Authors: Shuang Zhang, Ningning Dong, Lu Wang, Yu Lu, Xiaoyou Chen
Categories: Original Research, anti-tuberculosis drugs, drug-induced liver injury, hepatotoxicity, severe liver injury, risk factors
Source: Infection and Drug Resistance
Doi: 10.2147/IDR.S519211
Authors: Shuang Zhang, Ningning Dong, Lu Wang, Yu Lu, Xiaoyou Chen
Anti-tuberculosis drug-induced liver injury (ATB-DILI) is a common adverse reaction associated with tuberculosis (TB) treatment, significantly impacting treatment adherence and therapeutic outcomes. However, large-scale studies on hospitalized patients in China remain limited.
To characterize the clinical features and liver injury patterns in hospitalized TB patients with ATB-DILI and to identify risk factors associated with severe ATB-DILI.
We retrospectively reviewed 28,753 hospitalized TB patients at Beijing Chest Hospital from 2014 to 2023. ATB-DILI was diagnosed in 567 patients (2.0%) based on serum biochemical criteria and causality assessment. Demographic, clinical, and laboratory data were analyzed to characterize liver injury types and identify risk factors for severe cases. Subgroup analyses based on liver injury patterns were performed to further evaluate the association between age and severe ATB-DILI.
Overall, 567 cases with ATB-DILI (2.0%) were analyzed. Hepatocellular injury was the most common type (71.4%), followed by cholestatic (13.8%) and mixed (14.8%) injury patterns. Most patients (68.4%) were asymptomatic and diagnosed via routine biochemical monitoring; jaundice occurred in 18.2%. Patients with hepatocellular damage were significantly younger, while those with cholestatic injury were older (p < 0.001). Severe ATB-DILI occurred in 46 patients (8.1%), with advanced age (≥60 years) identified as an independent risk factor (OR = 2.45, 95% CI: 1.33–4.52, p = 0.004). Subgroup analysis showed that this association between age and severe ATB-DILI was significant in the hepatocellular injury type (unadjusted OR = 3.59, 95% CI: 1.61–8.02, p = 0.002), while no statistically significant association was observed in cholestatic or mixed types, which may reflect limited statistical power in these subgroups.
Routine liver function monitoring and age-specific risk assessment are essential for early identification and management of ATB-DILI in hospitalized TB patients.
Tuberculosis (TB) is a major global public health issue and ranks as the first deadliest infectious disease. According to the World Health Organization (WHO) Global Tuberculosis Report 2024,1 approximately 10.8 million people are expected to develop TB worldwide, with 1.25 million deaths attributed to the disease. As one of the countries with the highest TB burden, China accounts for approximately 6.8% of the new global TB cases and faces significant challenges, including high drug resistance rates, poor treatment adherence, and the presence of co-infections.
Anti-tuberculosis drug-induced liver injury (ATB-DILI) is one of the most common adverse events associated with TB treatment. The reported prevalence of ATB-DILI varies considerably, ranging from 2% to 28%, depending on diagnostic criteria and geographic region.2–4 One study demonstrated that ATB-DILI occurring during active TB treatment increases the risk of treatment failure and relapse by threefold.5 Among first-line anti-TB drugs, isoniazid, rifampicin, and pyrazinamide are the principal contributors to hepatotoxicity, while second-line drugs also play a role in hepatic adverse events.6,7 In Western countries, drug-induced liver injury (DILI) is commonly associated with antibiotics, cardiovascular agents, and non-steroidal anti-inflammatory drugs (NSAIDs). In contrast, anti-TB drugs are the predominant cause of DILI in many Eastern countries, likely reflecting differences in pharmacogenetic and immunological profiles.8 Furthermore, developing countries, which often lack robust pharmacovigilance systems and have patient populations with poorer baseline health status, report even higher rates of ATB-DILI. Despite its clinical significance, comprehensive and large-scale epidemiological data on ATB-DILI remain limited.
Recent studies have reported an increasing global trend in ATB-DILI incidence, rising from 5.1% in 1999 to 29.4% in 2020.9 In China, the number of reported DILI cases increased by 62% between 2012 and 2016, with anti-TB drugs being the most frequently identified cause.10 A single-center study in China reported that the incidence of drug-related liver injury increased from 2.21% in 2002 to 6.70% in 2022.11 These findings highlight the urgent need for enhanced drug safety surveillance and standardized diagnostic criteria for ATB-DILI. However, existing studies in China often suffer from limitations such as small sample sizes, inconsistent diagnostic approaches, and a lack of detailed investigations into liver injury types and their severity. To address these gaps, we conducted a retrospective analysis of hospitalized TB patients at Beijing Chest Hospital, Capital Medical University, from 2014 to 2023. This study aimed to evaluate the clinical characteristics of different ATB-DILI types and identified risk factors associated with severe ATB-DILI. To increase the likelihood that severe liver injury events were anti-TB drugs related, we excluded patients with preexisting liver and biliary tract diseases, as well as other potential causes of acute liver injury. The findings from this study provide novel insights into the prevention, monitoring, and management of ATB-DILI, particularly in high-burden TB settings like China.
