MicroRNA‐122 and cytokeratin‐18 have potential as a biomarkers of drug‐induced liver injury in European and African patients on treatment for mycobacterial infection

Aims Patients on antituberculosis (anti‐TB) therapy are at risk of drug‐induced liver injury (DILI). MicroRNA‐122 (miR‐122) and cytokeratin‐18 (K18) are DILI biomarkers. To explore their utility in this global context, circulating miR‐122 and K18 were measured in UK and Ugandan populations on anti‐TB therapy for mycobacterial infection. Methods Healthy subjects and patients receiving anti‐TB therapy were recruited at the Royal Infirmary of Edinburgh, UK (ALISTER—ClinicalTrials.gov Identifier: NCT03211208). African patients with human immunodeficiency virus–TB coinfection were recruited at the Infectious Diseases Institute, Kampala, Uganda (SAEFRIF—NCT03982277). Serial blood samples, demographic and clinical data were collected. In ALISTER samples, MiR‐122 was quantified using polymerase chain reaction. In ALISTER and SAEFRIF samples, K18 was quantified by enzyme‐linked immunosorbent assay. Results The study had 235 participants (healthy volunteers [n = 28]; ALISTER: active TB [n = 30], latent TB [n = 88], nontuberculous mycobacterial infection [n = 25]; SAEFRIF: human immunodeficiency virus‐TB coinfection [n = 64]). In the absence of DILI, there was no difference in miR‐122 and K18 across the groups. Both miR‐122 and K18 correlated with alanine transaminase (ALT) activity (miR‐122: R = .52, 95%CI = 0.42–0.61, P < .0001. K18: R =0.42, 95%CI = 0.34–0.49, P < .0001). miR‐122 distinguished those patients with ALT>50 U/L with higher sensitivity/specificity than K18. There were 2 DILI cases: baseline ALT, 18 and 28 IU/L, peak ALT 431 and 194 IU/L; baseline K18, 58 and 219 U/L, peak K18 1247 and 3490 U/L; baseline miR‐122 4 and 17 fM, peak miR‐122 60 and 336 fM, respectively. Conclusion In patients treated with anti‐TB therapy, miR‐122 and K18 correlated with ALT and increased with DILI. Further work should determine their diagnostic and prognostic utility in this global context‐of‐use.

The global burden of TB is unequally distributed, disproportionately affecting low-and middleincome countries particularly in Africa and Asia. One of the barriers to the effective treatment of TB are the adverse drug reactions experienced by patients on anti-TB medications, with 1 of the most common being drug-induced liver injury (DILI). 2 Three of the 4 first-line drugs used in the treatment of TB-isoniazid, rifampicin and pyrazinamide-are potentially hepatotoxic. 3 The estimate of the incidence of DILI in individuals undergoing anti-TB treatment for active TB varies from 2 to 33% depending on the cohort studied, drug regimen used, monitoring and reporting practices. 2,4,5 Individuals who experience DILI often need to stop treatment and, if clinically indicated, recommence once liver function tests (LFTs) return to normal.
However, for some individuals, re-exposure to the same drugs leads to reoccurrence of DILI, 6 and for others, liver injury progresses even after treatment has stopped. 7 Therefore, there is an unmet clinical need for new tools to improve the safety of this essential antimicrobial treatment.
Diagnosis of DILI relies on LFTs, with alanine aminotransferase (ALT) activity considered a gold standard for determining liver injury.
The DILI Expert Working Group defines DILI as ≥3× upper limit of normal (ULN) of ALT in the presence of symptoms, or ≥5× ULN ALT in the absence of symptoms. 8 Although ALT is currently the gold standard for determining DILI, there are issues associated with its use.
ALT is not specific to the liver and can provide false positive results, with elevations in ALT occurring after muscular damage following exercise 9 or subsequent to a myocardial infarction. 10 Furthermore, elevations in ALT are not specific to DILI 11 and can occur due to metabolic perturbations. 12,13 In paracetamol-overdose DILI, there is a delay between insult to the liver and rise in ALT, 14 meaning ALT is not optimal as a biomarker of DILI in this context. To address these challenges recent work has identified novel biomarkers capable of diagnosing, and in some cases predicting DILI.
MicroRNAs (miRNAs) are small noncoding RNAs regulate posttranscriptional gene expression. MiR-122 is a 22-nucleotide micro-RNA is highly expressed in, and highly specific for, the liver, with little to no expression in other tissues. In liver injury, miR-122 is released from necrotic hepatocytes, resulting in elevated miR-122 concentrations in the bloodstream. 15 Cytokeratin-18 (K18) is an intermediate filament protein responsible for maintaining the cytoskeletal structure in the liver and other epithelial cells and is reported to make up 5% of the liver's total protein content. 16 K18 is a mechanistic biomarker of liver injury, providing information on the pattern of cell death. In apoptosis, the release of a caspase-cleaved form of K18 (cc-K18) is an early event during cellular structural rearrangement, 17 whereas, in necrosis, the full-length form of K18 (FL-K18) is passively released upon cell death. 17 MiR-122 and K18 are able to predict DILI in patients who overdose on paracetamol earlier than standard LFTs 18,19 and provides enhanced hepatic specificity over other biomarkers. 20 Both novel biomarkers have regulatory support from the US Food and Drug Administration as biomarkers for DILI, 21 although to date their development has been largely limited to testing in western populations.
The aim of this study was to explore the properties of miR-122 and K18 in relevant European and African patients with mycobacterial infection (with and without HIV coinfection). Specifically, our aims were to determine whether the infective disease process and routine What is already known about this subject • Drug-induced liver injury (DILI) is a major concern in the treatment of tuberculosis (TB).
• There are new biomarkers for DILI (microRNA-122 and cytokeratin-18) that have the potential to add value because of enhanced specificity/sensitivity compared to current tests.
• These mechanistically informative DILI biomarkers have qualifying data from large USA and European studies.
However, these markers have not been tested in Africa.
Given the global burden of TB our aim was to explore biomarker utility in this global context of use.

