SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Subjects and Methods
  4. Results
  5. Discussion
  6. References

We prospectively investigated 615,532 diabetic patients and 614,871 age-matched and sex-matched control subjects selected from National Health Insurance claims for malignant neoplasms of liver and biliary tract (International Statistical Classification of Diseases and Related Health Problems, 9th edition, codes 155 and 156, respectively) between 2000 and 2006. The person-year approach with Poisson assumption was used to estimate the hazard rates. We also evaluated the age-specific and sex-specific relative risks of these two malignancies in relation to diabetes with Cox proportional hazard regression model with adjustment for potential confounders. The overall hazard rate of malignant neoplasm of the liver was 32.76 and 17.41 per 10,000 patient-years, respectively, for diabetic men and women; the corresponding figures for biliary tract neoplasm were much lower at 1.42 and 1.60 per 10,000 patient-years. Compared with control subjects, diabetic patients had a two-fold increased risk of malignant neoplasm of the liver, but this risk was attenuated by adjusting for selected clinical risk factors (hazard ratio [HR] 1.21; 95% confidence interval [CI] 1.17-1.25). Additionally, diabetic patients were associated with increased risk of biliary neoplasms with an approximate magnitude of 20%-30%, but the HR was attenuated and became insignificant after adjustment for clinical risk factors (HR 1.07; 95% CI 0.95-1.21). Diabetic patients with cirrhosis had the highest relative risk of liver neoplasm (HR 85.25; 95% CI 76.84-94.58), whereas those with cholangitis had the highest risk of biliary tract neoplasm (HR 70.30; 95% CI 51.95-95.12) compared with control subjects without any clinical risk factors. Conclusion: This population-based study confirms the association of diabetes with liver neoplasm and suggests that diabetic patients with certain clinical risk factors should be educated for strict adherence of liver neoplasm screening. (HEPATOLOGY 2010)

Primary tumor of the liver represents the sixth most common malignancy worldwide and the third most common cause of death from cancer.1 Although malignancies of the biliary tract are less common, their incidence and mortality have been on the rise worldwide.2 Diabetes, whose global prevalence has been rising,3 has been associated with increased risks of hepatocellular carcinoma4-18 and cholangiocarcinoma,5, 8, 19, 20 but some studies have not observed an association of diabetes mellitus with malignant neoplasm of liver21 or with biliary tract cancer.15, 16, 18, 21

A majority of previous studies were conducted with a case-control design,4, 6, 7, 9-12, 14, 18-20 and many of them had a limited number of study subjects. Some cohort studies5, 8, 13 recruited diabetic patients only from in-patient registries; others limited the study subjects to government employees and their dependents,15 middle-aged patients,16 and male diabetic patients.13 Moreover, some population-based cohort studies17, 21 localized their study subjects to regional areas rather than the whole national population. To our knowledge, no studies thus far have investigated the incidence and relative risk of malignant neoplasms of liver and biliary tract according to different age and sex stratifications.

In Taiwan, liver and biliary tract malignancies are the most common cancer in men and the third most common cancer in women.22 Lai et al.17 reported an increased risk of hepatocellular carcinoma in type 2 diabetic patients within the Keelung Community in northern Taiwan, but they did not further analyze the relationship between diabetes and biliary tract neoplasm. The aim of this study was to estimate the hazard rates and relative risks of malignant neoplasms of the liver and biliary tract in diabetic population according to sex and various age stratifications using a nationally representative diabetic cohort selected from National Health Insurance (NHI).

Subjects and Methods

  1. Top of page
  2. Abstract
  3. Subjects and Methods
  4. Results
  5. Discussion
  6. References

Identification of Study Subjects.

Taiwan's NHI program is a universal health program that was introduced in March 1995. By the end of 1996, approximately 96% of the total Taiwanese population had enrolled in the NHI program,23 and the state-run Bureau of National Health Insurance (BNHI) had contracted with 97% of hospitals as well as 90% of clinics all over the island.24 The BNHI accumulates all administrative and claims data for Taiwan, and the National Health Research Institute cooperates with the BNHI to establish an NHI research database. For the precision of the claim data, the BNHI performs expert reviews on a random sample of every 50-100 ambulatory and inpatient claims in each hospital and clinic quarterly, and false reports of diagnosis yield a severe penalty from the BNHI.25 With ethical approval from the National Health Research Institute, we used data for the ambulatory care claims, all inpatient claims, and updated registry for beneficiaries from 1997 to 2006 for this study. All NHI datasets can be interlinked with each individual personal identification number (PIN).

