Prognostic role of diabetes mellitus in hepatocellular carcinoma patients after curative treatments: a meta-analysis

Wei-Min Wang, Yang Xu, Xin-Rong Yang, Yao-Hui Wang, Hai-Xiang Sun and Jia Fan

Shanghai, China

Meta-analysis

Prognostic role of diabetes mellitus in hepatocellular carcinoma patients after curative treatments: a meta-analysis

Wei-Min Wang, Yang Xu, Xin-Rong Yang, Yao-Hui Wang, Hai-Xiang Sun and Jia Fan

Shanghai, China

BACKGROUND:The prognostic role of diabetes mellitus (DM) coexisting with hepatocellular carcinoma (HCC) remains controversial. To clarify its impact on survival in HCC patients after curative treatments, a meta-analysis was performed.

DATA SOURCES:Eligible studies were identified through multiple search strategies in the databases PubMed (MEDLINE), EMBASE, the Cochrane Library and ACP Journal Club between January 1950 and March 2010. Ten studies fulfilled the inclusion criteria, and data were aggregated comparing overall survival and recurrence-free survival in HCC patients according to DM status.

RESULTS:The pooled hazard ratios (HRs) estimate for overall survival was 1.34 (95% CI, 1.18-1.51;P<0.0001) and for recurrence-free survival was 1.48 (95% CI, 1.00-2.18;P<0.0001), showing a worse survival for HCC with coexisting DM. However, the patients with DM had a shorter survival time in HCV-related HCC (HR=1.71; 95% CI, 1.10-2.66;P=0.016), while HBV-related cases were not significantly different (HR=1.29; 95% CI, 0.69-2.40;P=0.182). Meanwhile, the coexistence of DM impaired overall survival in HCC patients with a small tumor burden (HR=1.63; 95% CI, 1.25-2.12;P<0.0001).

CONCLUSION:HCC patients with coexisting DM have a shorter survival time and a higher risk for tumor recurrence after curative treatments, while the precise value should be defined in more clinical trials with consistent methodology, especially prospective studies.

(Hepatobiliary Pancreat Dis Int 2011; 10: 346-355)

hepatocellular carcinoma; diabetes mellitus; prognostic factors; survival; meta-analysis

Introduction

Hepatocellular carcinoma (HCC) is the fifth most frequent malignant neoplasm, and its incidence and mortality rates have increased in recent years.[1]Although the survival of patients with HCC has been improved by advances in surgical techniques and perioperative management, long-term survival remains unsatisfactory owing to the high rate of recurrence and metastasis.[2,3]The dismal clinical outcome of HCC leads to an urgent need for in-depth understanding of the relevant factors affecting HCC prognosis after treatment. These may serve to guide decision-making for therapeutic strategies for HCC patients and improve their prognosis.

Diabetes mellitus (DM) is positively associated with risks of several common human malignancies, including cancers of the colon, breast, endometrium, pancreas and liver.[4]Compared with their non-diabetic counterparts, patients with pre-existing DM have a higher risk for developing HCC, and it has been suggested as a potential risk factor for HCC.[5-8]Given the higher risk of HCC in patients with DM, investigation of how preexisting DM may influence the prognosis of HCC after treatment is critical to decide the proper care of these patients. Recent researches have focused on the effects of comorbid conditions on the prognosis of HCC patients after treatment.[9-11]But potential interactions between DM and HCC are so complex that a conclusion remains controversial.[12-14]

To clarify a more precise effect of DM on HCC patients after curative treatments, we conducted a systematic review and meta-analysis to test the hypothesis that preexisting DM has an adverse effect on prognosis inpatients with HCC after curative treatments. Ten studies which fulfilled the selection criteria were included in a standard meta-analysis. Using this strategy to clarify the potential connection between DM and the prognosis of HCC patients may be helpful to clinical practice and enrichment of our knowledge.

Methods

Eligibility criteria

Studies must have evaluated the correlation of DM status with survival of HCC patients, and only trials providing information about overall survival (OS) and/or recurrence-free survival (RFS) were included. Studies not directly reporting hazard ratios (HRs) were allowed only if data were available for statistical estimation as described below. If the same patient populations were reported in several publications, only the most recent and/or complete one was included in the analysis to avoid overlap between cohorts. The criteria of DM diagnosis were based on the definitions described by the World Health Organization[15]or the American Diabetes Association.[16]

