Diabetes and hepatocellular cancer risk: Not only a matter of hyperglycemia


  • Potential conflict of interest: Dr. Marchesini advises Sanofi-Aventis and Roche. Dr. Forlani advises MSD and is on the speakers' bureau of Roche.

Wang P, Kang D, Cao W, Wang Y, Liu Z. Diabetes mellitus and risk of hepatocellular carcinoma: a systematic review and meta-analysis. Diabetes Metab Res Rev 2012;28:109-122. (Reprinted with permission).


Studies of diabetes and hepatocellular carcinoma (HCC) yielded inconsistent findings. This meta-analysis was conducted to examine the association between diabetes and risk of HCC. Studies were identified by searching PUBMED and MEDLINE database up to February 2011. Pooled risk estimates were calculated using the random-effects model. Potential sources of heterogeneity were explored by subgroup analyses. A total of 17 case-control studies and 32 cohort studies were included in the meta-analysis. The combined risk estimate of all studies showed a statistically significant increased risk of HCC prevalence among diabetic individuals (RR = 2.31, 95% CI: 1.87–2.84). The pooled risk estimate of 17 case-control studies (OR = 2.40, 95% CI: 1.85–3.11) was slightly higher than that from 25 cohort studies (RR = 2.23, 95% CI: 1.68–2.96). Metformin treatment was potentially protective. On the contrary, long duration of diabetes and sulfonylureas or insulin treatment could increase HCC risk. Also meta-analysis of 7 cohort studies found a statistically significant increased risk of HCC mortality (RR = 2.43, 95% CI: 1.66–3.55) for individuals with (versus without) diabetes. This meta-analysis shows that diabetes is associated with moderately increased risk of HCC incidence, as well as HCC mortality. Considering the rapidly increasing prevalence of diabetes mellitus, the study underlines the need for cancer prevention in diabetic individuals. Further investigation is needed to focus on the potential mechanism for the pathogenesis of HCC and the link between HCC and different types, severity, treatment and duration of diabetes.


On June 26, 2009, an earthquake shook the world of diabetes research. A series of papers published in Diabetologia, the official journal of the European Association for the Study of Diabetes, raised doubts on a possible link between insulin glargine and cancer. Glargine is the most widely prescribed insulin in the world, accounting for 37.5% of prescriptions in the United States. On the whole, the articles were inconclusive, and the accompanying editorial, explaining how insulin could increase cancer risk, ended up by urging further research.1 Since then, several data have refuted a specific role of glargine but confirmed a general and specific risk of cancer associated with diabetes and the role of drugs and metabolic factors.

The issue of diabetes and cancer was not new in 2009; it was initially raised by the 24-year follow-up of a very large Swedish diabetes cohort.2 Compared with nationwide rates, the standardized incidence ratio of primary liver cancer was 2.5 (95% confidence interval [CI], 2.3-2.6), and much higher for hepatocellular carcinoma (HCC) (incidence ratio, 4.1; 95% CI, 3.8-4.5). It was only moderately reduced after exclusion of patients with concomitant diseases favoring liver cancer, such as alcoholism, cirrhosis, and hepatitis.2 The authors discussed the possibility that the increased risk might be mediated by factors promoting cell proliferation (insulin, insulin-like growth factor-1) or by conditions associated with diabetes (alcohol, hemochromatosis, viral hepatitis, liver cell failure, fatty liver). In advanced cirrhosis, the so-called hepatogenous diabetes might indeed justify an excess of primary liver cancers,2 inflating the overall risk.

Obesity, metabolic syndrome, and fatty liver remain backstage of the association between diabetes and HCC. In a comprehensive review, Calle and Kaaks3 concluded for a relative risk of liver cancer associated with obesity between 1.5 and 4.0, but it was not possible to derive the population-attributable risk. In a very large adult population, the relative risk (RR) of mortality for primary liver cancer increased with increasing body mass index (BMI) up to 4.52 (95% CI, 2.94-6.94) in men and 1.68 (0.93-3.95) in women for a BMI ≥35 kg/m2.4

Excess body fat is also associated with hepatic triglyceride accumulation and nonalcoholic fatty liver disease (NAFLD). NAFLD is common in type 2 diabetes; over the years, fatty liver may be complicated by necroinflammation and fibrosis and progress to nonalcoholic steatohepatitis (NASH), bridging fibrosis, and cirrhosis, where HCC is more likely to develop.5

According to all these data, the results of Wang et al.6 are no surprise. They reviewed all the studies published up to February 2011 and identified 17 case-control studies and 32 cohort studies out of 1,891 articles fitting predefined criteria. The pooled meta-analysis identified an increased RR of primary liver cancer incidence among diabetic individuals (RR, 2.31; 95% CI, 1.87-2.84), similar between pooled case-control and cohort studies. The risk was maintained irrespective of geographic region, sex, presence of associated risk factors, and type of diabetes. Notably, the risk was cancelled by metformin (odds ratio [OR], 0.3; 95% CI, 0.2-0.6) and thiazolidinedione (OR, 0.3; 95% CI, 0.1-0.7) treatment, and increased in patients receiving sulfonylureas or insulin (OR, 3.1; 95% CI, 1.6-5.7). Finally, the meta-analysis restricted to seven cohort studies confirmed an increased risk of mortality (RR = 2.43) in diabetes. The authors concluded that a strict surveillance for liver cancer prevention is needed in diabetic subjects.

