Risk of HCC in Chronic HBV Patients With New Onset Diabetes
Risk of HCC in Chronic HBV Patients With New Onset Diabetes
This large population-based study with careful selection of new onset diabetes strongly suggests that diabetes accelerates HCC development in chronic hepatitis B patients. The finding of a temporal relationship by which diabetes promotes HCC development in chronic hepatitis B patients could infer a causal relationship. Diabetes preceded HCC development by many years in some cases, strongly support a causal association. Some patients developed diabetes early in the cohort, by many years before the development of HCC.
Previous studies indicated the synergistic effect of diabetes and chronic hepatitis B on HCC, however, these studies failed to address the time-relationship between HBV and diabetes on HCC development. Our data on large population showed that new onset diabetes superimposed on persistent HBV infection accelerated HCC development. In addition, the potential bias in determining the time-relationship between HBV and diabetes on HCC was minimised by using age-, gender- and inception point-matched nondiabetic chronic hepatitis B patients as controls. We had previously reported the increased risk of cirrhosis and its decompensation in chronic hepatitis B patients with new onset diabetes. In our study, diabetes was identified by ICD-9 codes. Previous study showed that diabetes and family history of diabetes, especially among nondiabetics, increased the risk of fibrosis and advanced fibrosis. We did not have data on family history of diabetes in our database. Although new onset diabetes may be a reflection of co-existing diabetes, the study design to enrol 'by definition' newly diagnosed diabetes provides evidence on time-relationship between diabetes and HCC. Furthermore, we performed additional analyses to show that the risk for HCC was independent of cirrhosis, as well as hyperlipidaemia, chronic hepatitis B treatment, statins therapy, comorbidity index and obesity.
The mechanisms by which diabetes leads to HCC in chronic hepatitis B remain unclear. Emerging evidence discloses hepatic steatosis, along with insulin resistance, may precede the development of diabetes. The hypothesis of the detrimental effect of diabetes on HCC has also been linked to insulin resistance and hepatic steatosis, include stimulation of the insulin-like growth factor 1, increased leptin, reduced adiponectin and imbalance in proinflammatory/anti-inflammatory cytokines. In addition, the chromosomal telomere attrition in diabetes induces DNA damage, replicative senescence and chromosomal instability in hepatocytes. Further studies are needed to evaluate the underlying pathogenesis of diabetes in causing HCC in chronic hepatitis B patients.
This study had several strengths. First, with the inclusion of 4179 chronic hepatitis B patients with a follow-up of up to 11 years in the analysis, we were able to demonstrate the increased risk of HCC in new onset diabetes that might be under- or overestimated in a smaller study. Second, using NHIRD to identify the large cohort of chronic hepatitis B patients has good generalisability. Third, this is the largest cohort study to evaluate the association between new onset diabetes superimposed on chronic hepatitis B and HCC in the literature.
This study had several limitations. First, there exists the possibility of misclassification of both the risk and the outcome of interest (diabetes and HCC) in large administrative data sets, so that the association cannot be verified. It is a nationwide practice to systematically screen chronic hepatitis B patients with ultrasound and alpha fetoprotein for HCC every 6–12 months. However, the diagnosis of diabetes was not systematically screened at regular intervals. While most of patients with diabetes diagnosis are likely to have actual diabetes, those without the diagnosis of diabetes may have had diabetes but were under-recognised or under-reported. This misclassification bias might thus tend to diminish the true effect of diabetes. Second, the validity of the 9th International Classification of Diseases codes to define other risk factors for HCC was unknown. Nevertheless, errors in coding or recording tend to occur in random due to the same codes were used in both groups of diabetes and nondiabetes patients and might have limited effect on the calculated hazard ratios. Taiwanese NHI covers all drug treatments for diabetes and almost all diabetic patients received treatment under NHI. Thus, the possibility of misclassification of diabetics under therapy or not is very low. The agreement between NHIRD and self-reported data of 15 660 subjects ranged between 82.5% and 97.3%. The positive predictive value of NHIRD for diabetes was 0.9 in a validation study with discharge notes, laboratory data and medication orders of a medical centre. Third, some variables related to diabetes or chronic hepatitis B might not be included in this study, such as the severity of diabetes, glucose level, HbA1c level, triglyceride level, body mass index, HBV DNA level, HBeAg and family history of HCC.