This study retrospectively analyzed the clinical data of patients with active TB who were hospitalized at Beijing Chest Hospital of Capital Medical University between January 2014 and December 2023. Beijing Chest Hospital is a tertiary institution specializing in tuberculosis care in China. Among a total of 28,753 hospitalized TB patients, 567 cases (2.0%) were confirmed as ATB-DILI and were included in the final analysis. We extracted data from electronic health records, which included demographic characteristics, inpatient diagnoses, medical histories, laboratory results, and medical treatment information. The study adhered to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines.12 Ethical approval was granted by the Institutional Review Board of Beijing Chest Hospital, and the requirement for informed consent was waived by the same committee.
Inclusion criteria for patients with active TB: (1) no restrictions on age or gender; (2) no restrictions on lesion sites, including pulmonary and extrapulmonary sites; (3) diagnosis of TB by clinical or laboratory criteria. Exclusion (1) patients with malignant tumors who are receiving concurrent radiotherapy, chemotherapy, or immunotherapy; (2) co-infection with non-TB mycobacteria; (3) co-infection with HIV; (4) pregnancy; (5) ICU patients were also excluded due to their complex clinical conditions and polypharmacy, which may confound the attribution of liver injury to anti-TB drugs. Patients with repeat hospitalizations were only included in the hospitalization where peak liver function was observed.
The diagnostic criteria were based on the Chinese guidelines for ATB-DILI and implemented through a two-step 13 First, liver injury was identified according to biochemical criteria, meeting either of the (1) alanine aminotransferase (ALT) ≥ 3 × upper limit of normal (ULN) and/or total bilirubin (TBIL) ≥ 2 × ULN; or (2) concurrent elevation of aspartate aminotransferase (AST), alkaline phosphatase (ALP), and TBIL, with at least one value ≥ 2 × ULN. Abnormal liver function is defined as an elevation of liver enzymes that does not meet the above criteria. The ULN values used in our laboratory are 40 IU/L for ALT, 40 IU/L for AST, 150 IU/L for ALP, and 20.5 μmol/L for TBIL. Second, to confirm anti-tuberculosis drugs as the causative agent, we applied the updated Roussel Uclaf Causality Assessment Method (RUCAM), a validated tool for DILI attribution.14,15 Patients with a RUCAM score of ≥6 were included in the study. To enhance diagnostic specificity, we rigorously excluded cases (1) cirrhosis, such as alcoholic cirrhosis, hepatitis B cirrhosis, and hepatitis C cirrhosis; (2) liver transplantation status or liver abscess; (3) intra- and extra-hepatic bile duct obstruction, such as biliary stones, bile duct stenosis, and bile duct obstruction or compression; (4) recent history of hypotension, right heart failure, septicemia, and total parenteral nutrition; (5) liver injury from non-anti-TB drugs, such as herbal medicines, cold remedies, NSAIDs; and (6) poorly documented medication history.
According to the R value [R value = (serum ALT/ALT ULN)/ (serum ALP/ALP ULN)].16,17 Cases were classified as “hepatocellular” if the R value was ≥5, “cholestatic” if the R value was ≤2, and “mixed” if the R value was 2–5.
Severe liver injury is defined based on the peak liver function index meeting any of the following 18 (1) elevated serum ALT or AST >3 × ULN, serum TBIL >2 × ULN, and serum ALP <2 × ULN; and (2) international normalized ratio (INR) ≥1.5, with TBIL levels >2 × ULN. Both definitions represent severe acute liver injury and are used by the US Food and Drug Administration’s Sentinel System to assess drug-induced acute liver injury in post-marketing clinical settings.19 Criterion (1) follows Hay’s Law biochemical standards,20,21 indicating hepatocellular injury severe enough to affect bilirubin excretion and increase the risk of mortality.22,23 Criterion (2) highlights liver function abnormalities that may occur in the late stages of acute liver failure, where hepatic transaminases may not yet be elevated enough to meet the first criterion. The potential preemptive use of anticoagulants should be excluded when assessing criterion (2).