Identification of studies

Articles were identified by an electronic search of PubMed (MEDLINE), EMBASE, the Cochrane Library and ACP Journal Club in July 2009 and then repeated in March 2010. The key words included terms for diabetes mellitus (e.g., diabetes, glucose intolerance, insulin resistance, hyperglycemia, hyperinsulinemia), hepatocellular carcinoma (e.g., liver cancer, liver neoplasm, liver malignant neoplasm), and prognosis (e.g., overall survival, disease-free survival, outcome, mortality, recurrence). The language was restricted to English, and studies were only on humans. Only full papers and published studies in the medical literature were included. Data from abstracts, review articles, editorials, case reports, and letters were not included. Characteristics of the studies were extracted from published articles and summarized in a uniform manner to aid comparison. All candidate articles were read independently by 4 reviewers (Wang WM, Xu Y, Yang XR and Fan J) and scored according to NOQAS.[17]This scoring system is described in a recent systematic review.[17]We only changed the contents of "comparability (1) a" to "study controls for tumor grade". When necessary, authors were contacted. Any discrepancies were resolved by a consensus in regular meetings attended by at least three-quarters of the investigators.

Statistical analysis

The association between DM status, OS and RFS was derived by summarizing the loge(HRi) (LnHRi) from the eligible studies with inverse variance weights. The LnHRi and the corresponding standard error [SE(LnHRi)] were used as data points for the meta-analysis. LnHRi was estimated from the presented parameters (O-E statistic and its variance, HR point estimate or its CI, events in each arm or total events, numbers randomized on each arm, P value, numbers at risk in each group) using the methods described by Parmar et al.[18]If no numerical data for the estimation of summary statistics were given, data were extracted manually from Kaplan-Meier survival curves to reconstruct the HR estimate. The rate of patients censored was presumed constant during the study follow-up.[18]When censoring data were presented, censored subjects were allocated to the appropriate time interval. Survival curves were read by Engauge Digitizer 4.1 software (http://digitizer.sourceforge.net/) after magniif cation to improve the accuracy. In the study by Nagasue et al,[19]insuf ficient data were presented to extract data of OS/RFS by any of the methods described above. No signi ficant difference was reported in survival outcome between the coexistence of DM and without DM in HCC.

For the meta-analysis, the HR was assigned to be 1.00, and a variance of LnHRi was assumed for similar sized studies (Kawamura et al[12]) to avoid selection bias.[20]P values quoted less than the speci fied threshold were assumed to be at the threshold, resulting in a conservative estimate of signi ficance level.[21]

The combined HR was displayed in Forest plots. And I2and Q estimates were performed in statistical assessment. If the I2value was >50% or the P value was less than 0.1, representing signi ficant heterogeneity, a Der Simonian-Laird random-effects model (D+L) was used. Otherwise, an inverse variance fixed-effects model (I-V) was selected.[22]Assessment of publication bias was performed using the Begg's/Egger's bias indicator test.[23]For all tests, a P value was two-sided and less than 0.05 was considered statistically signi ficant unless specially described.

All statistical analyses were conducted using the HR calculations spreadsheet in Microsoft Excel[24]and Stata 11.0 statistical software (Stata Corp, College Station, TX, USA).

Results

Eligible studies

Eighty-two potential articles were included through the search strategy. After viewing the contexts, 29 candidate studies were identified according to the requirements. Upon further analysis, 3 were aboutliver transplantation,[25-27]2 just reflected that DM is a risk factor for hepatic decompensation,[28,29]1 showed that DM is linked with acute renal failure, which is a poor predictor for HCC,[30]1 identified the risk factors predicting major postoperative complications,[31]3 were about glucose intolerance and insulin resistance,[32-34]and 19 related to survival/mortality stratified by DM status. Among the 19 potential studies, 9 were eliminated due to inadequate data[35-43]and duplicate records for analysis[13,44](Fig. 1). While the duplicate study separately analyzed the impact of DM on the OS and recurrence of HBV- or HCV-related HCC undergoing resection,[44]we selected it for the survival analysis of DM and the interaction with viral factors. Finally, 10 eligible studies were included in this meta-analysis.[9-14,19,45,46]All 4081 patients (sample size range 40-1713) in this study came from Asia. The mean age of most patients was 52.4-67.4 years (range 33.0-79.0), and 85% were male. The research period was from 1985 to 2008, and the median followup time was explicitly stated in 7 studies (range 18-54 months; median 33 months). Five trials (containing 1275 patients) analyzed the relationship of DM and RFS, and the other 5 studies were about the cumulative recurrence rate (RR). The characteristics of these studies are summarized in Table 1.

Survival analysis

Of the 10 eligible studies for pooling of OS/RFS data, none provided data for a direct estimation of LnHRi. In the analysis of the impact of DM status on OS of HCC patients, 8 HRi values were above 1.00, indicating that HCC patients with DM had a worse prognosis. Of these patients, 4 had a lower 95% CI above 1.00, suggesting a significant effect.

