The possibility that diabetes mellitus, a common condition in many populations, may increase the risk of hepatocellular carcinoma (HCC) has been raised in a number of cohort and case–control studies (112). In particular, this hypothesis received strong support from two large population-based cohort studies, one undertaken in Sweden (13) and the other in Denmark (14). Although most of the studies were methodologically adequate, few directly adjusted for tobacco smoking and alcohol intake; moreover, in none, to our knowledge, were the results specifically controlled for serologic indication of chronic infection with hepatitis B virus (HBV) and hepatitis C virus (HCV).

We have completed in Athens, Greece, a large case–control study of HCC in which information on medical history and life-style variables has been collected and chronic infection with HBV and/or HCV has been assessed. Details of the study were published previously (15,16). Briefly, during the period from January 1995 through December 1998, a total of 374 patients with incident HCC were admitted to one of three teaching hospitals in Athens; 333 (89%) of these patients with confirmed disease participated in the study. A total of 385 control subjects were identified from among patients who were hospitalized for eye, ear, nose, or throat conditions or for injuries and who had no history of cancer and no liver damage. Twenty-five (6%) of these patients refused to participate, which left 360 control subjects who were included in the analysis. The control subjects were generally matched with the case subjects by sex and age (±5 years). The mean age was the same (64.0 years) in both the HCC case subjects and the control subjects; 50 (15.0%) of the case subjects and 62 (17.2%) of the control subjects were female. The study protocol was approved by the University of Athens Medical School Committee of Ethics and Research.

All case and control subjects were interviewed in the hospital by use of a standardized questionnaire. A diagnosis of diabetes mellitus at least 1 year before the diagnosis of the current disease among case and control subjects was confirmed by examination of their respective medical records. Sera were obtained from each subject and were transferred on dry ice to the internationally certified BIOMED Laboratories in Athens for serologic testing for HBV and HCV markers (15). Randomly chosen coded sera from 97 HCC case subjects and 130 hospital control subjects were tested in the same laboratory for insulin-like growth factor-I (IGF-I) and its binding protein 3 (IGFBP-3). IGF-I was measured with the use of a chemiluminescence assay, and IGFBP-3 was measured with the use of a radioimmunoassay. Assays were done at the BIOMED Laboratories using assays produced by the Nichols Institute, San Juan Capistrano, CA. As part of their clinical management, 85 of the HCC case subjects also were tested for serum albumin levels and prothrombin time in the hospital laboratories.

An indicator of functional liver damage was constructed among case subjects by use of evidence of cirrhosis determined from available pathology slides or from medical records and also from prothrombin time (international normalized ratio [INR] units) and serum albumin levels (g/dL). We thought that a composite score was a more reliable indicator of liver damage, since a biopsy specimen was not available for all HCC subjects. This indicator ranged from 0 (no evidence of functional liver damage) to 3 (evidence of substantial functional liver damage). Thus, a patient with evidence of cirrhosis, whose albumin levels were less than or equal to 4.0 g/dL, and with a prothrombin time of 1.5 or more INR units was given a functional liver damage indicator of 3.

Statistical analysis was done by use of multivariate unconditional logistic regression models, in which HCC status was the dependent variable and the matching factors age (categorical) and sex were controlled for. The following potentially confounding variables were also adjusted for in the model: years of schooling (≥12 years versus <12 years), hepatitis viral status (infection with either or both viruses versus no infection), smoking status (ever versus never), and alcohol consumption (≥40 glasses per week versus <40 glasses per week). In the logistic regression analysis, the odds ratio (OR) provides an estimate of the association between HCC and the variable of interest; the 95% confidence interval (CI) around the OR indicates whether the association is statistically significant at the .05 level. To compare the means of continuous variables between two groups, we calculated mean differences and associated 95% CIs on the basis of the two-sided t distribution.

Table 1 shows the distribution of HCC case and control subjects by history of diabetes mellitus and by the potential confounders. As reported previously (15,16), chronic infection with HBV or HCV appears to substantially increase the risk of HCC, whereas both smoking and high intake of alcohol only modestly increase the risk. The OR linking history of diabetes mellitus with HCC adjusted only for age and sex (OR = 1.83; 95% CI = 1.18–2.84) was not much different from that adjusted for all potential confounders (OR = 1.86; 95% CI = 0.99–3.51). Adjustment for age and sex and for one additional confounder had only a marginal effect on the OR estimate (data not shown).