Diabetic ambulatory care claims record patients with diabetes-related diagnoses with International Statistical Classification of Diseases and Related Health Problems, 9th edition (ICD-9) code 250 or A-code A181. An individual was classified as a diabetic patient if she or he had an initial diabetes-related diagnosis at any time in 2000 and then experienced one or more additional diagnoses within the subsequent 12 months. The first and last outpatient visits within 1 year must be >30 days apart to avoid accidental inclusion of miscoded patients.26 To detect newly diagnosed malignant neoplasm cases, we excluded those patients admitted to the hospitals for any kinds of malignant neoplasm (ICD-9: 140-208) during 1997-1999 from our diabetic group. In Taiwan, BNHI issues major illness/injury certificates to all patients who suffer from malignant neoplasm, and these patients are exempt from copayment to the NHI if they are admitted for the illness associated with the related malignancy. To further ensure the accuracy of the diagnosis of malignant neoplasm, we excluded only those patients using major illness/injury certificates for the particular admissions. The final diabetic cohort consisted of 615,532 patients. The index date for patients in the diabetic group was the date of their first outpatient visit for diabetes care in 2000.

The control group was identified from the registry of beneficiaries, which accumulates information of all beneficiaries, including PIN, date of birth, sex, geographic area of each member's NHI units, and date of enrollment and withdrawal from each time between March 1995 and December 2006. The total number of beneficiaries as of January 1, 2000, was 22,176,542 with a mean age (± standard deviation) of 32.17 ± 20.40 years and a male/female ratio of 50.5:49.5. After excluding individuals included in diabetic ambulatory care claims and hospitalized for any type of malignancy (ICD-9: 140-208) using major illness/injury certificates between 1997 and 1999, we selected control subjects by way of an age-matched and sex-matched frequency-matching technique. Because of missing information on age or sex for 661 diabetic patients, we could only choose 614,871 control subjects in this analysis. The index date for subjects in the control group was the first date of enrollment to the NHI. If their first date of enrollment was before January 1, 2000, the index date was set as January 1, 2000, which was the starting point of follow-up.

The age of each study subject was determined by the difference in time between the index date and the date of birth. Additionally, the geographic area of each member's NHI unit, either the beneficiaries' residential area or location of their employment, was grouped into four geographic areas (North, Central, South, East) or two urbanization statuses (urban and rural) according to the National Statistics of Regional Standard Classification.27

Study Endpoints.

The inpatient claims include the records of all hospitalizations and provide various pieces of information, including PIN, date of birth, sex, date of admission and discharge, a maximum of five leading discharged diagnoses and four operation codes, partial amount of expenses paid by the beneficiaries for the admission, and so forth. With the unique PIN, we linked study subjects in both diabetic and control groups to the inpatient claim data from 2000 to 2006 to identify, if any, the first episode of primary or secondary diagnoses of malignant neoplasm of liver (ICD-9: 155) and biliary tract (ICD-9: 156) as the endpoints of this study. For the accuracy of the diagnoses of malignant neoplasm, we retrieved only those patients using major illness/injury certificates for that particular admission. Both outcomes were analyzed separately. The date of encountering each clinical endpoint of interest was the first day of hospitalization. The study period was from January 1, 2000, to December 31, 2006, a 7-year-period.

Clinical Risk Factors.

We identified clinical risk factors for malignant neoplasms of the liver14 (hepatitis B [ICD-9: 070.2, 070.3, V02.61], hepatitis C [ICD-9: 070.41, 070.44, 070.51, 070.54, V02.62], unspecified chronic hepatitis [ICD-9: 070.9, 571.4, 571.8, 571.9], alcoholic liver disease [ICD-9: 571.0, 571.1, 571.2, 571.3], cirrhosis [ICD-9: 571.5, 571.6]) and biliary tract19 (cholangitis [ICD-9: 575.8, 576.1], cholecystitis [ICD-9: 575.0, 575.11, 575.12], cholelithiasis [ICD-9: 574], choledocholithiasis [ICD-9: 574.5], biliary cirrhosis [ICD-9: 571.6]) from inpatient claims (1997-2006) and related ambulatory care claims of both diabetic and control subjects (1997-2006). We counted the above clinical risk factors occurring in individuals in both groups only when the dates of diagnosis for the selected illnesses (clinical risk factors) were noted before or the day on which the study subjects' endpoints or censoring took place. The following criteria were set as censoring dates for the study subjects. First, if a subject died in the hospital from causes other than the study endpoints, the date of censoring was the date of death. Second, if a subject did not encounter in-hospital mortality, the date of censoring was either the date of their last withdrawal from NHI or the date of termination of the study (December 31, 2006).