The forest plot in Fig. 2A shows the HRi and 95% CI for all studies in OS, and a pooled HR of 1.34 (95% CI, 1.18-1.51; P<0.0001) using a fixed-effects analysis, and of 1.37 (95% CI, 1.17-1.60; P<0.0001) in a randomeffects analysis. DM was confirmed as a poor prognostic factor of HCC in OS. No evidence of heterogeneity was found between studies (Q=10.78, I2=16.5%, P=0.291), and the result of the fixed-effects model was accepted preferentially. To assess the effect of assuming an HRi of 1.00 for OS, it[19]was excluded and reanalyzed. It remained unchanged in the pooled HR=1.35 (95% CI, 1.19-1.53, P<0.0001; Q=10.20, I2=21.6%, P=0.251). One study[9]with the largest weights was excluded, and HR was 1.46 (95% CI, 1.23-1.74, P<0.0001; Q=8.74, I2=8.4%, P=0.365). Similarly, the overall effect remained virtually unchanged (HR=1.32, 95% CI, 1.16-1.50, P<0.0001; Q=9.81, I2=18.4%, P=0.279) after removal of the lowerquality point study.[10]Integrating the HRi of 6 studies with curative treatments,[11-14,19,45]HR was 1.43 (95% CI, 1.18-1.75,P<0.0001;Q=8.18,I2=38.9%,P=0.147) shown by fixed-effects analysis (Table 2, Fig. 3).

Fig. 1. Flowchart of study identification, rejection and selection in the meta-analysis.

Table 1. Individual studies in meta-analysis

Limiting analysis to the 5 studies[11,12,19,45,46]assessing DM coexisting in Japanese HCC patients revealed the pooled HR to be 1.60 (Table 2). Four studies estimated this in a Chinese population,[9,13,14]and the summary HR was 1.21 (Table 2, Fig. 4).

Prognostic significance of DM only occurred in HCV-related HCC patients (P=0.016). HR was 1.29 in HBV-related HCCs, with evidence of study heterogeneity, while the HR was 1.71 in the HCV-related HCCs (Table 2, Fig. 5).

According to tumor size, the pooled HR was 1.63 and 0.67 in the groups with tumor diameter ≤5 cm and >5 cm respectively (Table 2).[13,14,46]DM coexistence impaired OS (HR=1.63, P<0.0001) in HCC patients with tumor diameter ≤5 cm (Fig. 6).

Impact of DM on RFS and RR

For RFS, 3 lower 95% CI of the HRi values were above 1.00.[11,13,45]For RR, 2 stated lower 95% CI >1.00 and DM worsened the RR.[11,12]HRi values were recorded for the included studies using available data and the techniques described above.

Fig. 2. A: Forest plot of HR for OS from HCC associated with coexisting DM status. DM coexistence was a predictor of worse survival of OS in HCC (HR=1.34, P<0.0001). B: Forest plot of HR for RFS from HCC associated with DM status in 5 studies. The summary HR value (HR=1.48, P<0.0001) implies a worse RR for the group with DM. C: Forest plot of HR for RR from HCC associated with DM status in 5 studies. The summary HR value (HR=1.29, P=0.022) implies a worse RR for the group with DM. It includes combined HR calculated using both the general inverse variance fixed effects (I-V) and Der Simonian-Laird random effects models (D+L) as well as the evaluation for heterogeneity (I2). Horizontal lines represent 95% CI. Black boxes indicate the HRi point estimate, and their areas are proportional to the weights of the studies. The broken line and diamond represent the summary estimate. The unbroken vertical line is at the null value (HR=1.0).

Table 2. Results of meta-analysis

Fig. 3. Forest plot of the HR of OS on subgroup analysis of those undergoing curative treatments. HCC patients with DM had a poor OS (pool HR=1.43, P<0.0001).

Fig. 4. Forest plot of the HR of OS on subgroup analysis based on genetic backgrounds.

In Fig. 2B, the forest plot displays the HRi and 95% CI for 5 studies in RFS, and the summary HR was 1.48 (Table 2) using a random-effects analysis. The heterogeneity was discovered using a restrictive I2test.[14]Omitting this study gave a similar result (HR=1.70, 95% CI, 1.35-2.15, P<0.0001; Q=4.00, I2=25.0%, P=0.262), except there was nearly no heterogeneity. The pooled HR was 1.58 (95% CI, 1.02-2.45, P<0.0001; Q=8.95, I2=66.5%, P=0.030) after excluding the study and assuming HR of 1.00.[19]If one study with the largest weights[45]was excluded, HR was 1.42 (95% CI, 1.80-2.53, P=0.046; Q=9.36, I2=68.0%, P=0.025). Similarly, the overall effect remained virtually unchanged (HR=1.54, 95% CI, 0.81-2.93, P=0.001; Q=8.71, I2=77.0%, P=0.013) when the lower quality point studies were removed.[13,19]

The forest plot in Fig. 2C displays the HRi and 95% CI for 5 studies in RR;[11-14,46]the pooled HR was 1.29 (Table 2).