Table 2 shows the mean IGF-I and IGFBP-3 levels among HCC case and control subjects, as well as the mean functional liver damage score among HCC case subjects by history of diabetes mellitus. Both IGF-I and IGFBP-3 levels were strikingly lower among HCC case subjects than among control subjects (mean IGF-I difference = 104.85 ng/mL [95% CI = 90.25–119.46]; mean IGFBP-3 difference = 1.56 μg/mL [95% CI = 1.36–1.76]). HCC case subjects with diabetes had lower levels of both IGF-I (P = .27) and IGFBP-3 (P = .02) in comparison to HCC case subjects without diabetes. The mean functional liver damage was higher in HCC case subjects with diabetes than in those without (P = .02).

This study is relatively large, and no obvious major bias is likely to have been introduced into the design and execution of the study because results concerning HBV, HCV, alcohol consumption, and tobacco smoking (15,16) are in line with those previously reported (7,17,18). In addition, all established confounding factors were adequately controlled for. Selection bias related to the use of hospital control subjects is a theoretical possibility, but it would be unlikely, since the disease under consideration, diabetes mellitus, was not part of the admission diagnosis in our investigation. Moreover, the median time between diagnosis of diabetes mellitus and the current disease was higher among the HCC case subjects than among the control subjects, which suggests that the association between diabetes mellitus and HCC occurrence also is not likely the result of selection bias related to the diagnosis of diabetes. It is remotely possible that the use of hospital control subjects could have led to an underestimation of the association between HCC and diabetes mellitus if, for whatever reason, the control subjects had a higher prevalence of diabetes than the source population giving rise to the case subjects. The opposite, however, seems improbable because HCC case subjects are bound to be hospitalized, irrespective of the coexistence of any other disease. Although the pathophysiologic characteristics of diabetes mellitus were not elaborated, type II diabetes is substantially more common than type I diabetes in Greece, as in the rest of Europe (19). There was no difference between diabetic and nondiabetic HCC case subjects with respect to whether information on cirrhosis was from pathology slides or from medical records or whether prothrombin time and serum albumin levels were measured.

Our study indicates that there is a positive association between a history of diabetes mellitus and the occurrence of HCC and that this relationship is not confounded by any of the known major risk factors for HCC. A number of possible mechanisms might explain this association (13). First, since most case subjects with diabetes are non-insulin dependent and are characterized by hyperinsulinemia, insulin or its precursors may interact with liver cells to stimulate mitogenesis or carcinogenesis (2023). The substantial reduction in IGF-I and IGFBP-3 levels among diabetic, as compared with nondiabetic, HCC case subjects in our study may reflect an intimate link between pancreatic and hepatocellular processes, the nature of which are poorly understood. Alternatively, the metabolic effects of diabetes may increase the risk of HCC through nonalcoholic steatohepatitis and cryptogenic cirrhosis (2426). An observation that loss of heterozygosity of p53 is statistically significantly more common among HCC patients with diabetes than among those without also provides evidence for a possible molecular mechanism for this association (27).

Lastly, the long pathophysiologic process of liver disease may increase insulin resistance in susceptible individuals, since patients with liver disease are known to have an increased prevalence of glucose intolerance (3,28,29). It is possible that mutations in hepatocyte nuclear factors cause diabetes though impairment of pancreatic beta-cell function (30). This hypothesis is supported by our observation that functional liver damage is higher among HCC case subjects with diabetes mellitus than among those without a history of this disease.

Table 1.