Statistical Methods.

We performed two major statistical analyses in this study. First, the age-specific and sex-specific hazard rate was estimated using person-years as the denominator under the Poisson assumption. Second, to determine the independent effects of diabetes on the risks of malignant neoplasms of the liver and biliary tract, we used Cox proportional hazard regression models with age, sex, geographic area, urbanization statuses, and related clinical risk factors adjusted simultaneously in the model. We adjusted geographic variables for possible geographic variations in the incidence of hepatocellular carcinoma.28 Furthermore, we explored the relative hazards of malignant neoplasm of the liver and biliary tract in relation to diabetes accompanied by the selected clinical risk factors individually with Cox proportional hazard regression models with age, sex, geographic area, and urbanization statuses adjusted in the model. All statistical analyses were performed with SAS version 9.1 (SAS Institute, Cary, NC). A P value <0.05 was considered statistically significant.

Results

  1. Top of page
  2. Abstract
  3. Subjects and Methods
  4. Results
  5. Discussion
  6. References

Characteristics of the Study Subjects.

The mean (± standard deviation) age of the diabetic group was 60.09 ± 12.73 years, whereas that of the control subjects was 60.00 ± 12.84 years. The percentages of people aged <45, 45-64, and >64 years were 11.3%, 48.3%, and 40.4% in both the control group and the diabetic population. The ratio of men to women was 51.9:48.1 in both groups. The details of geographic and clinical risk factors distribution are shown in Table 1. The median time of follow-up was similar at 6.9 years for both groups.

Table 1. Characteristics of the Study Subjects
Variables*Control Group, n (%)Diabetic Group, n (%)
  • *

    Inconsistency between the total population and the population added for individual variables was due to missing information.

General characteristics
 Age (years)
  <4569,617 (11.32)69,825 (11.34)
  45-64296,810 (48.27)297,142 (48.27)
  >64248,444 (40.41)248,562 (40.39)
  Mean age   (± standard deviation)60.00 (12.84)60.09 (12.73)
 Sex
  Female319,308 (51.93)319,310 (51.93)
  Male295,563 (48.07)295,566 (48.07)
 Geographic area
  Northern269,239 (44.29)269,920 (44.41)
  Central151,693 (24.96)141,321 (23.25)
  Southern168,995 (27.80)178,627 (29.39)
  Eastern17,938 (2.95)17,944 (2.95)
 Urbanization status
  Urban area407,323 (66.81)415,154 (68.16)
  Rural area202,343 (33.19)193,949 (31.84)
Clinical risk factors
 Hepatitis B
  No592,094 (96.30)583,344 (94.77)
  Yes22,777 (3.89)32,188 (5.23)
 Hepatitis C
  No599,384 (97.48)589,782 (95.82)
  Yes15,487 (2.52)25,750 (4.18)
 Unspecified chronic hepatitis
  No452,976 (73.67)354,888 (57.66)
  Yes161,895 (26.33)260,644 (42.34)
 Alcoholic liver disease
  No604,804 (98.29)595,601 (96.76)
  Yes10,067 (1.71)19,931 (3.24)
 Cirrhosis
  No594,230 (96.64)574,716 (93.37)
  Yes20,641 (3.36)40,816 (6.63)
 Cholecystitis
  No607,954 (98.88)604,119 (98.15)
  Yes6,917 (1.12)11,413 (1.85)
 Cholangitis
  No606,879 (98.70)604,275 (98.17)
  Yes7,992 (1.30)11,257 (1.83)
 Cholelithiasis
  No559,856 (91.05)535,981 (87.08)
  Yes55,015 (8.95)79,551 (12.92)
 Choledocholithiasis
  No606,032 (98.56)604,008 (98.13)
  Yes8,839 (1.44)11,524 (1.87)
 Biliary cirrhosis
  No614,471 (99.93)614,859 (99.89)
  Yes400 (0.07)673 (0.11)
Total614,871 (100.00)615,532 (100.00)

Incidence and Relative Hazards of Malignant Neoplasm of the Liver.