Obesity increased the risk of HCC. Furthermore, obesity and alcohol have synergistic effects to increase the risk of HCC, including in hepatitis B surface antigen-positive men. Whether the synergism between obesity and alcohol uncovers incident diabetes remains unknown. Our study excluded patients with alcoholic liver disease and cirrhosis. In addition, the NHIRD database did not have data on alcohol use. Nonetheless, the Cox proportional hazard analysis showed that the increased risk of HCC in this cohort was independent of obesity. Furthermore, the risk of HCC was also independent of other metabolic risk factors including hyperlipidaemia and hypertension (Table S2 http://onlinelibrary.wiley.com/store/10.1111/apt.13191/asset/supinfo/apt13191-sup-0003-TabS1-S2-FigLegends.doc?v=1&s=fa8351f6f6ea4748fe47c09475f4356d02e87be0). In this cohort, obesity, hyperlipidaemia and hypertension were not a significant risk for HCC (Table S2 http://onlinelibrary.wiley.com/store/10.1111/apt.13191/asset/supinfo/apt13191-sup-0003-TabS1-S2-FigLegends.doc?v=1&s=fa8351f6f6ea4748fe47c09475f4356d02e87be0).
The strongest factor associated with HCC was cirrhosis (Table 3 and Table S2 http://onlinelibrary.wiley.com/store/10.1111/apt.13191/asset/supinfo/apt13191-sup-0003-TabS1-S2-FigLegends.doc?v=1&s=fa8351f6f6ea4748fe47c09475f4356d02e87be0), which is consistent with the current knowledge. Kaplan–Meier survival analysis showed a significantly higher cumulative incidences of HCC (log-rank test, P = 0.039, Figure S1) with a RR of 1.567 (95% CI, 1.042–2.357, P = 0.030) among chronic hepatitis B cirrhotics with new onset diabetes as compared to those without diabetes. In contrast, the cumulative incidences of HCC among noncirrhotics chronic hepatitis B patients with new onset diabetes was comparable to those without diabetes (log-rank test, P = 0.608, Figure S2). Thus, the risk for HCC was more significant among chronic hepatitis B cirrhotics with new onset diabetes.
(Enlarge Image)
Figure S1.
Cumulative incidences of hepatocellular carcinoma in cirrhotic patients with new onset diabetes and nondiabetes cohort. For cumulative incidence of hepatocellular carcinoma, Kaplan–Meier method was used for the analysis.
(Enlarge Image)
Figure S2.
Cumulative incidences of hepatocellular carcinoma in noncirrhotic patients with new onset diabetes and nondiabetes cohort. For cumulative incidence of hepatocellular carcinoma, Kaplan–Meier method was used for the analysis.
The Kaplan–Meier survival analysis showed that risk of HCC among diabetics was higher among those who did not receive anti-viral treatment (Table 2). It is well-known that serum HBV DNA levels is a determinant for anti-viral treatment, however, serum HBV DNA is not included in NHIRD. Whether the presence of diabetes in chronic hepatitis B is an indication for anti-viral treatment needs further study with inclusion of serum HBV DNA levels. The Kaplan–Meier survival analysis also showed that the risk for HCC was higher among diabetics without hyperlipidaemia or obesity (Table 2). Further analysis showed that the increased HCC incidence among chronic hepatitis B with new onset diabetes was independent of hyperlipidaemia and obesity (Table 3). Our unpublished data showed that the risk for HCC was lower among nondiabetic chronic hepatitis B patients with new onset hyperlipidaemia. Thus, each component of metabolic syndrome has differential effect on the risk for HCC. In addition, Cox proportional hazards analysis did not show decreased risk of HCC with anti-viral treatment (Table 3), this might be due to small sample size of patients receiving treatment (7.91% in new onset diabetes cohort and 6.88% in nondiabetes cohort) (Table 1).
In conclusion, after adjustment for age, gender, hyperlipidaemia, chronic hepatitis B treatment, statins therapy, cirrhosis, comorbidity index and obesity, new onset diabetes is still an independent predictor for HCC development in chronic hepatitis B patients. Chronic hepatitis B patients who develop diabetes during follow-up are at an increased risk of HCC over time. Every effort should be made in prevention of diabetes in chronic hepatitis B patients to lower the risk of HCC. New onset diabetes in chronic hepatitis B patients should be actively treated as a part of prevention of HCC.