All of these results indicated that DM was an important predictor for HCC patients after curative treatments.

Assessment of publication bias

Fig. 5. Forest plot of the HR of OS on subgroup analysis based on viral status. HCC patients with DM and HCV had a poor OS (pool HR=1.71, P=0.016).

Fig. 6. Forest plot of the HR of OS on subgroup analysis based on tumor diameter. HCC patients with DM had a poor OS if diameter≤5 cm (pooled HR=1.63, P<0.0001).

Fig. 7. Funnel plot of the HR of OS (A), RFS (B), and RR (C) from HCCs associated with DM status in the studies. It displays HR on a log scale against its standard error (SE) for each study included in the meta-analysis. The vertical line indicates the pooled estimate of the LnHR, with the sloping lines representing the expected 95% CI for a given SE. Under the assumption of no heterogeneity between studies, 95% of studies lie between or on these two lines.

No evidence of overt publication bias was found in these studies by using visual assessment of the funnel plot (Fig. 7A), or formal evaluation using Begg's (P=0.916) and Egger's tests (P=0.748) (Fig. 8A). Similarly, there was no significant publication bias in the RFS and RR studies (P=1.000, 0.816 and 0.734,=0.389, respectively) (Fig. 8B, C) although one study was outside the pseudo 95% CI line in the funnel plot (Fig. 7B, C).

Discussion

DM is a global problem with devastating human, social and economic impact. Its prevalence has reached epidemic proportions.[47]Growing evidence shows that DM is an independent risk for several common human malignancies, including HCC.[4]Meanwhile, HCC patients, commonly with chronic liver disease and a cirrhosis background, are predisposed to DM. As a result, identifying the effect of DM on the prognosis of HCC can guide clinical decision-making on therapy and prevention. In this study, we used meta-analysis to investigate the effect of DM as a coexisting condition on the longterm outcome of HCC patients based on 10 eligible published studies. Pooled estimates showed that DM as a concomitant disease correlated with poor prognosis and high RR in Asian HCC patients who underwent curative treatments. Many published studies have investigated possible mechanisms of the influence of DM on HCC. First, patients with DM always have insulin resistance which can cause compensatory hyperinsulinemia. It has been reported that hyperinsulinemia is associated with an increasing growth rate of HCC.[48]Meanwhile, Dellon et al[49]reported that hyperinsulinemia can promote the phosphorylation and activation of AKT and ERK pathways via interaction with the insulin receptor, which may play important roles in tumor development and progress. Furthermore, diabetics have a disorder in insulin-like growth factor-1 (IGF-1) and insulin-like growth factor binding protein-1 (IGFBP-1) levels.[50]Once they are combined, this induces a conformational change which results in autophosphorylation to convert to the active form.[51]This event triggers the initiation of multiple downstream signaling pathways, including the mitogen-activated protein kinase (MAPK) and phosphatidylinositol 3-kinase (PI3K) signaling cascades, which results in cellular proliferation and inhibition of apoptosis.[51-53]Finally, as a result of hyperglycemia, oxidative stress is a key in the pathogenesis and complications of DM.[54]Free radicals caused by oxidative stress mediate endothelial cell dysfunction, which plays an important role in the various stages of tumor progression and metastasis.[55]

Fig. 8. Detecting publication bias on OS (A), RFS (B), and RR (C) using the Egger's and Begg's funnel plot.

Further, we explored the prognostic role of DM in HCC subgroups with different virus infection backgrounds, and found that patients with DM had a shorter survival time in HCV-related HCC, while HBV-related cases were not significantly different. This may be related to different modes of pathogenesis of HBV- and HCV-related HCC. The prevalence of DM and insulin resistance is higher in patients with HCV infection than in those with HBV.[56]Chronic inflammation and oxidative stress are closely associated with the death and regeneration of hepatocytes in HCV-related HCC.[57]Patients with DM in these cohorts may have a worse progression of liver fibrosis and an increase in the development of HCC in HCV-related liver disease.[58]Moreover, a series of hepatic molecules regulating glucose metabolism are modulated by HCV infection. HCV core-induced suppressor of cytokine signaling 3 promotes proteosomal degradation of insulin receptor substrate (IRS) 1 and 2 through ubiquitination; then the disruption of IRS1 and IRS2, respectively results in insulin resistance and DM.[59]While in HBV-related liver disease, integration of the viral genome into the host DNA appears to induce HCC[60]and the coexistence of DM may have little synergistic effect on such a mechanism. In HBV carriers, chronic HBV infection seems not to be related to insulin resistance or hepatic steatosis.[61]But these results also should be interpreted with prudence considering the small number of included studies.

