposity and steroid hormone concentrations. Obesity is inversely related to testosterone concentrations in males
Moreover, findings from a meta-analysis indicated that high testosterone concentrations were associated with a higher risk of type 2 DM in females but with a lower risk in males.
Given insulin resistance is one of the important mechanisms linking obesity to risk of PLC, a reduction in testosterone concentrations due to obesity in males may be one reason for the stronger association of obesity with PLC risk in males than in females. Adiposity is positively associated with circulating concentrations of estradiol in postmenopausal females and males.
In addition, epidemiological studies have demonstrated a clear male predominance in the diagnosis of NAFLD, while higher prevalence rates of NAFLD in postmenopausal women.
Other potential explanations for the sex difference might be related to sex-specific differences in exposure to risk factors. Men are more likely to be infected with hepatitis B and C viruses, consume alcohol and smoke cigarettes.
In a subgroup analysis stratified by Asians and non-Asians, we found that obesity was associated with a significantly higher risk of PLC among non-Asians than that among Asians. The difference in the prevalence of obesity and incidence of PLC between Asians and non-Asians may partly explain the difference between Asians and non-Asians.
We found a higher summary RR estimate of PLC in EBW individuals with HCV infection than that in general population with EBW (P = 0.059). Although it is well established that patients with HCV infection are at increased risk for the development of HCC, it is unclear what role steatosis plays in the development of HCC. Prior data have shown an association of chronic HCV infection and hepatic steatosis in a large portion of cases, which has also been shown to be a risk factor for liver disease progression.
In humans, steatosis may be a common mediator of HCC in hepatitis C through enhanced oxidative stress, increased susceptibility to apoptosis and activation of subsinusoidal stellate cells.
In summary, findings of this meta-analysis provide evidence that EBW may significantly increase PLC risk, and suggest that PLC may, at least partly, be prevented by maintaining a healthy body weight.
Conflict of interest statement
None declared.
References
Polesel J.
Zucchetto A.
Montella M.
et al.
The impact of obesity and diabetes mellitus on the risk of hepatocellular carcinoma.
Metabolic factors and subsequent risk of hepatocellular carcinoma by hepatitis virus infection status: a large-scale population-based cohort study of Japanese men and women (JPHC Study Cohort II).
Effects of excess weight on cancer incidences depending on cancer sites and histologic findings among men: Korea National Health Insurance Corporation Study.
Relationship of insulin-like growth factors system gene polymorphisms with the susceptibility and pathological development of hepatocellular carcinoma.
The duration of follow-up in one study
00217-1&id=gr1.jpg){kind=link}
00217-1&id=gr2.jpg){kind=link}