Eosinophilia is defined as a serum eosinophil count exceeding 4% to 6% of the total white blood cell count, depending on the normal reference range for each hospital. In our laboratory, the upper normal limit is 5%. Lymphopenia is defined as serum lymphocyte levels <10%, both based on blood tests conducted when diagnosing ATB-DILI. The rash is an acute skin condition involving skin texture or color changes, which may present as inflammation or hypersensitivity. Allergy at the time of ATB-DILI diagnosis is identified if any of the following symptoms are drug fever, rash, eosinophilia, lymphopenia, or arthralgia.24
Risk factors investigated for severe ATB-DILI included age, gender, history of alcohol consumption, smoking history, comorbidities (such as diabetes and hypertension), allergic reactions, time from drug exposure to DILI recognition, type of liver injury, use of prophylactic hepatoprotective agents, and prior history of ATB-DILI. The interval between drug exposure and DILI onset was defined as the number of days from the initiation of anti-TB therapy (day 0) to the appearance of the first clinical symptom, physical sign, or abnormal laboratory finding indicative of liver injury.15
Data analysis was performed using SPSS version 29.0 software. Quantitative data are presented as either the mean ± standard deviation (Mean ± SD) or median with interquartile range (Median [IQR]), based on the data distribution. The independent samples t-test or Mann–Whitney U-test was applied appropriately for comparisons between the two groups. For comparisons involving multiple groups, one-way analysis of variance (ANOVA) or the Kruskal–Wallis test was used. Categorical variables are presented as frequencies and percentages, with group differences assessed using the chi-square test (χ²) or Fisher’s exact test. Variables with p < 0.1 in univariate analysis were included in a multivariate logistic regression model, with stepwise selection for variable inclusion. A two-tailed p-value of less than 0.05 was considered statistically significant. Post hoc power analysis was conducted using PASS 2021 based on the observed odds ratios and actual sample sizes. Furthermore, subgroup analyses were performed to evaluate the association between age and severe ATB-DILI within each liver injury subtype (hepatocellular, cholestatic, and mixed) using univariate logistic regression models. To address multiple comparisons in subgroup analyses, Bonferroni correction was applied (adjusted α=0.0167).
A total of 28,753 hospitalized patients with active TB who received anti-TB treatment were initially assessed for eligibility. Among them, 3233 patients (11.2%) exhibited abnormal liver function based on blood biochemical tests, and 930 (3.2%) were clinically diagnosed with liver injury. After excluding 363 cases with alternative causes of liver injury, as determined by the RUCAM criteria, 567 patients (2.0% of the total population) were confirmed or highly suspected to have ATB-DILI. These 567 cases constituted the final sample included in the present study. The detailed inclusion and exclusion process is illustrated in Figure 1. Figure 1Patient inclusion and exclusion flowchart for ATB-DILI.