We also found that coexisting DM impaired OS of HCC patients with a small tumor burden, which implied that close surveillance and controlling the level of blood glucose may improve the prognosis of these patients. However, the results still need to be interpreted with caution due to the heterogeneity and poor representation of contributing studies. For RFS and RR, the prognostic role of DM in subgroups with different virus infection backgrounds and tumor diameter could not be analyzed owing to an absence of adequate data.

Many other factors, such as tumor stage, liver function and degree of liver fibrosis may affect the survival of patients with DM. Unfortunately these data were not explicitly reported in the studies. Furthermore, meta-regression should be carried out to reveal any significant associations between the most important factors (study size, year of publication, length of follow-up, type of DM, method of data presentation and extraction) with outcome. However, this was not performed owing to a small number of studies included in the analysis. Several studies noted that anti-diabetic therapies can affect the survival of HCC patients, but no abundant data were received for meta-analysis. An Italian study indicated that patients with DM, particularly males and treated with insulin, had an increased frequency of HCC.[62]In a population-based cohort study, diabetics exposed to sulfonylureas and insulin had a significantly increased risk of cancer-related mortality compared to those exposed to metformin.[63]The former anti-diabetic drugs might facilitate tumor development by increasing circulating insulin levels, but metformin as an insulinsensitizer against hyperinsulinemia may reduce the risk of cancer.[64]

Heterogeneity or publication bias between studies should be noted in assessing the validity of clinical research studies because it detects potentially important differences using a small number of primary studies analyzed in each group. No evidence of significant heterogeneity was found in OS and RFS analysis. To lower or even avoid bias, we identified all the key published analyses and used rigid inclusion criteria in selecting studies. Some potential sources of heterogeneity or bias in this meta-analysis may be attributed to therapies for DM and HCC, tumor stage, inadequate blinding of survival data from assessors, estimate of HR, ethnic background and language.

In summary, the meta-analysis results support the hypothesis that coexisting DM status is a prognostic factor for a poor outcome in HCC patients, especially those with HCV infection. However, studies that are better designed, especially homogeneous HCC cohorts analyzed prospectively, are required to unequivocally estimate the precise prognostic effect of DM coexistence in HCC patients. Moreover, the underlying mechanism of the effect of DM on the prognosis of HCC should be elucidated clearly so that an effective strategy can be identified to reduce the negative impact after operation.

Acknowledgement

We thank Nai-Qing Zhao and Tian-Shu Liu for their instruction on the statistics.

Funding:This study was supported by grants from the National Basic Research Program of China (973 Program, 2011CB504001), the National Natural Science Foundation of China (30873039, 81030038, 81071661 and 81000927), the National Key Sci-Tech Special Project of Infectious Diseases (2008ZX10002-022), and the Shanghai Rising-Star Follow-up Program Fund (10QH1400500).

Ethical approval:Not needed.

Contributors:XY, YXR and FJ proposed the study. WWM and YXR wrote the first draft. WWM, WYH and SHX analyzed the data. All authors contributed to the design and interpretation of the study and to further drafts. FJ is the guarantor.

Competing interest:No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article.

1 Parkin DM, Bray F, Ferlay J, Pisani P. Global cancer statistics, 2002. CA Cancer J Clin 2005;55:74-108.

2 Llovet JM, Beaugrand M. Hepatocellular carcinoma: present status and future prospects. J Hepatol 2003;38:S136-149.

3 Tang ZY, Ye SL, Liu YK, Qin LX, Sun HC, Ye QH, et al. A decade's studies on metastasis of hepatocellular carcinoma. J Cancer Res Clin Oncol 2004;130:187-196.

4 Strickler HD, Wylie-Rosett J, Rohan T, Hoover DR, Smoller S, Burk RD, et al. The relation of type 2 diabetes and cancer. Diabetes Technol Ther 2001;3:263-274.

5 Lai MS, Hsieh MS, Chiu YH, Chen TH. Type 2 diabetes and hepatocellular carcinoma: A cohort study in high prevalence area of hepatitis virus infection. Hepatology 2006;43:1295-1302.

6 Lagiou P, Kuper H, Stuver SO, Tzonou A, Trichopoulos D, Adami HO. Role of diabetes mellitus in the etiology of hepatocellular carcinoma. J Natl Cancer Inst 2000;92:1096-1099.

