Risk of liver cirrhosis in patients with tuberculosis: a nationwide cohort study.

(1)

Risk of liver cirrhosis in patients with tuberculosis:

a nationwide cohort study

Yen-Chun Peng

^*,†

, Cheng-Li Lin

^‡

, Wan-Yun Hsu

^§

, Chi-Sen Chang

^*

, Hong-Zen Yeh

^*,†

and Chia-Hung Kao

^¶,**

Introduction

Worldwide, chronic infections cause serious problems for healthcare systems. Both liver cirrhosis and tuberculosis (TB) are chronic inflammatory diseases with dismal outcomes.

Because Taiwan is a hepatitis endemic area, the interaction of liver cirrhosis with TB is intriguing. Hepatitis B virus (HBV) infection is endemic in Asia, Africa, Southern Europe and the Pacific Islands [1]. Hepatitis C virus (HCV) infection is the other global health burden that affects about 160–170 million people worldwide. The natural courses of HBV and HCV infections are the sequences of chronic hepatitis, cirrhosis and hepatocellular carcinoma (HCC). Both of HBV and HCV infections are primary causes of liver cirrhosis and an increasing incidence of

HCC [2,3]. Patients with TB are infected by the Mycobacterium tuberculosis pathogen. After the primary infections, most of the patients infected with M. tuberculosis have latent and asymptomatic infections. The risk of active disease status is approximately

5% [4,5].

Data on the interaction of TB, chronic hepatitis and cirrhosis are limited. Most of the previous reports focused on TB in patients with cirrhosis [6–11]. Recent reports focused on drug toxicities in the treatment of TB in patients with cirrhosis [12,13]. The liver disease linking TB and cirrhosis is primary biliary cirrhosis (PBC). These studies investigated its association with antibodies related to TB infection. It has been suggested that the disease may be related to infectious agents.

Previous studies revealed that sera from patients with PBC

reacted against mycobacterium [6,14].

(2)

The pathogenesis of viral hepatitis related to cirrhosis is

unique. The effects of TB on cirrhosis have not been thoroughly

investigated. Moreover, several questions regarding the chronic inflammatory course of cirrhosis and TB have not been

addressed: Is the TB inflammation process enhanced in the presence of liver cirrhosis? Does co-infection with chronic liver inflammation and TB enhance liver cirrhosis? Do anti-TB regimens affect the pathogenesis of liver cirrhosis?

To address these questions, we conducted a nationwide cohort study to analyse the risk of developing liver cirrhosis among the patients with TB in Taiwan.

Materials and methods

The National Health Insurance (NHI) programme was implemented in Taiwan in March 1995. This universal programme

covers approximately 99% of the 23_74 million people living in Taiwan, and it involves contracts with 97% of hospitals and clinics in Taiwan [15]. The National Health Insurance Research Database (NHIRD), a large database derived from the Bureau of National Health Insurance programme, is maintained by the National Health Research Institutes (NHRI). Data are encrypted and depersonalized before release for public access. The Longitudinal Health Insurance Database 2000 (LHID2000) contains

all of the original claims data of 1 million beneficiaries randomly sampled from the 2000 Registry for Beneficiaries of the

NHIRD, which contains the registration data of every NHI programme beneficiary between 1996 and 2000. According to the NHRI in Taiwan, there are no significant differences in sex, age or healthcare costs between data in the LHID2000 and those in the original NHIRD. The NHRI followed all of the aforementioned randomly sampled patients up to 2011. The International

Classification of Disease, 9th Revision, Clinical

Modification (ICD-9-CM) was used to identify patients’ health status. This study was exempted by the Institutional Review Board (CMU-REC-101-012).

Sampled participants

In this retrospective nationwide population-based cohort study,

we identified patients over 20 years of age with newly diagnosed

TB (ICD-9 Codes 011–012) from both outpatient and

(3)

inpatient claim files between 1 January 1998 and 31 December 2010 to serve as the TB cohort. The date of the first medical visit for TB was identified as the index date. Patients with a history of liver cirrhosis (ICD-9 Codes 571.2, 571.5 and 571.6) before the index date or with incomplete age or sex information were excluded. Patients without TB were randomly selected from the LHID2000 to serve as the controls during the same study period and were frequency matched to the TB cohort at a ratio of 1 : 4 for age (every 5 year), sex and index year. Controls were subjected to the same exclusion criteria as the TB cohort. The

records of comorbidities were obtained before the index date for each participant. Comorbidities included hepatitis B (ICD-9 Codes V02.61, 070.20, 070.22, 070.30 and 070.32), hepatitis C (ICD-9 Codes V02.62, 070.41, 070.44, 070.51 and 070.54), diabetes (ICD-9 Code 250), chronic renal disease (ICD-9 Code 585),

chronic obstructive pulmonary disease (COPD) (ICD-9 Codes 490–496) and cancer (ICD-9 Codes 140–208).

Main outcome

Each study participant was followed from the index date until diagnosis of liver cirrhosis, loss of follow-up, death or withdrawal from the NHI programme or until 31 December 2011

(whichever occurred first).