Discussion
This large population-based study with careful selection of new onset diabetes strongly suggests that diabetes accelerates HCC development in chronic hepatitis B patients. The finding of a temporal relationship by which diabetes promotes HCC development in chronic hepatitis B patients could infer a causal relationship. Diabetes preceded HCC development by many years in some cases, strongly support a causal association. Some patients developed diabetes early in the cohort, by many years before the development of HCC.
Previous studies indicated the synergistic effect of diabetes and chronic hepatitis B on HCC, however, these studies failed to address the time-relationship between HBV and diabetes on HCC development. Our data on large population showed that new onset diabetes superimposed on persistent HBV infection accelerated HCC development. In addition, the potential bias in determining the time-relationship between HBV and diabetes on HCC was minimised by using age-, gender- and inception point-matched nondiabetic chronic hepatitis B patients as controls. We had previously reported the increased risk of cirrhosis and its decompensation in chronic hepatitis B patients with new onset diabetes. In our study, diabetes was identified by ICD-9 codes. Previous study showed that diabetes and family history of diabetes, especially among nondiabetics, increased the risk of fibrosis and advanced fibrosis. We did not have data on family history of diabetes in our database. Although new onset diabetes may be a reflection of co-existing diabetes, the study design to enrol 'by definition' newly diagnosed diabetes provides evidence on time-relationship between diabetes and HCC. Furthermore, we performed additional analyses to show that the risk for HCC was independent of cirrhosis, as well as hyperlipidaemia, chronic hepatitis B treatment, statins therapy, comorbidity index and obesity.
The mechanisms by which diabetes leads to HCC in chronic hepatitis B remain unclear. Emerging evidence discloses hepatic steatosis, along with insulin resistance, may precede the development of diabetes. The hypothesis of the detrimental effect of diabetes on HCC has also been linked to insulin resistance and hepatic steatosis, include stimulation of the insulin-like growth factor 1, increased leptin, reduced adiponectin and imbalance in proinflammatory/anti-inflammatory cytokines. In addition, the chromosomal telomere attrition in diabetes induces DNA damage, replicative senescence and chromosomal instability in hepatocytes. Further studies are needed to evaluate the underlying pathogenesis of diabetes in causing HCC in chronic hepatitis B patients.
This study had several strengths. First, with the inclusion of 4179 chronic hepatitis B patients with a follow-up of up to 11 years in the analysis, we were able to demonstrate the increased risk of HCC in new onset diabetes that might be under- or overestimated in a smaller study. Second, using NHIRD to identify the large cohort of chronic hepatitis B patients has good generalisability. Third, this is the largest cohort study to evaluate the association between new onset diabetes superimposed on chronic hepatitis B and HCC in the literature.
This study had several limitations. First, there exists the possibility of misclassification of both the risk and the outcome of interest (diabetes and HCC) in large administrative data sets, so that the association cannot be verified. It is a nationwide practice to systematically screen chronic hepatitis B patients with ultrasound and alpha fetoprotein for HCC every 6–12 months. However, the diagnosis of diabetes was not systematically screened at regular intervals. While most of patients with diabetes diagnosis are likely to have actual diabetes, those without the diagnosis of diabetes may have had diabetes but were under-recognised or under-reported. This misclassification bias might thus tend to diminish the true effect of diabetes. Second, the validity of the 9th International Classification of Diseases codes to define other risk factors for HCC was unknown. Nevertheless, errors in coding or recording tend to occur in random due to the same codes were used in both groups of diabetes and nondiabetes patients and might have limited effect on the calculated hazard ratios. Taiwanese NHI covers all drug treatments for diabetes and almost all diabetic patients received treatment under NHI. Thus, the possibility of misclassification of diabetics under therapy or not is very low. The agreement between NHIRD and self-reported data of 15 660 subjects ranged between 82.5% and 97.3%. The positive predictive value of NHIRD for diabetes was 0.9 in a validation study with discharge notes, laboratory data and medication orders of a medical centre. Third, some variables related to diabetes or chronic hepatitis B might not be included in this study, such as the severity of diabetes, glucose level, HbA1c level, triglyceride level, body mass index, HBV DNA level, HBeAg and family history of HCC.