7 El-Serag HB, Tran T, Everhart JE. Diabetes increases the risk of chronic liver disease and hepatocellular carcinoma. Gastroenterology 2004;126:460-468.

8 Gao C, Yao SK. Diabetes mellitus: a "true" independent risk factor for hepatocellular carcinoma? Hepatobiliary Pancreat Dis Int 2009;8:465-473.

9 Huo TI, Hsu CY, Huang YH, Hsia CY, Lin HC, Lee PC, et al. Diabetes mellitus as an independent prognostic predictor and its association with renal dysfunction in patients with hepatocellular carcinoma. Liver Int 2010;30:198-207.

10 Amarapurkar DN, Patel ND, Kamani PM. Impact of diabetes mellitus on outcome of HCC. Ann Hepatol 2008;7:148-151.

11 Komura T, Mizukoshi E, Kita Y, Sakurai M, Takata Y, Arai K, et al. Impact of diabetes on recurrence of hepatocellular carcinoma after surgical treatment in patients with viral hepatitis. Am J Gastroenterol 2007;102:1939-1946.

12 Kawamura Y, Ikeda K, Arase Y, Yatsuji H, Sezaki H, Hosaka T, et al. Diabetes mellitus worsens the recurrence rate after potentially curative therapy in patients with hepatocellular carcinoma associated with nonviral hepatitis. J Gastroenterol Hepatol 2008;23:1739-1746.

13 Huo TI, Wu JC, Lui WY, Huang YH, Lee PC, Chiang JH, et al. Differential mechanism and prognostic impact of diabetes mellitus on patients with hepatocellular carcinoma undergoing surgical and nonsurgical treatment. Am J Gastroenterol 2004; 99:1479-1487.

14 Poon RT, Fan ST, Wong J. Does diabetes mellitus influence the perioperative outcome or long term prognosis after resection of hepatocellular carcinoma? Am J Gastroenterol 2002;97:1480-1488.

15 No authors listed. Diabetes mellitus. Report of a WHO Study Group. World Health Organ Tech Rep Ser 1985;727: 1-113.

16 Expert Committee on the Diagnosis and Classification of Diabetes Mellitus. Report of the expert committee on the diagnosis and classification of diabetes mellitus. Diabetes Care 2003;26:S5-20.

17 Schoenleber SJ, Kurtz DM, Talwalkar JA, Roberts LR, Gores GJ. Prognostic role of vascular endothelial growth factor in hepatocellular carcinoma: systematic review and metaanalysis. Br J Cancer 2009;100:1385-1392.

18 Parmar MK, Torri V, Stewart L. Extracting summary statistics to perform meta-analyses of the published literature for survival endpoints. Stat Med 1998;17:2815-2834.

19 Nagasue N, Kohno H, Tachibana M, Yamanoi A, Ohmori H, El-Assal ON. Prognostic factors after hepatic resection for hepatocellular carcinoma associated with Child-Turcotte class B and C cirrhosis. Ann Surg 1999;229:84-90.

20 Egger M, Smith GD. Bias in location and selection of studies. BMJ 1998;316:61-66.

21 Popat S, Matakidou A, Houlston RS. Thymidylate synthase expression and prognosis in colorectal cancer: a systematic review and meta-analysis. J Clin Oncol 2004;22:529-536.

22 Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. BMJ 2003;327:557-560.

23 Egger M, Davey Smith G, Schneider M, Minder C. Bias in meta-analysis detected by a simple, graphical test. BMJ 1997; 315:629-634.

24 Tierney JF, Stewart LA, Ghersi D, Burdett S, Sydes MR. Practical methods for incorporating summary time-to-event data into meta-analysis. Trials 2007;8:16.

25 John PR, Thuluvath PJ. Outcome of patients with new-onset diabetes mellitus after liver transplantation compared with those without diabetes mellitus. Liver Transpl 2002;8:708-713.

26 Moon JI, Barbeito R, Faradji RN, Gaynor JJ, Tzakis AG. Negative impact of new-onset diabetes mellitus on patient and graft survival after liver transplantation: Long-term follow up. Transplantation 2006;82:1625-1628.

27 Xu X, Ling Q, He ZL, Gao F, Zheng SS. Post-transplant diabetes mellitus in liver transplantation: Hangzhou experience. Hepatobiliary Pancreat Dis Int 2008;7:465-470.

28 Huo TI, Lui WY, Huang YH, Chau GY, Wu JC, Lee PC, et al. Diabetes mellitus is a risk factor for hepatic decompensation in patients with hepatocellular carcinoma undergoing resection: a longitudinal study. Am J Gastroenterol 2003;98: 2293-2298.