The distribution of anti-TB regimens among the 567 ATB-DILI patients, categorized by liver injury pattern, severity, and allergic manifestations, is presented in Table 1. The median age was 44 years (IQR: 28.0–62.0), with a predominance of male patients (356, 62.8%). Most patients (67.0%) received prophylactic hepatoprotective agents. Comorbid conditions included diabetes mellitus (15.0%) and hypertension (14.1%). A prior history of ATB-DILI was documented in 7.2% of patients, and 6.9% reported previous allergic reactions to anti-TB drugs. Most patients (85.5%) were undergoing initial treatment, while 14.5% were receiving retreatment. Regarding TB classification, pulmonary TB was diagnosed in 49.4%, extrapulmonary TB in 21.0%, and both forms in 29.6% of patients. The median latency from drug exposure to liver injury onset was 20 days (IQR: 11.0–36.0). Clinical manifestations included gastrointestinal symptoms (35.1%), jaundice (18.2%), drug rash (9.5%), drug fever (9.5%), and other discomforts (3.9%). Notably, 68.4% (388 patients) were asymptomatic and were identified solely through abnormal liver biochemical markers during routine monitoring (Tables 2 and 3).Table 1Distribution of the Anti-TB Regimens in 567 Cases of ATB-DILI According to the Type of Liver Damage, Severity of Hepatic Injury, and Presence of AllergyDrug RegimensTotal Cases(N)Type of Liver Injury (N)Allergy(N)Severe Liver Injury(N)HepatocellularCholestaticMixedHRZE24520314289614HRZE+ Lfx554357194HRZE + Mfx311876103HR ± FQs57261714239HRftZE ± FQs39257793HRftE ± FQs258143134HZ ± FQs14112140RZ ± FQs10100071H ± FQs431030Includes Z, not HR531121Other regimens include second-line drugs8256917357Notes: The HRZE group includes 11 patients who received the HRZE fixed-dose combination (FDC), a single-tablet formulation containing isoniazid (H), rifampicin (R), pyrazinamide (Z), and ethambutol (E).Abbreviations: H, isoniazid; R, rifampicin Z, pyrazinamide; E, ethambutol; Lfx, levofloxacin; Mfx, moxifloxacin; FQs, fluoroquinolones, including levofloxacin and moxifloxacin; Rft, rifapentine. Table 2Characteristics of Subjects With Different Patterns of ATB-DILIEntire Cohort(N=567)Hepatocellular(N= 406, 71.6%)Cholestatic(N= 77, 13.6%)Mixed(N=84, 14.8%)p-valueAge, y, IQR44.0 (28.0–62.0)37.0 (26.0–57.0)66.0 (46.0–75.5)47.5 (29.0–66.8)<0.001Age group, y (%)<0.001 <30162 (28.6)134 (33.0)6 (7.8)22 (26.2) 30–59251 (44.3)191 (47.0)28 (36.4)32 (38.1) ≥60154 (27.2)81 (20.0)43 (55.8)30 (35.7)Male sex (%)356 (62.8)250 (61.6)49 (63.6)57 (67.9)0.548Location (%)0.715 Urban347 (61.2)252 (62.1)44 (57.1)51 (60.7) Rural220 (38.8)154 (37.9)33 (42.9)33 (39.3)Body mass index, kg/m², IQR20.7 (18.6–23.3)20.8 (18.6–23.2)20.1 (18.4–23.2)20 (18.7–23.7)0.455Prophylactic use of hepatoprotectants (%)380 (67.0)265 (65.3)57 (74.0)58 (69.0)0.305Previous history of ATB-DILI (%)41 (7.2)30 (7.4)6 (7.8)5 (6.0)0.880A history of allergy to anti-TB drugs (%)39 (6.9)34 (8.4)2 (2.6)3 (3.6)0.101Current smoking (%)148 (26.1)100 (24.6)25 (32.5)23 (27.4)0.342Alcohol use (%)73 (12.9)50 (12.3)12 (15.6)11 (13.1)0.733Type of treatment (%)0.069 Initial treatment485 (85.5)356 (87.7)62 (80.5)67 (79.8) Re-treatment82 (14.5)50 (12.3)15 (19.5)17 (20.2)Comorbidities (%) Diabetes mellitus85 (15.0)53 (13.1)17 (22.1)15 (17.9)0.092 Renal Insufficiency14 (2.5)5 (1.2)4 (5.2)5 (6.0)0.006 Hypertension80 (14.1)39 (9.6)23 (29.9)18 (21.4)<0.001 Cardiovascular disease45 (7.9)21 (5.2)10 (13.0)14 (16.7)<0.001 Fatty Liver Disease^a^63 (11.1)47 (11.6)8 (10.4)8 (9.5)0.842 Biliary tract disease^b^25 (4.4)15 (3.7)6 (7.8)4 (4.8)0.229Current TB (%)0.331 Pulmonary TB280 (49.4)209 (51.5)35 (45.5)36 (42.9) Extrapulmonary TB119 (21.0)79 (19.5)16 (20.8)24 (28.6) Both pulmonary and extrapulmonary TB168 (29.6)118 (29.1)26 (33.8)24 (28.6)Notes: a and b are both patients with abnormal findings by imaging (ultrasound or abdominal CT) during hospitalization for patients with fatty liver and biliary tract disease with normal baseline liver function. Table 3Clinical Pictures and Laboratory