The Wang et al. study has its strengths; it is a large study, and the data are consistent with those in other reports. In an analysis of all persons diagnosed with HCC and intrahepatic cholangiocarcinoma (ICC) between 1993 and 2005 in the Surveillance, Epidemiology, and End Results (SEER)-Medicare database, Welzel et al.7 identified 3,649 HCC and 743 ICC cases to be compared with 195,953 persons living in the same areas. The metabolic syndrome was significantly more common among persons with HCC (37.1%) and ICC (29.7%) than in the comparison group (17.1%). In multiple logistic regression analyses, adjusted for confounders, the metabolic syndrome remained significantly associated with HCC (OR, 2.13; 95% CI, 1.96-2.31) and, less significantly, ICC (OR, 1.56; 95% CI, 1.32-1.83). Notably, impaired fasting glucose/diabetes was the feature most closely associated with HCC (OR, 2.90; 95% CI, 2.71-3.10).

Many therapeutic options have been approved for diabetes, and many more drugs might be soon available, acting at different levels and with different safety profiles (Table 1); they might also modulate cancer risk. Metformin remains the first-line treatment in type 2 diabetes; its use has been associated with the increased reduced liver volume, reduced liver fat, normalization of liver enzymes in NAFLD/NASH,8 stable or reduced body weight, and reduced overall cancer risk,9 as confirmed by Wang et al.'s meta-analysis. When metformin fails, add-on therapy with either insulin secretagogues (sulfonylureas or glinides) or insulin stimulates the mitogen-activated protein kinase pathway favoring mitogenesis and cell proliferation, in keeping with the increased HCC risk.6 Thiazolidinediones are a valid alternative. They promote insulin sensitivity without increasing insulin levels, improve liver histology (decreased liver fat and necroinflamation, stable or improved fibrosis),10 and are now considered the best option for treating NAFLD/NASH. Reduced HCC risk is in keeping with their role in NAFLD, but their use may be limited by cardiovascular risk (rosiglitazone)11 and an unexpected risk of bladder cancer (pioglitazone).12

Table 1. Effects of Drug Treatment in Type 2 Diabetes
Drug ClassHbA1c DecreaseHypoglycemiaWeight ChangeCardiovascular RiskContraindicationsLiver FatCancer Risk
  • Metformin was used as a comparator in the evaluation of other drugs. Note that the different classes also have a highly variable safety profile on hypoglycemia and cardiovascular risk.

  • Abbreviations: DPP-4, dipeptidyl peptidase-4; GLP-1, glucagon-like peptide-1; HbA1c, glycosylated hemoglobin; ND, no data; SGLT-2, sodium-glucose transporter-2. Symbols: =, no change; upward arrow, increased; downward arrow, reduced.

  • *

    Increased cardiovascular risk, in relation to insulin dose.

  • Increased cancer risk in observational studies, in relation to increased circulating insulin concentrations, insulin resistance, or insulin dose.

  • Increased cardiovascular risk in cohort studies.

  • §

    Increased cardiovascular risk associated with rosiglitazone use, unchanged or reduced cardiovascular risk with pioglitazone.

  • ||

    Warnings for bladder cancer with pioglitazone.

  • No data; considering their mechanism of action, in principle the risk is expected to be similar to sulfonylureas.

  • #

    See text; no additional risk of primary liver cancer, but alerts for cancers at different sites.

  • **

    Dose to be reduced with progressive renal failure.

Metformin≈1.0=↓Renal failure
Insulin, long-acting1.0-1.5↑↑↑↑*=
Insulin, rapid-acting1.5-2.0↑↑↑↑*=
Sulfonylureas1.0-1.5Renal failure=
Thiazolinediones0.5-1.4↑↑§Heart failure; fracture risk↓↓=↓||
α-Glucosidase inhibitors0.5-0.8=↓NDND
GLP-1 agonists0.5-1.0↓↓NDRenal failure#
DPP-4 inhibitors0.6-0.8=ND**ND#
SGLT-2 inhibitors0.5-0.8=↓NDNDND#

What else? The new classes of glucagon-like peptide-1 (GLP-1) agonists (incretins) or dipeptidyl peptidase-4 (DPP-4) inhibitors (gliptins) are a valid alternative. Incretins mimic GLP-1, favoring insulin release in response to glucose, whereas gliptins prevent GLP-1 degradation. They do not cause hypoglycemia, and phase II-III studies indicate a good safety profile on cardiovascular risk. GLP-1 agonists (exenatide, liraglutide, but soon also lixisenatide, dulaglutide, taspoglutide, and albiglutide) reduce body weight and liver fat;13 DPP-4 inhibitors (sitagliptin, vildagliptin, saxagliptin, but soon linagliptin and alogliptin) are weight-neutral, but have nonetheless been associated with reduced liver fat accumulation.14 In addition, the new class of sodium-glucose transporter-2 inhibitors (SGLT-2, also called glifozins) has neutral or favorable effects on body weight. For a few drugs, alerts have been issued regarding specific cancer risks (pancreatic cancer for exenatide, thyroid medullary carcinoma for liraglutide, breast cancer for dapaglifozim), but there is no solid evidence. What should we expect regarding HCC risk? The obesity and diabetes epidemics are driving the NAFLD epidemic. Although the overall risk of HCC in NAFLD is lower than in liver disease of different etiology, the very high number of cases will constitute a worldwide challenge for National Health Systems. The diabetes-associated progression of NASH/cryptogenic cirrhosis to HCC is slow, but effective treatments are limited by comorbidities, which also reduce the inclusion in liver transplant lists.15 New diabetes treatments might help decrease the future burden of diabetes-associated HCC.