Association of hepatocellular carcinoma (HCC) with history of diabetes mellitus and other risk factors in a case–control study in Athens, Greece*

 HCC case subjects, No. (%) (n = 333) Control subjects, No. (%) (n = 360) OR† (95% CI) Multivariate adjusted OR‡ (95% CI) 
*OR = odds ratio; CI = confidence interval. 
†Adjusted for age and sex only. 
‡Adjusted for age, sex, history of diabetes mellitus, years of schooling, hepatitis virus infection, smoking status, and alcohol consumption. 
§There are three case subjects and one control subject without information on history of schooling. 
∥There are eight control subjects with missing data on hepatitis virus infection. 
History of diabetes mellitus     
 No 273 (82.0) 321 (89.2) 1.00 (referent) 1.00 (referent) 
 Yes 60 (18.0) 39 (10.8) 1.83 (1.18–2.84) 1.86 (0.99–3.51) 
Schooling, y§     
 <12 251 (76.1) 262 (73.0) 1.00 (referent) 1.00 (referent) 
 ≥12 79 (23.9) 97 (27.0) 0.84 (0.59–1.19) 0.96 (0.55–1.67) 
Hepatitis B or C virus infection∥     
 Neither 83 (24.9) 13 (3.7) 1.00 (referent) 1.00 (referent) 
 Either or both 250 (75.1) 339 (96.3) 98.6 (51.9–187.3) 110.9 (57.2–215.0) 
Smoking status     
 Never smoker 98 (29.4) 126 (35.0) 1.00 (referent) 1.00 (referent) 
 Ever smoker 235 (70.6) 234 (65.0) 1.30 (0.90–1.85) 1.64 (0.91–2.95) 
Alcohol consumption, glasses/wk     
 <40 252 (75.7) 304 (84.4) 1.00 (referent) 1.00 (referent) 
 ≥40 81 (24.3) 56 (15.6) 1.72 (1.17–2.54) 2.40 (1.33–4.30)  
 HCC case subjects, No. (%) (n = 333) Control subjects, No. (%) (n = 360) OR† (95% CI) Multivariate adjusted OR‡ (95% CI) 
*OR = odds ratio; CI = confidence interval. 
†Adjusted for age and sex only. 
‡Adjusted for age, sex, history of diabetes mellitus, years of schooling, hepatitis virus infection, smoking status, and alcohol consumption. 
§There are three case subjects and one control subject without information on history of schooling. 
∥There are eight control subjects with missing data on hepatitis virus infection. 
History of diabetes mellitus     
 No 273 (82.0) 321 (89.2) 1.00 (referent) 1.00 (referent) 
 Yes 60 (18.0) 39 (10.8) 1.83 (1.18–2.84) 1.86 (0.99–3.51) 
Schooling, y§     
 <12 251 (76.1) 262 (73.0) 1.00 (referent) 1.00 (referent) 
 ≥12 79 (23.9) 97 (27.0) 0.84 (0.59–1.19) 0.96 (0.55–1.67) 
Hepatitis B or C virus infection∥     
 Neither 83 (24.9) 13 (3.7) 1.00 (referent) 1.00 (referent) 
 Either or both 250 (75.1) 339 (96.3) 98.6 (51.9–187.3) 110.9 (57.2–215.0) 
Smoking status     
 Never smoker 98 (29.4) 126 (35.0) 1.00 (referent) 1.00 (referent) 
 Ever smoker 235 (70.6) 234 (65.0) 1.30 (0.90–1.85) 1.64 (0.91–2.95) 
Alcohol consumption, glasses/wk     
 <40 252 (75.7) 304 (84.4) 1.00 (referent) 1.00 (referent) 
 ≥40 81 (24.3) 56 (15.6) 1.72 (1.17–2.54) 2.40 (1.33–4.30)  
Table 2.

Mean insulin-like growth factor-I (IGF-I) and insulin-like growth factor-binding protein 3 (IGFBP-3) levels among hepatocellular carcinoma (HCC) case subjects and control subjects and mean functional liver damage score among HCC case subjects by history of diabetes mellitus (DM)