Statistical analysis

Distributions of demographic characteristics including sex, age, monthly income, urbanization level and comorbidities were compared between the TB and the general population using the chi-square test for categorical variables and the t-test for continuous variables. Furthermore, we estimated sex-, age-,monthly

income-, urbanization level- and comorbidity-specific liver cirrhosis incidence rate by population person-years for the two

cohorts. Univariable and multivariable Cox proportional hazards regression models were used to assess the influence of TB on liver cirrhosis risk based on the hazard ratio (HR) and 95%

confidence intervals (CIs). Adjustment of the multivariable

model to control for age, sex, monthly income, urbanization level

and comorbidities revealed a significant difference (Table 1). The

cumulative incidence of liver cirrhosis was estimated using the

Kaplan–Meier method, and a log-rank test was used to compare

(4)

the survival curves of the TB and control. All analyses were performed using

^SAS

System version 9.3 for Windows (SAS Institute Inc., Cary, NC, USA), and the results were considered statistically significant if the two-tailed P-values were < 0_05.

Results

Demographic data

Among the records of claims data from 1 January 2000 to 31 December 2010, we identified 9339 patients with TB (TB cohort) and 37 356 frequency-matched participants without TB (general population). More than half (67_1%) were men, and the average age at baseline was approximately 61 years.

The two cohorts exhibited similar distributions of age and sex.

The distribution of monthly income and urbanization level varied, although the differences were nonsignificant. Compared with the general population, patients in the TB cohort

exhibited a higher prevalence of comorbidities in patients with hepatitis B, hepatitis C, diabetes, chronic renal disease, COPD and cancer (P < 0_001). The mean follow-up time was

5_27 years [standard deviation (SD) = 3_56] for the TB cohort and 6_18 years (SD = 3_24) for the general population (data not shown) Incidences of liver cirrhosis in tuberculosis and general population

The liver cirrhosis cumulative incidence curve in Fig. 1 shows that the TB cohort had a significantly higher risk of liver cirrhosis compared with the general population (log-rank test,

P < 0_001). During the 49 299 (230 867) person-years of followup for the TB (non-TB) cohort, the overall incidence of liver

cirrhosis was significantly higher in the TB cohort than in the general population (3_83 vs. 2_02 per 1000 person-year; crude HR =1_88; 95% CI = 1_59–2_23; P < 0_001), with an adjusted HR of 1_79 (95% CI = 1_50–2_14; P < 0_001) (Table 2).

Risk of liver cirrhosis in tuberculosis and general population

In the gender subgroup analysis, liver cirrhosis incidence was

higher in men than in women in both cohorts. The relative

adjusted HR of liver cirrhosis was 2_08 (95% CI = 1_71–2_52) in

men. The age-specific incidence of liver cirrhosis increased with

age in both cohorts. TB patients younger than 49 years of age

(5)

exhibited a 3_49-fold increased risk of liver cirrhosis compared with that of the general population (95% CI = 2_38–5_11;

P < 0_001). In the TB patients, the relative adjusted HRs of liver cirrhosis for all monthly income groups were significantly higher thanthoseof thepatients withoutTB(P < 0_001).The urbanization level-specific incidence of liver cirrhosis increasedwithdecreasing urbanization level in both cohorts.TheTBpatients with or without comorbidities were associated with significantly higher risk of liver cirrhosis compared with the non-TB patients (P < 0_001).

Univariable and multivariable analysis

Table 3 shows the results of univariable and multivariable Cox proportional hazard analyses for the association between liver cirrhosis and TB. The risk of developing liver cirrhosis was 1_81-fold higher for men than for women (95% CI = 1_50–2_19), and a 1_02-fold (95% CI = 1_01–1_02) increase was observed

with age (every 1 year). The lower monthly income group exhibited an adjusted HR of 1_26 for liver cirrhosis compared

with those with a higher monthly income (95% CI = 1_02–1_57).

The adjusted HR for liver cirrhosis was 1_55 (95% CI = 1_24–

1_96) for those living in the lowest urbanization level compared to those living in areas with the highest urbanization level. The risk for liver cirrhosis was higher for patients with a comorbidity of hepatitis C (adjusted HR = 5_65; 95% CI = 4_27–7_47), hepatitis B (adjusted HR = 2_52; 95% CI = 1_84–3_46), chronic renal disease (adjusted HR = 1_84; 95% CI = 1_30–2_61) and diabetes (adjusted HR = 1_61; 95% CI = 1_35–1_92).

Trends of liver cirrhosis by stratified follow-up years We calculated the incidence and adjusted HRs according to the length of the follow-up after TB was diagnosed (Table 4). The incidence rates of liver cirrhosis were significantly higher in the first 2 years for both the TB

cohort (5_36 per 1000 person-year) and the general population (2_10 per 1000 person-year) (P < 0_001). The adjusted HR for liver cirrhosis increased significantly with follow-up

duration. Tuberculosis and the risk of liver cirrhosis in hepatitis B or hepatitis C infection

Table 5 shows the risk of liver cirrhosis in TB patients with

hepatitis B or hepatitis C infection as determined by Cox proportional

(6)

hazard regression. In the various categories of hepatitis B or hepatitis C, TB increased the risk for cirrhosis in patients with either hepatitis B (adjusted HR = 1_91; 95%

CI = 1_05–3_47) or hepatitis C (adjusted HR = 2_56; 95%

CI = 1_37–4_78).