Tests With Different Patterns of ATB-DILIEntire Cohort(N=567)Hepatocellular(N=406, 71.6%)Cholestatic(N=77, 13.6%)Mixed(N=84, 14.8%)p-valueDays between exposure and DILI recognition, d, IQR20.0 (11.0–36.0)20.0 (11.0–36.0)20.0 (9.5–41.5)17.0 (9.3–36.8)0.679Clinical manifestations (%) Digestive Tract Symptoms199 (35.1)135 (33.3)37 (48.1)27 (32.1)0.037 Drug Rash54 (9.5)39 (9.6)7 (9.0)8 (9.5)0.990 Drug Fever54 (9.5)41 (10.1)3 (3.9)10 (11.9)0.154 Jaundice103 (18.2)29 (7.1)54 (70.1)20 (23.8)<0.001 Other discomforts22 (3.9)13 (3.2)7 (9.1)2 (2.4)0.056 No complaints388 (68.4)316 (77.8)19 (24.7)53 (63.1)<0.001Blood routine examination WBC, ×10^9^/L, IQR5.8 (4.7–7.7)5.6 (4.6–7.0)7.3 (5.3–1.2)6.5 (4.9–8.7)<0.001 HB, g/L, IQR125.0 (112.0–139.0)129.0 (117.0–141.3)109.0 (96.0–125.0)123.0 (105.0–133.8)<0.001 PLT, ×10^9^/L, IQR226.0 (180.0–296.0)223.5 (181.8–286.5)216.0 (158.5–306.5)242.5 (181.5–368.8)0.089 Absolute eosinophils, ×10^9^/L, IQR0.13 (0.06–0.26)0.14 (0.07–0.28)0.06 (0.01–0.12)0.14 (0.06–0.25)<0.001Biochemical indicators, peak values, IQR ALT, U/L, IQR186.0 (135.0–306.0)229.5 (157.0–391.0)37.0 (17.0–78.5)144 (122.3–176.5)<0.001 AST, U/L, IQR174.0 (103.0–314.0)200 (119.5–348.5)88.5 (35.5–162.8)145.5 (77.8–250.3)<0.001 ALP, U/L, IQR96.0 (75.0–137.0)88.0 (71.0–107.0)162.0 (94.5–288.0)136.0 (112.0–184.0)<0.001 GGT, U/L, IQR74.4 (45.8–129.4)66.0 (43.1–102.8)126.6 (53.4–310.0)117.9 (68.3–208.0)<0.001 TBIL, μmol/L, IQR14.6 (9.5–29.0)12.8 (8.9–18.5)48.1 (36.0–58.6)16.1 (9.1–40.2)<0.001 TBA, μmol/L, IQR9.3 (4.7–21.7)8.3 (4.4–17.3)20 (7.2–77.9)11.2 (5.6–30.2)<0.001 Albumin, g/L, IQR36 (31.5–39.5)37.4 (33.2–40.6)29.0 (26.0–33.2)34.1 (29.2–38.0)<0.001 Serum Iron, μmol/L, IQR11.2 (6.8–16.5)11.9(7.8–16.6)8.7 (5.3–14.2)9.8 (5.4–16.9)0.015 INR, IQR1.1 (1.0–1.2)1.1(1.0–1.2)1.1 (1.0–1.3)1.0 (1.0–1.1)0.037 PTA, %, IQR78.3 (64.6–88.6)78.8 (65.0–89.6)75.3 (59.0–85.5)80.2 (65.9–89.6)0.375Fits Hy’s law (%)46 (8.1)27 (6.7)11 (14.3)8 (9.5)0.070Acute liver failure (%)8 (1.4)7 (1.7)0 (0)1 (1.2)0.849Severe liver injury (%)46 (8.1)27 (6.7)11 (14.3)8 (9.5)0.070Notes: Gastrointestinal symptoms include abdominal discomfort, nausea, vomiting, poor appetite, and diarrhea; Other discomforts include malaise, fatigue, joint pain, and bleeding spots on the skin.Abbreviations: WBC, white blood cell; HB, hemoglobin; PLT, platelet; GGT, γ-glutamyl transferase; TBA, total bile acids; PTA, prothrombin time activity.
As presented in Tables 2 and 3, the hepatocellular type of liver injury was the most prevalent, accounting for 71.4% of cases, while cholestatic and mixed types comprised 13.8% and 14.8%, respectively. Patients with hepatocellular injury were significantly younger (p < 0.001) and less likely to exhibit clinical jaundice, with the majority (77.8%) being asymptomatic. In contrast, patients with cholestatic liver injury were older and more likely to present with pronounced jaundice (69.2%) and prominent gastrointestinal symptoms. The mixed liver injury type demonstrated intermediate characteristics between hepatocellular and cholestatic injuries in terms of patient age. No statistically significant differences were observed among the liver injury types in terms of gender, body mass index, prophylactic hepatoprotective use, incubation period, smoking, alcohol consumption, types of TB or severity of liver injury (all p > 0.05). Laboratory analysis revealed that patients with hepatocellular injury had significantly higher levels of ALT, albumin, and serum iron (all p < 0.05), whereas patients with cholestatic or mixed injury exhibited higher levels of ALP, GGT, TBIL, and TBA (all p < 0.001).