29 Shirabe K, Shimada M, Gion T, Hasegawa H, Takenaka K, Utsunomiya T, et al. Postoperative liver failure after major hepatic resection for hepatocellular carcinoma in the modern era with special reference to remnant liver volume. J Am Coll Surg 1999;188:304-309.

30 Huo TI, Wu JC, Huang YH, Chiang JH, Lee PC, Chang FY, et al. Acute renal failure after transarterial chemoembolization for hepatocellular carcinoma: a retrospective study of the incidence, risk factors, clinical course and long-term outcome. Aliment Pharmacol Ther 2004;19:999-1007.

31 Shimada M, Matsumata T, Akazawa K, Kamakura T, Itasaka H, Sugimachi K, et al. Estimation of risk of major complications after hepatic resection. Am J Surg 1994;167: 399-403.

32 Wang DD, Zhao YM, Wang L, Ren G, Wang F, Xia ZG, et al. Preoperative serum retinol-binding protein 4 is associated with the prognosis of patients with hepatocellular carcinoma after curative resection. J Cancer Res Clin Oncol 2011;137: 651-658.

33 Sumie S, Kawaguchi T, Komuta M, Kuromatsu R, Itano S, Okuda K, et al. Significance of glucose intolerance and SHIP2 expression in hepatocellular carcinoma patients with HCV infection. Oncol Rep 2007;18:545-552.

34 Imai K, Takai K, Nishigaki Y, Shimizu S, Naiki T, Hayashi H, et al. Insulin resistance raises the risk for recurrence of stageihepatocellular carcinoma after curative radiofrequency ablation in hepatitis C virus-positive patients: A prospective,case series study. Hepatol Res 2010;40:376-382.

35 Yanaga K, Matsumata T, Hayashi H, Shimada M, Urata K, Suehiro T, et al. Effect of diabetes mellitus on hepatic resection. Arch Surg 1993;128:445-448.

36 Shimada M, Takenaka K, Fujiwara Y, Gion T, Shirabe K, Yanaga K, et al. Risk factors linked to postoperative morbidity in patients with hepatocellular carcinoma. Br J Surg 1998;85:195-198.

37 Khan MM, Saito S, Takagi S, Ohnishi H, Izumi H, Sakauchi F, et al. Relationship between hepatocellular carcinoma and impaired glucose tolerance among Japanese. Hepatogastroenterology 2006;53:742-746.

38 Martin JH, Coory MD, Valery PC, Green AC. Association of diabetes with survival among cohorts of Indigenous and non-Indigenous Australians with cancer. Cancer Causes Control 2009;20:355-360.

39 Takamatsu S, Noguchi N, Kudoh A, Nakamura N, Kawamura T, Teramoto K, et al. Influence of risk factors for metabolic syndrome and non-alcoholic fatty liver disease on the progression and prognosis of hepatocellular carcinoma. Hepatogastroenterology 2008;55:609-614.

40 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.

41 Coughlin SS, Calle EE, Teras LR, Petrelli J, Thun MJ. Diabetes mellitus as a predictor of cancer mortality in a large cohort of US adults. Am J Epidemiol 2004;159:1160-1167.

42 Matsumata T, Taketomi A, Kawahara N, Higashi H, Shirabe K, Takenaka K. Morbidity and mortality after hepatic resection in the modern era. Hepatogastroenterology 1995;42:456-460.

43 Shimada M, Hasegawa H, Gion T, Shirabe K, Taguchi K, Takenaka K, et al. Risk factors of the recurrence of hepatocellular carcinoma originating from residual cancer cells after hepatectomy. Hepatogastroenterology 1999;46: 2469-2475.

44 Huo TI, Wu JC, Lui WY, Lee PC, Huang YH, Chau GY, et al. Diabetes mellitus is a recurrence-independent risk factor in patients with hepatitis B virus-related hepatocellular carcinoma undergoing resection. Eur J Gastroenterol Hepatol 2003;15:1203-1208.

45 Ikeda Y, Shimada M, Hasegawa H, Gion T, Kajiyama K, Shirabe K, et al. Prognosis of hepatocellular carcinoma with diabetes mellitus after hepatic resection. Hepatology 1998;27: 1567-1571.

46 Toyoda H, Kumada T, Nakano S, Takeda I, Sugiyama K, Kiriyama S, et al. Impact of diabetes mellitus on the prognosis of patients with hepatocellular carcinoma. Cancer 2001;91:957-963.

47 Wild S, Roglic G, Green A, Sicree R, King H. Global prevalence of diabetes: estimates for the year 2000 and projections for 2030. Diabetes Care 2004;27:1047-1053.

48 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.