Obesity increased the risk of HCC. Furthermore, obesity and alcohol have synergistic effects to increase the risk of HCC, including in hepatitis B surface antigen-positive men. Whether the synergism between obesity and alcohol uncovers incident diabetes remains unknown. Our study excluded patients with alcoholic liver disease and cirrhosis. In addition, the NHIRD database did not have data on alcohol use. Nonetheless, the Cox proportional hazard analysis showed that the increased risk of HCC in this cohort was independent of obesity. Furthermore, the risk of HCC was also independent of other metabolic risk factors including hyperlipidaemia and hypertension (Table S2 http://onlinelibrary.wiley.com/store/10.1111/apt.13191/asset/supinfo/apt13191-sup-0003-TabS1-S2-FigLegends.doc?v=1&s=fa8351f6f6ea4748fe47c09475f4356d02e87be0). In this cohort, obesity, hyperlipidaemia and hypertension were not a significant risk for HCC (Table S2 http://onlinelibrary.wiley.com/store/10.1111/apt.13191/asset/supinfo/apt13191-sup-0003-TabS1-S2-FigLegends.doc?v=1&s=fa8351f6f6ea4748fe47c09475f4356d02e87be0).
The strongest factor associated with HCC was cirrhosis (Table 3 and Table S2 http://onlinelibrary.wiley.com/store/10.1111/apt.13191/asset/supinfo/apt13191-sup-0003-TabS1-S2-FigLegends.doc?v=1&s=fa8351f6f6ea4748fe47c09475f4356d02e87be0), which is consistent with the current knowledge. Kaplan–Meier survival analysis showed a significantly higher cumulative incidences of HCC (log-rank test, P = 0.039, Figure S1) with a RR of 1.567 (95% CI, 1.042–2.357, P = 0.030) among chronic hepatitis B cirrhotics with new onset diabetes as compared to those without diabetes. In contrast, the cumulative incidences of HCC among noncirrhotics chronic hepatitis B patients with new onset diabetes was comparable to those without diabetes (log-rank test, P = 0.608, Figure S2). Thus, the risk for HCC was more significant among chronic hepatitis B cirrhotics with new onset diabetes.
(Enlarge Image)
Figure S1.
Cumulative incidences of hepatocellular carcinoma in cirrhotic patients with new onset diabetes and nondiabetes cohort. For cumulative incidence of hepatocellular carcinoma, Kaplan–Meier method was used for the analysis.
(Enlarge Image)
Figure S2.
Cumulative incidences of hepatocellular carcinoma in noncirrhotic patients with new onset diabetes and nondiabetes cohort. For cumulative incidence of hepatocellular carcinoma, Kaplan–Meier method was used for the analysis.
The Kaplan–Meier survival analysis showed that risk of HCC among diabetics was higher among those who did not receive anti-viral treatment (Table 2). It is well-known that serum HBV DNA levels is a determinant for anti-viral treatment, however, serum HBV DNA is not included in NHIRD. Whether the presence of diabetes in chronic hepatitis B is an indication for anti-viral treatment needs further study with inclusion of serum HBV DNA levels. The Kaplan–Meier survival analysis also showed that the risk for HCC was higher among diabetics without hyperlipidaemia or obesity (Table 2). Further analysis showed that the increased HCC incidence among chronic hepatitis B with new onset diabetes was independent of hyperlipidaemia and obesity (Table 3). Our unpublished data showed that the risk for HCC was lower among nondiabetic chronic hepatitis B patients with new onset hyperlipidaemia. Thus, each component of metabolic syndrome has differential effect on the risk for HCC. In addition, Cox proportional hazards analysis did not show decreased risk of HCC with anti-viral treatment (Table 3), this might be due to small sample size of patients receiving treatment (7.91% in new onset diabetes cohort and 6.88% in nondiabetes cohort) (Table 1).
In conclusion, after adjustment for age, gender, hyperlipidaemia, chronic hepatitis B treatment, statins therapy, cirrhosis, comorbidity index and obesity, new onset diabetes is still an independent predictor for HCC development in chronic hepatitis B patients. Chronic hepatitis B patients who develop diabetes during follow-up are at an increased risk of HCC over time. Every effort should be made in prevention of diabetes in chronic hepatitis B patients to lower the risk of HCC. New onset diabetes in chronic hepatitis B patients should be actively treated as a part of prevention of HCC.
Source...