The overall and age-specific and sex-specific hazard rate of malignant neoplasm of the liver are presented in Table 2. The overall hazard rate for diabetic men and women was 32.76 and 17.41 per 10,000 patient-years, respectively, while the corresponding figures for control men and women were 16.51 and 9.09 per 10,000 patient-years. Except for diabetic men aged >64 years, the hazard rate increased with age in both diabetic and control groups irrespective of sex.

Table 2. Overall and Age-Specific and Sex-Specific Hazard Rates of Malignant Neoplasm of the Liver (ICD-9: 155) in Diabetic and Control Groups
Sex and Age*Control GroupDiabetic Group
No. of PatientsNo. of EventsHR per 10,000 Patient-Years (95% CI)No. of PatientsNo. of EventsHR per 10,000 Patient-Years (95% CI)
  • *

    Inconsistency between the total population and the population added for individual variables was due to missing information.

  • Based on Poisson assumption.

Male      
 <45 years40,5371324.90 (4.07-5.74)40,53735013.66 (12.23-15.10)
 45-65 years141,8991,62117.33 (16.49-18.18)141,8993,37437.71 (36.43-38.98)
 >65 years113,1271,42819.75 (18.73-20.78)113,1292,29333.44 (32.08-34.81)
 Total295,5633,18116.51 (15.93-17.08)295,5666,01732.76 (31.93-33.59)
Female      
 <45 years29,080190.98 (0.59-1.53)29,079331.78 (1.23-2.49)
 45-65 years154,9117407.19 (6.67-7.71)154,9111,47814.94 (14.18-15.71)
 >65 years135,3171,15513.06 (12.31-13.82)135,3182,00223.73 (22.69-24.77)
 Total319,3081,9149.09 (8.68-9.49)319,3103,51317.41 (16.83-17.98)
Overall614,8715,09512.63 (12.28-12.98)615,5329,53024.70 (24.20-25.19)

Compared with the control subjects, diabetic men and women showed significantly elevated risks of malignant neoplasm of the liver with a hazard ratio (HR) 1.99 and 95% confidence interval (CI) 1.90-2.08 (men) and HR 1.90 and 95% CI 1.79-2.01 (women), whereas adjustments were made only for age, sex, and geographic statuses. If we further adjusted for clinical risk factors in the model, the HRs were attenuated to 1.20 (95% CI 1.15-1.26) and 1.22 (95% CI 1.15-1.29) in diabetic men and women, respectively. If we excluded those patients with malignant neoplasm of the liver diagnosed within 6 months of the index date, the adjusted HRs were essentially the same as those estimates without such exclusion (HR 1.19, 95% CI 1.15-1.23). Because there was a significant interaction of diabetes with age (P < 0.0001) for both men and women, we performed the stratified analysis to estimate the age-specific HRs for each sex. The diabetic patients with younger ages had higher HRs, but they became insignificant after adjustment of clinical risk factors. The highest age-specific HR was observed for diabetic men and women aged >65 years (HR 1.27, 95% CI 1.19-1.36 [men]; HR 1.26, 95% CI 1.17-1.36 [women]) with inclusion of clinical risk factors in the model. (Table 3).

Table 3. Overall and Age-Specific and Sex-Specific Relative Hazards of Malignant Neoplasm of the Liver (ICD-9: 155) in Association With Diabetes
Sex and Age*Control GroupDiabetic GroupCrude HR (95% CI) in Association With DiabetesAdjusted HR (95% CI) in Association With DiabetesAdjusted HR (95% CI) in Association With Diabetes
No. of PatientsNo. of EventsNo. of PatientsNo. of Events
  • *

    Inconsistency between the total population and the population added for individual variables was due to missing information.

  • Based on Cox proportional hazard regression with adjustment for age, sex, geographic area, and urbanization status.

  • Based on Cox proportional hazard regression with adjustment for age, sex, geographic area, urbanization status, hepatitis B, hepatitis C, unspecifed chronic hepatitis infection, alcoholic liver disease, and liver cirrhosis.