 HCC case subjects Control subjects 
 History of DM No DM History of DM No DM 
*CI = confidence interval. 
†Based on equal variances. 
‡Based on unequal variances. 
No. of subjects 14 83 20 110 
IGF-I, ng/mL     
 Mean 54.14 70.02 174.30 172.27 
 Mean difference (95% CI)* 15.88 (−12.60 to 44.37)†  2.03 (−18.99 to 23.05)‡  
IGFBP-3, μg/mL     
 Mean 0.63 0.85 2.33 2.38 
 Mean difference (95% CI) 0.22 (0.04 to 0.41)‡  0.06 (−0.42 to 0.53)†  
No. of subjects 60 273   
Functional liver damage score     
 Mean 1.52 1.30 — — 
 Mean difference (95% CI) 0.21 (0.04 to 0.38)†  —  
 HCC case subjects Control subjects 
 History of DM No DM History of DM No DM 
*CI = confidence interval. 
†Based on equal variances. 
‡Based on unequal variances. 
No. of subjects 14 83 20 110 
IGF-I, ng/mL     
 Mean 54.14 70.02 174.30 172.27 
 Mean difference (95% CI)* 15.88 (−12.60 to 44.37)†  2.03 (−18.99 to 23.05)‡  
IGFBP-3, μg/mL     
 Mean 0.63 0.85 2.33 2.38 
 Mean difference (95% CI) 0.22 (0.04 to 0.41)‡  0.06 (−0.42 to 0.53)†  
No. of subjects 60 273   
Functional liver damage score     
 Mean 1.52 1.30 — — 
 Mean difference (95% CI) 0.21 (0.04 to 0.38)†  —  
Supported by a grant from the Europe Against Cancer Program of the European Union.
BIOMED Laboratories (Athens, Greece) provided scientific and technical advice and undertook all analyses free of charge.

References

1
Kessler II. Cancer mortality among diabetics.
J Natl Cancer Inst
 
1970
;
44
:
673
–86.
2
Ragozzino M, Melton LJ 3d, Chu CP, Palumbo PJ. Subsequent cancer risk in the incidence cohort of Rochester, Minnesota, residents with diabetes mellitus.
J Chronic Dis
 
1982
;
35
:
13
–9.
3
Kingston ME, Ali MA, Atiyeh M, Donnelly RJ. Diabetes mellitus in chronic active hepatitis and cirrhosis.
Gastroenterology
 
1984
;
87
:
688
–94.
4
Lawson DH, Gray JM, McKillop C, Clarke J, Lee FD, Patrick RS. Diabetes mellitus and primary hepatocellular carcinoma.
Q J Med
 
1986
;
61
:
945
–55.
5
Lu SN, Lin TM, Chen CJ, Chen JS, Liaw YF, Chang WY, et al. A case–control study of primary hepatocellular carcinoma in Taiwan.
Cancer
 
1988
;
62
:
2051
–5.
6
La Vecchia C, Negri E, D'Avanzo B, Boyle P, Franceschi S. Medical history and primary liver cancer.
Cancer Res
 
1990
;
50
:
6274
–7.
7
Yu MC, Tong MJ, Govindarajan S, Henderson BE. Nonviral risk factors for hepatocellular carcinoma in a low-risk population, the non-Asians of Los Angeles County, California.
J Natl Cancer Inst
 
1991
;
83
:
1820
–6.
8
La Vecchia C, Negri E, Franceschi S, D'Avanzo B, Boyle P. A case–control study of diabetes mellitus and cancer risk.
Br J Cancer
 
1994
;
70
:
950
–3.
9
Hadziyannis S, Tabor E, Kaklamani E, Tzonou A, Stuver S, Tassopoulos N, et al. A case–control study of hepatitis B and C virus infections in the etiology of hepatocellular carcinoma.
Int J Cancer
 
1995
;
60
:
627
–31.
10
La Vecchia C, Negri E, Decarli A, Franceschi S. Diabetes mellitus and the risk of primary liver cancer.
Int J Cancer
 
1997
;
73
:
204
–7.
11
Braga C, La Vecchia C, Negri E, Franceschi S. Attributable risks for hepatocellular carcinoma in northern Italy.
Eur J Cancer
 
1997
;
33
:
629
–34.
12
Shibata A, Fukuda K, Nishiyori A, Ogimoto I, Sakata R, Tanikawa K. A case–control study on male hepatocellular carcinoma based on hospital and community controls.
J Epidemiol
 
1998
;
8
:
1
–5.
13
Adami HO, Chow WH, Nyren O, Berne C, Linet MS, Ekbom A, et al. Excess risk of primary liver cancer in patients with diabetes mellitus.
J Natl Cancer Inst
 