Severe liver injury in this study was evaluated based on Hy’s law and diagnostic criteria for acute liver failure. The findings revealed that Hy’s law was effective in identifying patients with acute liver failure. Among the 567 patients diagnosed with ATB-DILI, 46 (8.1%) were identified as having severe liver injury. The median age of patients with severe liver injury was 58 years (IQR: 39.0–70.3), which was significantly older than the median age of 43 years (IQR: 28.0–60.0) for patients with non-severe liver injury (p < 0.001). The severe liver injury group exhibited higher proportions of cholestatic and mixed liver injury types compared to the hepatocellular type (p = 0.043). No significant differences (p > 0.05) were observed between severe and non-severe ATB-DILI groups in terms of gender, urban versus rural residency, use of prophylactic hepatoprotective drugs, prior history of ATB-DILI, smoking history, alcohol consumption, or types of TB (Table 4).Table 4Demographic Characteristics, Clinical and Laboratory Findings With Non-Severe ATB-DILI Versus Patients With Severe ATB-DILIEntire Cohort(N=567)Non-Severe ATB-DILI(N= 521, 91.9%)Severe ATB-DILI(N=46, 8.1%)p-valueAge, y, IQR44.0 (28.0–62.0)43.0 (28.0–60.0)58.0(39.0–70.3)<0.001Male sex (%)356 (62.8)329 (63.1)27 (58.7)0.549Location (%)0.320 Urban347 (61.2)322 (61.8)25 (54.3) Rural220 (38.8)199 (38.2)21 (45.7)Body mass index, kg/m², IQR20.7 (18.6–23.3)20.6 (18.6–23.2)21.6 (18.4–24.1)0.196Prophylactic use of hepatoprotectants (%)380 (67.0)350 (67.2)30 (65.2)0.786Previous history of ATB-DILI (%)41 (7.2)37 (7.1)4 (8.7)0.565Current smoking (%)148 (26.1)132 (25.3)16 (34.8)0.162Alcohol use (%)73 (12.9)66 (12.7)7 (15.2)0.621Comorbidities (%) Diabetes mellitus85 (15)77 (14.8)8 (17.4)0.634 Hypertension80 (14.1)70 (13.4)10 (21.7)0.121 Fatty Liver Disease63 (11.1)55 (10.6)8 (17.4)0.157Days between exposure and DILI recognition, IQR20.0 (11.0–36.0)20.0 (10.0–36.0)28.0 (13.8–39.3)0.187Treatment type, (%)0.143 Initial treatment485 (85.5)449 (86.2)36 (78.3) Re-treatment82 (14.5)72 (13.8)10 (21.7)Jaundice (%)103 (18.2)57 (10.9)46 (100.0)<0.001Allergy (%)221 (39.0)200 (38.4)21 (45.7)0.333Pattern of liver injury (%)0.043 Hepatocellular damage406 (71.6)379 (72.7)27 (58.7) Cholestatic & Mixed damage161 (28.4)142 (27.3)19 (41.3)Current TB (%)0.422 Pulmonary TB280 (49.4)258 (49.5)22 (47.8) Extrapulmonary TB119 (21.0)112 (21.5)7 (15.2) Both pulmonary and extrapulmonary TB168 (29.6)151 (29.0)17 (37.0)Blood routine examination, IQR WBC, ×10^9^/L, IQR5.8 (4.7–7.7)5.8 (4.7–7.5)6.2 (4.5–8.9)0.603 HB, g/L, IQR125.0 (112.0–139.0)125.0 (112.0–139.0)123.0 (115.8–258.0)0.922 PLT, ×10^9^/L, IQR226.0 (180.0–296.0)229.0 (183.5–299.5)186.5 (131.8–258.0)0.003Biochemical indicators, peak values, IQR ALT, U/L, IQR187.0 (136.0–306.0)184.0 (137.5–297.0)272.0 (87.8–511.0)0.175 AST, U/L, IQR174.0 (103.0–314.0)168.0 (100.0–292.5)324.5 (168.0–675.5)<0.001 AST/ALT, IQR0.9 (0.6–1.4)0.9 (0.6–1.4)1.4 (0.8–2.7)<0.001 ALP, U/L, IQR96.0 (75.0–137.0)94.0 (74.0–129.5)131.5 (103.8–164.5)<0.001 TBIL, μmol/L, IQR14.6 (9.5–29.0)13.7 (9.2–23.0)55.4 (49.1–89.8)<0.001 TBA, μmol/L, IQR9.3 (4.7–21.7)8.5 (4.4–17.9)124.9 (32.2–234.8)<0.001 INR, IQR1.1 (1.0–1.2)1.0 (1.0–1.1)1.3 (1.2–1.5)<0.001 Albumin, g/L, IQR36 (31.5–39.5)36.5 (31.9–39.9)30.8 (28.2–35.9)<0.001 Serum Iron, μmol/L, IQR11.2 (6.8–16.5)10.7 (6.7–15.8)13.7 (9.2–24.4)0.008
Multivariable logistic regression analysis was performed on variables with P < 0.1 in the univariate analysis (including age ≥60 years and liver injury type). The analysis identified age ≥60 years as an independent risk factor for severe ATB-DILI (OR = 2.45, 95% CI: 1.33–4.52, p = 0.004). Post hoc power analysis for the logistic regression model indicated sufficient statistical power (81.8%) to detect the observed effect size (Table 5). We further assessed the relationship between age and severe ATB-DILI in different subgroups of liver injury types. Among patients with hepatocellular injury, those age ≥60 years had a significantly higher risk of severe ATB-DILI compared to those age <60 years (unadjusted OR = 3.59, 95% CI: 1.61–8.02, p = 0.002). In contrast, no statistically significant association was observed in patients with cholestatic or mixed liver injury patterns (Figure 2).Table 5Non-Biochemical