Background
Tuberculosis (TB) remains one of the most deadly infectious diseases worldwide [
1]. In 2015, approximately 10.4 million new cases of TB were diagnosed globally. Despite a decrease in the mortality rate in 2015 from 2000, TB caused approximately 1.8 million deaths in 2015 [
1]. Although effective anti-TB drugs are readily available, the treatment of TB is not always successful because treatment-emergent adverse events—such as hepatotoxicity, peripheral neuropathy, gastrointestinal upset, hyperuricemia, optic neuritis, and cutaneous reactions [
2]—often lead to treatment interruption and further dissemination of the TB bacilli.
Hepatotoxicity is the most common adverse event associated with TB treatment, with an incidence rate between 10.2% and 18.9% in Taiwan, and it is potentially life-threatening [
3‐
6]. The risk of drug-induced liver injury during anti-TB treatment ranges from 5% to 33% according to the American Thoracic Society [
7]. Clinical presentations may include low-grade fever, general malaise, poor appetite, nausea, vomiting, abdominal distension, icteric conjunctiva, and elevated serum aminotransferases and bilirubin levels. These symptoms and signs are typically indistinguishable from those of acute biliary events (ABEs) such as cholelithiasis, biliary obstruction, acute cholecystitis, and cholangitis. Delayed or incorrect diagnosis of ABEs may cause unnecessary anti-TB treatment interruption, or if left untreated, may lead to intra-abdominal complications, such as acute pancreatitis, sepsis, or perforation [
8]. However, the association between ABEs and anti-TB treatment has not been elucidated thus far. Therefore, in this paper, we first report four cases of ABEs during anti-TB treatment, and subsequently, we investigate the incidence rate and risk factors of ABEs during anti-TB treatment in a nationwide TB cohort.
Discussion
To our knowledge, this was the first study to investigate ABEs during anti-TB treatment. The incidence rate of ABEs during anti-TB treatment in the case study cohort was 0.11% (4/3686), which was similar to the finding in the nationwide cohort (0.12%). ABEs do not appear to onset at a particular time. The risk factors for ABEs are older age and comorbid DM, with the latter occurring in half of the patients in the case series.
Because ethambutol and streptomycin are the only first-line TB therapies that are not associated with hepatotoxicity, abnormal liver function remains common during anti-TB treatment. Transient asymptomatic elevation of aminotransferase levels was observed in 10% to 20% of patients who have received isoniazid [
11]. Furthermore, transient asymptomatic hyperbilirubinemia was detected in 0.6% of patients receiving rifampin [
2]. The incidence of symptomatic and severe hepatotoxicity varies between different studies, ranging from 5% to 33% [
7]. However, the clinical presentations of hepatotoxicity are typically indistinguishable from those of ABEs. Distinguishing hepatotoxicity from ABEs has not been investigated in this field. Therefore, when a patient who is treated for TB presents symptoms of an ABE—right upper quadrant pain, nausea, and vomiting with hyperbilirubinemia— the patient may be misdiagnosed with drug-induced hepatitis by the primary care physician. If the ABE is mild or self-limited, such as a passing stone, the physician may attribute the patient’s recovery to discontinuation of anti-TB treatment, further delaying the diagnosis of an ABE. Because some ABE cases should be managed aggressively with invasive procedures (more than two-thirds in the present nationwide cohort), delayed diagnosis may increase morbidity and mortality [
12]. The findings of this study emphasize that although ABEs are not common, increasing awareness of these potential complications and practicing differential diagnosis are crucial for early diagnosis and proper management of ABEs.
From the nationwide cohort in this study, we discovered that older age and comorbid DM are two significant risk factors for ABEs during anti-TB treatment. These findings are similar to those of a previous multicenter study performed in Italy [
13], which discovered that older age and DM are both independent risk factors for cholelithiasis. Epidemiological studies on the risk of gallbladder disease among patients with DM have yielded inconsistent results. In a case–control study [
14], the prevalence of DM was significantly higher in patients with gallstone than in the control group (11.6% vs. 4.8%, odds ratio 2.55 [1.39–4.67]). In studies demonstrating a positive association between DM and gallbladder disease or gallstones, the risk ratios have been reported to lie in the range 1.68–2.09 [
15‐
18]. Other studies, however, have not found an association [
19,
20]. The results of a recent meta-analysis suggested that a diagnosis of DM may increase the relative risk of gallbladder disease by 56% [
21].
Patients with ABEs during anti-TB treatment may have asymptomatic cholelithiasis before treatment. When treatment begins, isoniazid and rifampin may inhibit the bile salt export pump [
22] and accelerate the formation of cholesterol stone [
23], with the previous asymptomatic stone used as a nucleus, resulting in a higher risk of ABEs. Cholelithiasis may be caused by either cholesterol gallstones, which are a mixture of cholesterol, bile salt, and phospholipids, or pigmented gallstones. The relative proportion of each component predicts if the mixture of cholesterol, bile salt, and phospholipids forms crystals, micelles, or vesicle micelles [
24]. Crystal formation is associated with higher proportion of bile salt. The crystal becomes the nucleus of a cholesterol gallstone. Insulin treatment has been shown to increase the biliary saturation index, resulting in cholesterol precipitation and gallstone formation [
25,
26].
Another possible mechanism for the increased risk of an ABE in patients with DM may be decreased gallbladder motility as a result of denervation [
27,
28]. The underlying pathogenesis is still unclear. One possibility is fewer cholecystokinin receptors on the gallbladder wall due to diabetic autonomic neuropathy leading to poor response to cholecystokinin stimulation [
27]. In addition to impaired cholinergic innervation, cholecystoparesis may result from increased dopaminergic activity [
29‐
32].
We discovered that duration of rifamycin use in the ABE group was significantly shorter than that in the non-ABE group. This finding is most likely to be confounded by indication. Among the first-line anti-TB drugs, only rifampin has been reported to be associated with cholestasis, which typically occurs in the first month of treatment [
7]. Rifampin inhibits the major bile salt exporter pump and blocks bilirubin uptake [
33]. In addition, rifampin competitively affects the clearance of bilirubin from the sinusoidal membrane or canaliculular level in a dose-dependent manner [
34,
35]. Therefore, when the bilirubin level increases, primary care physicians tend to discontinue rifamycin treatment first. However, it was impossible to confirm the effect of rifampin on the development of ABEs in this retrospective study. Because it is unethical to treat TB without using rifamycin, a prospective interventional study comparing different doses of rifampin could be used to address this issue.
The present study had some limitations. First, because of the retrospective design of the study, the incidence of ABEs was probably underestimated in the hospital-based cohort because of missing mild and transient events. However, the effect of this underestimation may not be large because serum transaminase and bilirubin assessments in the second, fourth, and eighth weeks of treatment are recommended by TB treatment guidelines in Taiwan [
36], and the percentage of follow-up tests performed for liver function in the hospital-based cohort was > 90%. Second, because the NHIRD cohort study was based solely on claims data, ABE incidence may have been underestimated because of noncomprehensive data. Finally, the causal relationships between ABEs, rifampin treatment, and comorbid DM were not addressed. Further prospective studies are necessary to clarify these issues.