Male       
 <45 years40,53713240,5373502.74 (2.24-3.35)2.68 (2.19-3.29)0.94 (0.76-1.16)
 45-65 years141,8991,621141,8993,3742.16 (2.03-2.29)2.15 (2.02-2.28)1.15 (1.08-1.22)
 >65 years113,1271,428113,1292,2931.68 (1.57-1.79)1.73 (1.62-1.85)1.27 (1.19-1.36)
 Total295,5633,181295,5666,0171.97 (1.89-2.06)1.99 (1.90-2.08)1.20 (1.15-1.26)
Female       
 <45 years29,0801929,079331.83 (1.04-3.22)1.83 (1.04-3.22)0.75 (0.41-1.37)
 45-65 years154,911740154,9111,4782.08 (1.91-2.28)2.06 (1.88-2.25)1.16 (1.05-1.27)
 >65 years135,3171,155135,3182,0021.79 (1.67-1.93)1.80 (1.67-1.93)1.26 (1.17-1.36)
 Total319,3081,914319,3103,5131.91 (1.80-2.01)1.90 (1.79-2.01)1.22 (1.15-1.29)
Overall614,8715,095615,5329,5301.94 (1.88-2.01)1.95 (1.89-2.02)1.21 (1.17-1.25)

Incidence and Relative Hazards of Malignant Neoplasm of the Biliary Tract.

For malignant neoplasm of the biliary tract, the overall hazard rate estimated for diabetic men and women, respectively, was 1.42 and 1.60 per 10,000 patient-years. The corresponding data for control men and women, respectively, were 1.09 and 1.30 per 10,000 patient-years (Table 4). The hazard rate increased with age irrespective of sex and diabetic status. The sex-specific HRs were 1.29 (95% CI 1.07-1.55) and 1.22 (95% CI 1.03-1.43), respectively, for diabetic men and women if we adjusted age, sex, and geographic statuses in the model. However, if we included additional clinical risk factors in the model, the results were no longer significant in both sexes. At the same time, the adjusted HRs estimated from the data excluding biliary neoplasm diagnosed within 6 months of index dates were similar to those estimates without exclusion (HR 1.07, 95% CI 0.94-1.22 with adjustment of clinical risk factors). Again, we noted a significant interactive effect of diabetes and age on risk of biliary tract neoplasm for both men and women (P < 0.0001). Compared with control subjects with the same sex, the relative hazards of biliary tract cancer significantly increased in diabetic men and women aged 45-64 years, but the risks became insignificant with addition of clinical risk factors to the model (Table 5).

Table 4. Overall and Age-Specific and Sex-Specific Hazard Rates of Malignant Neoplasm of the Biliary Tract (ICD-9: 156) in the Diabetic and Control Groups
Sex and Age*Control GroupDiabetic Group
No. of PatientsNo. of EventsHR per 10,000 Patient-Years (95% CI)No. of PatientsNo. of EventsHR per 10,000 Patient-Years (95% CI)
  • *

    Inconsistency between the total population and the population added for individual variables was due to missing information.

  • Based on Poisson assumption.

Male      
 <45 years40,53730.11 (0.02-0.33)40,53740.16 (0.04-0.39)
 45-65 years141,899800.85 (0.67-1.05)141,8991181.31 (1.08-1.56)
 >65 years113,1271281.76 (1.46-2.07)113,1291412.04 (1.70-2.38)
 Total295,5632111.09 (0.94-1.24)295,5662631.42 (1.25-1.59)
Women      
 <45 years29,08040.21 (0.01-0.52)29,07930.16 (0.03-0.47)
 45-65 years154,911900.87 (0.70-1.07)154,9111191.20 (0.99-1.43)
 >65 years135,3171802.03 (1.73-2.33)135,3182022.38 (2.05-2.71)
 Total319,3082741.30 (1.15-1.45)319,3103241.60 (1.42-1.77)
Overall614,8714851.20 (1.09-1.31)615,5325871.51 (1.39-1.63)
Table 5. Overall and Age- and Sex-Specific Relative Hazards of Malignant Neoplasm of the Biliary Tract (ICD-9: 156) in Association With Diabetes
Sex and Age*Control GroupDiabetic GroupCrude HR (95% CI) in Association With DiabetesAdjusted HR (95% CI) in Association With DiabetesAdjusted HR (95% CI) in Association With Diabetes
No. of PatientsNo. of EventsNo. of PatientsNo. of Events
  • *

    Inconsistency between the total population and the population added for individual variables was due to missing information.

  • Based on Cox proportional hazard regression with adjustment for age, sex, geographic area, and urbanization status.

  • Based on Cox proportional hazard regression with adjustment for age, sex, geographic area, urbanization status, and status of cholangitis, cholecystitis, cholelithiasis, choledocholithiasis, and biliary cirrhosis.