49 Dellon ES, Shaheen NJ. Diabetes and hepatocellular carcinoma: associations, biologic plausibility, and clinical implications. Gastroenterology 2005;129:1132-1134.

50 Akturk M, Arslan M, Altinova A, Ozdemir A, Ersoy R, Yetkin I, et al. Association of serum levels of IGF-I and IGFBP-1 with renal function in patients with type 2 diabetes mellitus. Growth Horm IGF Res 2007;17:186-193.

51 Lee HC, Tian B, Sedivy JM, Wands JR, Kim M. Loss of Raf kinase inhibitor protein promotes cell proliferation and migration of human hepatoma cells. Gastroenterology 2006; 131:1208-1217.

52 Alexia C, Fallot G, Lasfer M, Schweizer-Groyer G, Groyer A. An evaluation of the role of insulin-like growth factors (IGF) and of type-I IGF receptor signalling in hepatocarcinogenesis and in the resistance of hepatocarcinoma cells against druginduced apoptosis. Biochem Pharmacol 2004;68:1003-1015.

53 Radcliff K, Tang TB, Lim J, Zhang Z, Abedin M, Demer LL, et al. Insulin-like growth factor-I regulates proliferation and osteoblastic differentiation of calcifying vascular cells via extracellular signal-regulated protein kinase and phosphatidylinositol 3-kinase pathways. Circ Res 2005;96: 398-400.

54 Rosen P, Nawroth PP, King G, Moller W, Tritschler HJ, Packer L. The role of oxidative stress in the onset and progression of diabetes and its complications: a summary of a Congress Series sponsored by UNESCO-MCBN, the American Diabetes Association and the German Diabetes Society. Diabetes Metab Res Rev 2001;17:189-212.

55 Tesfamariam B, Cohen RA. Free radicals mediate endothelial cell dysfunction caused by elevated glucose. Am J Physiol 1992;263:H321-326.

56 Imazeki F, Yokosuka O, Fukai K, Kanda T, Kojima H, Saisho H. Prevalence of diabetes mellitus and insulin resistance in patients with chronic hepatitis C: comparison with hepatitis B virus-infected and hepatitis C virus-cleared patients. Liver Int 2008;28:355-362.

57 Lauer GM, Walker BD. Hepatitis C virus infection. N Engl J Med 2001;345:41-52.

58 Fong DG, Nehra V, Lindor KD, Buchman AL. Metabolic and nutritional considerations in nonalcoholic fatty liver. Hepatology 2000;32:3-10.

59 Kawaguchi T, Yoshida T, Harada M, Hisamoto T, Nagao Y, Ide T, et al. Hepatitis C virus down-regulates insulin receptor substrates 1 and 2 through up-regulation of suppressor of cytokine signaling 3. Am J Pathol 2004;165:1499-1508.

60 Brechot C, Pourcel C, Louise A, Rain B, Tiollais P. Presence of integrated hepatitis B virus DNA sequences in cellular DNA of human hepatocellular carcinoma. Nature 1980;286: 533-535.

61 Wang CC, Hsu CS, Liu CJ, Kao JH, Chen DS. Association of chronic hepatitis B virus infection with insulin resistance and hepatic steatosis. J Gastroenterol Hepatol 2008;23:779-782.

62 Donadon V, Balbi M, Casarin P, Vario A, Alberti A. Association between hepatocellular carcinoma and type 2 diabetes mellitus in Italy: potential role of insulin. World J Gastroenterol 2008;14:5695-5700.

63 Bowker SL, Majumdar SR, Veugelers P, Johnson JA. Increased cancer-related mortality for patients with type 2 diabetes who use sulfonylureas or insulin. Diabetes Care 2006;29:254-258.

64 Evans JM, Donnelly LA, Emslie-Smith AM, Alessi DR, Morris AD. Metformin and reduced risk of cancer in diabetic patients. BMJ 2005;330:1304-1305.

Received November 6, 2010

Accepted after revision March 4, 2011

Author Affiliations: Institute of Biomedical Sciences, Fudan University, Shanghai 200032, China (Wang WM and Sun HX); Key Laboratory of Carcinogenesis and Cancer Invasion, Ministry of Education, and Liver Cancer Institute, Zhongshan Hospital, Fudan University, Shanghai 200032, China (Xu Y, Yang XR and Fan J); Fudan University Shanghai Cancer Center, Shanghai 200032, China (Wang YH)

Jia Fan, MD, PhD, Liver Cancer Institute, Zhongshan Hospital, Fudan University, Shanghai 200032, China (Tel: 86-21-64037181; Email: jiafan99@yahoo.com)

? 2011, Hepatobiliary Pancreat Dis Int. All rights reserved.