Discussion

After adjusting for confounding factors, we observed an adjusted HR of 2_52 for liver cirrhosis in patients with hepatitis B and 5_65 for patients with hepatitis C. Our investigation provides crucial evidence from a large database with essentially no loss to follow-up. In the database, TB and liver cirrhosis are registered based on clinical evidence and follow-up, and TB treatment is mandatory. Our findings show that a combination of TB infection and chronic hepatitis resulted in increased risk of cirrhosis in Taiwan.

Most susceptible hosts with primary HBV infection are asymptomatic, particularly young children. Persistent HBV infection may lead to continuous viral replication, continuous viremia and histological liver injury. Liver cirrhosis is a condition of regeneration nodules and fibrosis that develops in

approximately 20% of patients with chronic hepatitis B infection [2,16]. Treating viral hepatitis may reverse the process of cirrhosis and HCC. Chronic hepatitis B patients receiving nucleotide analogue therapy exhibited a significantly lower incidence of HCC compared with untreated patients [17]. In HCV infection, when chronic infection is established, spontaneous viral clearance is rare, and the HCV continually damages

liver tissue [3,18]. Both chronic HBV and HCV infections are

associated with an increased risk for liver cirrhosis and subsequent HCC [2,3,16]. In contrast to treatment for reversing liver

cirrhosis or fibrosis, a combined infection with TB increased the risk of liver cirrhosis in patients with HBV or HCV infection.

Liver cirrhosis and fibrosis involve a dynamic natural course and are potentially reversible [19–21]. The host response to TB infection relies on the host, bacterial factors and how these

factors interact [22]. Persistent and recurrent liver injury progressing to fibrosis is a complex process involving several factors,

such as host immunity and genetics, viral factors and drug

(7)

treatment [2,23]. HBV is not directly cytotoxic to hepatocytes, and host immune responses to viral antigens expressed on

infected hepatocytes are the principal determinants of hepatocellular injury [2]. HCV is also not directly cytopathic to hepatocytes,

and it is mainly related to immune-mediated

mechanisms characterized by T-cell responses [3]. The findings of this study reveal a more significant cirrhosis risk in TB patients with hepatitis C than in those with hepatitis B.

Tuberculosis leads to an increase of HIV replication and

accelerates progression of HIV infection, possibly resulting in a similar situation in which TB increases the risk of cirrhosis in HBV and HCV infections [5]. Combined bacterial and viral infections, particularly in patients with chronic inflammatory status, may alter immunity activation, resulting in persistent liver injury and disease progression.

As anti-TB treatment is essential, hepatic toxicity by antimicrobial agents is a potential issue that needs to be addressed

[5,13]. Patients with diagnosed and registered TB are required by law to receive anti-TB treatment. Liver fibrosis and cirrhosis are reversible, and thus, chemoprevention is a possible modality that

can be applied to treat chronic HBV and HCV infection. In contrast to the role of antiviral treatment, anti-TB treatment

increases the risk of liver injury and may exacerbate liver inflammation and fibrosis, resulting in increased risk of liver cirrhosis.

Tuberculosis is associated with PBC [14,24], which is a progressive cholestatic liver disease characterized by immunemediated destruction of biliary epithelial cells in small intrahepatic bile ducts. It has been hypothesized that mycobacteria are related to PBC because mycobacteria infection is involved in granuloma formation. The autoantibody anti-mitochondrial antibody has also been detected in patients with TB, but this antibody is nonspecific. Because infection with TB may trigger PBC, the mode of interaction between TB and cirrhosis warrants further investigation.

There were some limitations in this study. First, several unmeasured confounding factors, such as body mass index,

smoking status, alcohol intake and drug history, which are associated

with cirrhosis, are not included in the LHID2000. Second,

(8)

anti-TB medications as well as antiviral medications for treating chronic hepatitis are critical factors in the pathogenesis of cirrhosis, and these factors were not examined in this study.We

could not determine the distribution of health factors and antiviral treatment in TB patients or in the control, and thus, we did not adjust the regression model to account for the effects of these factors on the development of cirrhosis. However, these factors may be considered equal in both cohorts because of the large number of cases and the treatment guidelines. Third, because our study employed a retrospective cohort design involving patients with TB and HBV or HCV infection, the data may have been subject to certain selection biases. Fourth,we observed significant differences in comorbidities between TB patients and the general population. The TB cohort exhibited a higher frequency of comorbidities, which might also be a crucial confounding factor

between both cohorts. Fifth, the diagnosis of liver cirrhosis was based on ultrasound or CT scan images, combined with clinical evidence of cirrhosis, including laboratory data (bilirubin, PT, albumin), esophageogastric varices, ascites and encephalopathy.

Liver cirrhosis diagnosed by histology from liver biopsy is not essential in the nationwide database.