Male       
 <45 years40,537340,53741.43 (0.32-6.40)1.45 (0.32-6.49)0.80 (0.17-3.75)
 45-65 years141,89980141,8991181.51 (1.13-2.00)1.48 (1.11-1.98)1.21 (0.91-1.62)
 >65 years113,127128113,1291411.17 (0.92-1.49)1.17 (0.91-1.49)1.05 (0.82-1.34)
 Total295,563211295,5662631.30 (1.08-1.56)1.29 (1.07-1.55)1.12 (0.93-1.35)
Female       
 <45 years29,080429,07930.83 (0.18-3.70)0.80 (0.18-3.62)0.51 (0.11-2.38)
 45-65 years154,91190154,9111191.37 (1.04-1.80)1.35 (1.03-1.78)1.12 (0.85-1.48)
 >65 years135,317180135,3182021.15 (0.94-1.41)1.15 (0.94-1.41)0.99 (0.81-1.22)
 Total319,308274319,3103241.22 (1.04-1.43)1.22 (1.03-1.43)1.04 (0.88-1.22)
Overall614,871485615,5325871.25 (1.11-1.41)1.25 (1.10-1.41)1.07 (0.95-1.21)

Overall Relative Hazards for Diabetic Patients Accompanied by Selected Clinical Risk Factors.

Table 6 presents the overall relative hazards of malignant neoplasm of the liver and biliary tract in relation to diabetes accompanied by various clinical risk factors. Diabetic subjects without any clinical risk factors had an increased risk of liver and biliary tract malignant neoplasm by a magnitude of 72% and 45%, respectively. Additionally, the highest HR of malignant neoplasm of liver was associated with cirrhosis (HR 85.25, 95% CI 76.84-94.58), whereas that of malignant neoplasm of the biliary tract was associated with cholangitis (HR 70.30, 95% CI 51.95-95.12).

Table 6. Overall Numbers and Relative Hazards of Malignant Neoplasm of the Liver (ICD-9: 155) and Biliary Tract (ICD-9: 156) in Relation to Diabetes Accompanied by Selected Clinical Risk Factors
Study GroupMalignant NeoplasmHR (95% CI)*
NoYes
  • *

    Based on Cox proportional hazard regression with adjustment for age, sex, geographic area, and urbanization status.

Malignant neoplasm of the liver   
 Control subjects without any clinical risk factors437,0795841.00
 Diabetic subjects without any clinical risk factors335,5197721.72 (1.54-1.92)
 Diabetic subjects with hepatitis B infection only29,4592,72918.82 (16.18-21.90)
 Diabetic subjects with hepatitis C infection only22,5493,20127.95 (23.84-32.77)
 Diabetic subjects with unspecified chronic hepatitis only253,1287,5165.34 (4.85-5.87)
 Diabetic subjects with alcoholic liver disease only18,6871,24410.17 (8.17-12.64)
 Diabetic subjects with cirrhosis only33,9826,83485.25 (76.84-94.58)
Malignant neoplasm of the biliary tract   
 Control subjects without any clinical risk factor555,0451521.00
 Diabetic subjects without any clinical risk factor528,2772081.45 (1.17-1.79)
 Diabetic subjects with cholecystitis only11,3367718.63 (12.17-28.53)
 Diabetic subjects with cholangitis only11,08517270.30 (51.95-95.12)
 Diabetic subjects with cholelithiasis only79,2562957.99 (6.38-10.00)
 Diabetic subjects with choledocholithiasis only11,41910524.63 (17.83-34.02)
 Diabetic subjects with biliary cirrhosis only670310.22 (1.36-76.85)

Discussion

  1. Top of page
  2. Abstract
  3. Subjects and Methods
  4. Results
  5. Discussion
  6. References

Diabetes and Malignant Neoplasm of the Liver.

Like a previous Taiwanese study,17 our study showed that the incidence of primary malignant neoplasm of liver increased with age. Our results also indicated that the incidence of primary malignant neoplasm of the liver in the diabetic patients was significantly higher (about two-fold) than that of age-matched and sex-matched control subjects. The highest age-specific hazard ratio was observed in diabetic men aged 45-64 years and in diabetic women aged >64 years.