Variables Independently Associated With Severe ATB-DILIIndependent VariablesCoefficientSEWald χ2OR (95% CI)p-valueAge ≥ 600.900.318.222.45 (1.33–4.52)0.004Note: Post hoc power was 81.8%.Abbreviations: CI, confidence interval; OR, odds ratio. Figure 2Effect of age on severe ATB-Dili according to different types of liver injury. Shown are subgroup-specific odds ratios for all the patients and for those who were hepatocellular, cholestatic, or mixed liver injury. Odds ratios are plotted as squares, the horizontal lines represent 95% confidence intervals. Bold p-values indicate statistical significance (p < 0.05). Post hoc power analysis indicated that the statistical power for detecting associations in the hepatocellular group was 89.7% (or 76.2% after Bonferroni correction), whereas the cholestatic and mixed groups had limited power (23.1% and 34.5%, respectively).
In this Chinese hospitalized cohort of 28,753 patients with active TB, 2.0% had ATB-DILI. However, this finding may not be directly generalizable to other populations due to differences in region, genetics, lifestyle, and medical practices. The observed result is lower than the cumulative incidence of ATB-DILI reported in a previous Chinese cohort study, which was 2.55% (95% CI: 2.04%-3.06%),25 and the 5.4% incidence found in another cohort study of 3155 patients.26 These differences may be attributed to variations in study design and methodology. In this study, a retrospective analysis combined with a rigorous review of electronic medical records and causality assessments ensured the inclusion of only liver injury cases occurring during hospitalization. Consequently, the overall incidence of ATB-DILI may have been underestimated. Furthermore, the study encompassed patients receiving both first-line and second-line anti-TB regimens, reflecting real-world medication practices and potentially mitigating drug-related biases. In comparison, international studies report varying ATB-DILI a single-center study in the UK found a prevalence of 6.9%,27 a study in Thailand reported 6.4%,28 and a tertiary hospital in South Korea documented an incidence of 11.9%.29 These discrepancies are likely due to differences in study populations, diagnostic criteria, and treatment protocols, highlighting the challenges of cross-study comparisons.
The diagnosis of DILI remains challenging, primarily relying on exclusionary criteria, which can result in misdiagnosis.30 This study addressed the limitations of traditional DILI reporting systems by systematically excluding non-drug-related causes of liver injury. A UK study previously demonstrated that nearly 50% of adverse liver reactions initially attributed to drugs were ultimately found to have non-drug-related causes upon thorough evaluation.31 Additionally, prescribing habits in China, including the widespread use of traditional Chinese medicine and herbal therapies, as well as high rates of self-medication, add complexity to DILI diagnosis. In this study, 70 patients were found to have liver injury caused by non-anti-TB drugs, while liver injury in another 102 patients occurred before anti-TB treatment, emphasizing the importance of careful evaluation.
Early recognition and management of DILI are critical for improving patient outcomes.32 Most ATB-DILI events (87%) occur within the first two months of anti-TB treatment when four first-line drugs are typically used in combination.27 Routine liver function monitoring every two weeks has been shown to detect ATB-DILI early and may also help predict late-onset DILI in patients undergoing anti-tuberculosis treatment.33 Observational studies have also demonstrated that regular liver function monitoring every 2–4 weeks during the initial weeks of treatment can facilitate early detection of DILI, prevent progression to severe liver injury, and ultimately improve patient outcomes.34 In our study, the median time between the initiation of treatment and the onset of liver injury was 20 days (IQR: 11–36). Notably, 68.4% of patients exhibited no apparent clinical symptoms, and liver injury in these cases was identified solely through routine biochemical surveillance. This supports previous evidence and underscores the importance of early liver function tests in diagnosing ATB-DILI.