1996
;
88
:
1472
–7.
14
Wideroff L, Gridley G, Mellemkjaer L, Chow WH, Linet M, Keehn S, et al. Cancer incidence in a population-based cohort of patients hospitalized with diabetes mellitus in Denmark.
J Natl Cancer Inst
 
1997
;
89
:
1360
–5.
15
Kuper H, Tzonou A, Kaklamani E, Hadziyannis S, Tasopoulos N, Lagiou P, et al. Hepatitis B and C viruses in the etiology of hepatocellular carcinoma; a study in Greece using third-generation assays.
Cancer Causes Control
 
2000
;
11
:
171
–5.
16
Kuper H, Tzonou A, Kaklamani E, Hsieh CC, Lagiou P, Adami HO, et al. Tobacco smoking, alcohol consumption and their interaction in the causation of hepatocellular carcinoma.
Int J Cancer
 
2000
;
85
:
498
–502.
17
International Agency for Research on Cancer (IARC). Hepatitis viruses. Vol 59. IARC Monogr Eval Carcinog Risks Hum. Lyon (France): IARC; 1994.
18
International Agency for Research on Cancer (IARC). Alcohol drinking. Vol 44. IARC Monogr Eval Carcinog Risks Hum. Lyon (France): IARC; 1988.
19
Raptis S. Internal medicine [in Greek] (ISBN 960-340-080-7). Athens (Greece): Parisianos; 1998. p. 25–35.
20
Flyvbjerg A, Orskov H. Growth hormone, insulin-like growth factors and diabetes. In: Alberti KG, Krall LP, editors. The diabetes annual 5. Amsterdam (The Netherlands): Elsevier Science Publishers; 1990. p. 642–56.
21
Kahn CR, White MF. Molecular mechanism of insulin action. In: DeGroot LJ, editor. Endocrinology. Philadelphia (PA): Saunders; 1995. p. 1373–88.
22
Tabor E. Tumor suppressor genes, growth factor genes, and oncogenes in hepatitis B virus-associated hepatocellular carcinoma.
J Med Virol
 
1994
;
42
:
357
–65.
23
Moore MA, Park CB, Tsuda H. Implications of the hyperinsulinaemia–diabetes–cancer link for preventive efforts.
Eur J Cancer Prev
 
1998
;
7
:
89
–107.
24
Silverman JF, O'Brien KF, Long S, Leggett N, Khazanie PG, Pories WJ, et al. Liver pathology in morbidly obese patients with and without diabetes.
Am J Gastroenterol
 
1990
;
85
:
1349
–55.
25
Caldwell SH, Oelsner DH, Iezzoni JC, Hespenheide EE, Battle EH, Driscoll CJ. Cryptogenic cirrhosis: clinical characterization and risk factors for underlying disease.
Hepatology
 
1999
;
29
:
664
–9.
26
Marceau P, Biron S, Hould FS, Marceau S, Simard S, Thung SN, et al. Liver pathology and the metabolic syndrome X in severe obesity.
J Clin Endocrinol Metab
 
1999
;
84
:
1513
–7.
27
Hsu HC, Peng SY, Lai PL, Sheu JC, Chen DS, Lin LI, et al. Allelotype and loss of heterozygosity of p53 in primary and recurrent hepatocellular carcinomas. A study of 150 patients.
Cancer
 
1994
;
73
:
42
–7.
28
Mason AL, Lau JY, Hoang N, Qian K, Alexander GJ, Xu L, et al. Association of diabetes mellitus and chronic hepatitis C virus infection.
Hepatology
 
1999
;
29
:
328
–33.
29
Allison ME, Wreghitt T, Palmer CR, Alexander GJ. Evidence for a link between hepatitis C virus infection and diabetes mellitus in a cirrhotic population.
J Hepatol
 
1994
;
21
:
1135
–9.
30
Iwasaki N, Ogata M, Tomonaga O, Kuroki H, Kasahara T, Yano N, et al. Liver and kidney function in Japanese patients with maturity-onset diabetes of the young.
Diabetes Care
 
1998
;
21
:
2144
–8.