Compared with the age- and sex-matched control group, the overall risk of malignant neoplasm of the liver in our diabetic population was modestly increased even after adjustment for various clinical risk factors. There were a few previous studies that adjusted for clinical risk factors in their multivariable analyses. The association between diabetes and hepatic cancer lost significance after controlling for cirrhosis in one case-control study,10 but not so in the other two studies.6, 7 In one cohort study, the diabetes and hepatocellular carcinoma association became weaker but was still statistically significant after adjustment for age, sex, alcoholic liver disease, and viral hepatitis status.13 The risk was reported to be higher in diabetic men than in diabetic women,5, 8, 15 but in some studies the risk of hepatocellular carcinoma was similar in both sexes,7, 16, 18 as was our observation. Further age-stratified analysis demonstrated that older diabetic patients (>65 years) were at the most elevated relative risk of malignant neoplasm of the liver irrespective of sex even after adjustment of age, sex, geographic, and urbanization statuses as well as various clinical risk factors. In one previous study, El-serag et al.13 also reported that older age (per 10 years) increased risk of hepatocellular carcinoma with HR 9.61 (95% CI 4.69-19.68) in hospitalized veterans.

We also found that diabetic patients without any clinical risk factors still increased the risk of malignant neoplasm of the liver as compared with the nondiabetic counterpart. The possible biological mechanism associated with diabetes and primary malignant neoplasm of liver has not been clearly elucidated. Diabetes predisposes to nonalcoholic fatty liver disease.29 Nonalcoholic steatohepatits, a severe form of nonalcoholic fatty liver disease is regarded as an entity that can progress to cirrhosis29 as well as primary liver cancers.30, 31 Moreover, type 2 diabetes is a risk factor of hepatitis C,32 which can also progress to cirrhosis and hepatocellular carcinoma.33 Additionally, type 2 diabetes has been found to be associated with insulin resistance, compensatory hyperinsulinemia9, 11 and elevated level of insulin-like growth factor,11 which may interact with liver cells to stimulate mitogenesis and carcinogenesis. In some studies, insulin use in diabetes was also reported to be associated with hepatocellular carcinoma4 and in turn increased cancer-related mortality.34

Compared with control subjects without clinical risk factors, diabetic patients with hepatitis B and C in our study had significantly increased risk of malignant neoplasm of the liver by magnitudes comparable with those of previous studies.11, 14 The HR of diabetic patients with cirrhosis in our findings was also higher than those with alcoholic liver disease including cirrhosis, which was similar to the findings from a population-based United States study.14 Screening of every diabetic patient for hepatic neoplasm might not be cost-effective, because these outcomes are rare even among diabetic patients. However, the HRs of diabetic patients with hepatitis B, hepatitis C, and cirrhosis were significantly increased enough that diabetologists should educate patients with those clinical risk factors for strict adherence to the present liver cancer screening program. Although some other potential confounding factor such as obesity35 might be responsible for the increased risk of liver cancer rather than diabetes itself, one previous study7 that adjusted for body mass index (BMI) in a multivariate analysis found that BMI had no effect on the significant association of diabetes and hepatocellular carcinoma. Stattin et al.21 also reported that adjustment for BMI had no material effect on risk estimates of hyperglycemia and cancer risk. A recent Taiwanese study36 prospectively followed 2,903 male hepatitis B virus surface antigen-positive government employees for a mean of 14.7 years, and reported a significant increase in the risk of hepatocellular carcinoma (HR 1.48, 95% CI 1.04-2.12) in overweight men (BMI between 25.0 and 29.9 kg/m2). The HR increased to 1.96 (95% CI 0.72-5.38) in obese men (BMI ≥30.0 kg/m2). This study thus concluded that excess body weight is involved in the transition from healthy hepatitis B carrier state to hepatocellular carcinoma among men. Nonetheless, our study demonstrated that, even in the absence of hepatitis B, diabetic patients were still at a significantly greater risk of liver cancer. Because no anthropometric data are available from the NHI data, we were unable to empirically assess the extent to which obesity would confound the relationship between diabetes and liver cancer observed in our study.

Diabetes and Malignant Neoplasm of Biliary Tract.

The incidence of biliary tract cancers of diabetic patients was scarcely investigated in the literature. Irrespective of diabetic status, the incidence of biliary tract neoplasm increased with age, and they were higher in men than in women except in those diabetic patients >64 years. Without adjustment for clinical risk factors, we observed a 20%-30% increased risk of biliary tract neoplasm in diabetic patients, which was comparable with some previous studies,5, 8 but lower than some others.19 In our study, the risk estimates of diabetic women and diabetic men were similar to the findings of Wideroff et al.8 Further age stratifications revealed that only those subjects aged 45-64 years had significant increased risks compared with the age-matched and sex-matched control group in both sexes, but its significance was lost after adjustment for additional clinical risk factors. Prior studies5, 8, 19, 20 that reported association of diabetes and biliary cancer did not adjust for clinical risk factors in the multivariate analyses. Some previous case-control studies indicated an increased risk of gallbladder cancer in obese women.37 Additionally, Grainge et al.20 reported that a BMI ≥30 was associated with mild increased risk of cholangiocarcinoma. Because the relative risk estimates of biliary tract cancer noted in our study were close to null after adjustment for certain known clinical risk factors for biliary tract cancer, the potential confounding by obesity should not be substantial.