In our cohort, hepatocellular injury was the predominant form of ATB-DILI, accounting for 71.4% of cases. This is consistent with previous DILI studies in China, which reported hepatocellular injury in 51.4% of cases, though our proportion is notably higher.35 We further found that the hepatocellular pattern was more common in younger patients, while cholestatic injury tended to occur in older individuals. This age-related pattern of liver injury aligns with findings from other DILI studies,24,36 but contrasts with a study conducted in Pakistan.37 However, the exact underlying mechanisms require further investigation. Previous studies have shown that cholestatic DILI is strongly associated with chronic DILI progression and adverse outcomes, including acute liver failure and increased mortality.11 Malnutrition and low hemoglobin levels have also been linked to a higher risk of severe ATB-DILI.38 In our study, severe ATB-DILI occurred in 46 patients (8.1%), with advanced age (≥60 years) identified as an independent risk factor.
To further explore the interaction between age and severe ATB-DILI, we performed subgroup analyses. A significant association between advanced age and severe ATB-DILI was observed in the hepatocellular subgroup (OR = 3.59, p = 0.002), while the association did not reach statistical significance in the cholestatic or mixed types. However, post hoc power analysis revealed that these subgroups were underpowered (cholestatic: 23.1%, 34.5%), and the absence of significance should be interpreted with caution. The borderline interaction p-value (p = 0.075) indicates a potential modification effect by injury type. This emphasizes the importance of age-based risk stratification, particularly in patients with hepatocellular-type ATB-DILI. These findings contribute nuanced evidence to an area with limited prior data. While hepatocellular injury appears more common and strongly associated with age-related severity, we cannot exclude similar risks in other subtypes due to statistical limitations. Future studies with larger subgroup sizes are warranted to validate these trends.
Several studies have highlighted hyperbilirubinemia as an independent predictor of mortality in patients with DILI.39 In the present study, jaundice was observed in 18.2% of patients, which is like the 15.9% reported in another study,26 but lower than in other centers that reported higher mortality rates.39 This difference may be attributable to an earlier diagnosis of liver injury in the current study. Our study found no statistically significant difference in the use of prophylactic hepatoprotective drugs between patients with severe and non-severe ATB-DILI. This phenomenon may be attributed to inter-individual variability and differential responses to hepatoprotective agents among patients with different patterns of liver injury. The efficacy of prophylactic hepatoprotective drugs has been a topic of considerable debate. Some retrospective studies have failed to demonstrate a significant reduction in the risk of ATB-DILI associated with their use,40 while others have reported potential benefits in preventing liver injury during anti-tuberculosis treatment.41,42 These conflicting findings highlight the need for further research to clarify the role of hepatoprotective drugs in managing ATB-DILI. Moreover, while ATB-DILI generally has a favorable prognosis, approximately 5% of cases may progress to acute liver failure.26 In our study, 46 patients (8.1%) fulfilled Hy’s law criteria, and 8 patients (1.4%) were diagnosed with acute liver failure. A multicenter prospective study in the United States over a decade ago identified standard anti-TB treatment as the leading cause of drug-induced acute liver failure, with 40% of cases requiring liver transplantation and 32% resulting in death.43 However, with improved medical care and increased awareness of ATB-DILI, a lower incidence of severe liver injury was observed in our study, consistent with previous reports.25,26
This study has several limitations. First, as a single-center retrospective analysis lacking long-term follow-up data, it is subject to selection bias and limited generalizability. We were unable to assess clinical outcomes such as mortality or liver function recovery, which restricts the comprehensive evaluation of disease severity. Second, the cohort consisted of patients from a tertiary care hospital, whose characteristics may not reflect those in primary or community settings. Third, reliance on inpatient data may have excluded milder or outpatient ATB-DILI cases, potentially underestimating the true prevalence. Additionally, the small number of severe cases in the cholestatic and mixed subgroups limited statistical power, raising the possibility of type II error. Future multicenter studies with larger and more diverse populations are needed to validate the age-related risk across all liver injury types.
In conclusion, this retrospective study of hospitalized TB patients in China revealed that ATB-DILI occurs in approximately 2.0% of cases, with hepatocellular injury being the predominant pattern. Most patients were asymptomatic and diagnosed through routine biochemical monitoring. Advanced age (≥ 60 years) was identified as an independent risk factor for severe ATB-DILI. Subgroup analysis showed that this association was significant in the hepatocellular injury type, while the associations in the cholestatic and mixed types did not reach statistical significance, this may reflect limited statistical power in these subgroups. These findings suggest that in clinical practice, enhanced liver function monitoring and early identification of liver injury should be prioritized in elderly patients receiving anti-tuberculosis therapy. Future research should prioritize prospective, multicenter designs to validate these associations and improve risk stratification tools.