In our study, we observed that diabetes with cholecystitis, cholangitis, cholelithoasis, choledocholithiasis, or biliary cirrhosis significantly increased the risk of malignant neoplasm of the biliary tract compared with control subjects without any clinical risk factors. Those risk estimates were similar to those reported in previous studies19 that explored the risk factors for cholangiocarcinoma.

Methodological Consideration and Conclusion.

There were several methodological strengths in our study. First, the diabetic and control groups were retrieved from the NHI database, which is population-based and highly representative, causing little possibility of recall and selection bias. In addition, there is little likelihood of nonresponse and loss to follow-up of cohort members. Second, one of the potential advantages of using insurance claim datasets in clinical research is easy access to the longitudinal records for a large sample of patients from different geographic areas.38 Third, the large number of study subjects also made it possible for us to make age-stratified and sex-stratified analyses without compromising the required sample size. Fourth, because the diagnostic procedures of liver and biliary tract cancers can be dependent on medical resources and physicians' behavior, adjustment for geographic area and urbanization level made it possible in reducing such geographic-related and urbanization-related confounding factors. Finally, we excluded those patients with all types of malignancy 3 years before the index date so that we could obtain relatively accurate estimates of incidence and relative risks of malignant neoplasms of the liver and biliary tract.

In spite of the above strengths, several limitations should be noted in our study. First, exclusive reliance on the claim data might have resulted in potential disease misclassification bias in our study. The accuracy of a single diabetes diagnosis in the NHI claim data in 2000 was reported to be 74.6%,39 but we used at least two diabetes-related diagnoses with the first and the last visits >30 days apart, which may largely reduce the likelihood of disease misclassification. Despite that, the control group might still have been mixed up with new onset or undiagnosed diabetes. Furthermore, because we only selected those patients with major illness/injury certificates for the accuracy of diagnosis of malignancy, we might have missed some patients who had been waiting for the pathological diagnosis and had not received major illness/injury certificates. Such misclassification bias, however, was likely to be nondifferential, which tends to underestimate rather overestimate the true relative risks.40 Second, we were unable to differentiate between type 1 and type 2 diabetes in our study, which also limits specific interpretations of the study results. Third, we could not determine the BMI, duration and treatment regimens of diabetes, smoking, alcohol consumption, and other socioeconomic characteristics in our study population, which might have also confounded the study results. Fourth, our exclusive reliance on inpatient claims for the diagnosis of liver and biliary cancers would have missed some of the patients who were not hospitalized, which could underestimate the incidence rate, but it would have had little influence on the relative risk estimates of those cancers associated with diabetes. Finally, screening or surveillance bias might be a concern in our study, because there are more frequent physician contacts for the diabetic patients. To assess whether the significant association of diabetes with malignant neoplasm of the liver was due to frequent surveillance of disease among diabetic patients, we limited our control subjects to hypertensive subjects (ICD-9: 401-405) who can also be considered as frequent clinic visitors. The results showed that the recalculated point estimates of HRs for malignant neoplasm of liver (HR 1.28) and biliary tract (HR 1.02) were very similar to the original estimates (HR 1.21 in Table 3 and 1.07 in Table 5, respectively), suggesting little surveillance bias in our study.

In conclusion, over a 7-year study period, diabetic men and women in Taiwan were observed to have modestly increased risks of malignant neoplasms of liver, but the statistical significance of the association between diabetes and biliary tract cancer was lost after adjustment of various known clinical risk factors for biliary tract cancer. Additionally, compared with control subjects without any clinical risk factors, accompanying cirrhosis had the highest risk of liver neoplasm in diabetic patients, whereas those with cholangitis had the highest risk of biliary tract neoplasm. Awareness of the relationship between diabetes and hepatic neoplasm is essential to diabetic patients, especially those with known clinical risk factors for hepatic neoplasm, and these patients should be well educated for strict adherence to a liver cancer screening program.

References

  1. Top of page
  2. Abstract
  3. Subjects and Methods
  4. Results
  5. Discussion
  6. References