What this study adds
• The baseline concentrations of cytokeratin-18 did not substantially differ across UK and Ugandan populations on treatment for TB and human immunodeficiency virus.
MicroRNA-122 did not differ across healthy subjects and UK patients treated for mycobacterial infection.
• These biomarkers significantly correlated with alanine transaminase activity (current liver injury marker) and were elevated with DILI. MicroRNA-122 distinguished those patients with alanine transaminase >50 U/L with higher sensitivity/specificity than K18.
• MicroRNA-122 and cytokeratin-18 have potential in this context-of-use and should be taken forward into larger studies, which could provide data for formal qualification. management with anti-TB medicines affect these biomarkers in the absence of DILI and to characterise how miR-122 and K18 change in relation to ALT and in cases of DILI.

| Healthy subjects
As a control group to test whether circulating miR-122 and K18 were affected by active TB, latent TB or NTM infection, adults with no medical complaints and no medication use were recruited and blood was drawn with informed consent.

| ALISTER clinical study
Participants were recruited at the Royal Infirmary of Edinburgh. Adults Patients with NTM infection had grown at least 2 cultures with nontuberculous mycobacterium and had clinical signs of pulmonary disease. 22,23 Active and latent TB patients were treated following World Health Organisation guidelines. 24,25 Patients with susceptible active TB were treated with isoniazid, rifampicin, ethambutol and pyrazinamide for an initiation phase of 2 months, followed by rifampicin and isoniazid for a continuation phase of 4 months. 24 Patients with latent TB were treated with either a combination of isoniazid and rifampicin for 3 months, or isoniazid or rifampicin alone for 6 months. 25 Patients with Mycobacterium avium complex infection were treated with the recommended regimen of rifampicin, ethambutol and clarithromycin for 2 years. 23 Dependent on the NTM species, disease severity, resistance profile of the infection and tolerance of the individual, drugs were replaced or added, including isoniazid, moxifloxacin, azithromycin and amikacin. 22,23

| SAEFRIF clinical trial
The SAEFRIF clinical trial (NCT03982277) was performed at the Infectious Disease Institute at Makerere University, Uganda. The data presented in this paper provide preliminary data on liver safety from 64 trial participants. A full report on predefined study end-points will be provided on trial completion. Ethical approval for this study was sought from Joint Clinical Research Council ethics committee, the National Drug Authority and the Uganda National Council for Science and Technology. All participants signed an informed consent form prior to study enrolment.
Inclusion criteria were HIV-infected patients aged ≥18 years who were due to initiate rifampicin-containing therapy for newly diagnosed active TB, and were either already taking or planning to start efavirenz-based or dolutegravir-based antiretroviral therapy. Exclusion criteria were patients who have rifampicin resistant TB, pregnant women, women of reproductive age on dolutegravir who decline the use of effective contraceptive, patients with liver disease, ALT > 5× ULN or glomerular filtration rate < 50 mL/min. The trial protocol is described by Nabisere et al. 26

| Data and blood samples
For the ALISTER study and SAEFRIF trial demographic and clinical data from the participants were recorded from medical records and clinical trial records. Liver function test results (ALT) were recorded from each clinic visit. Blood samples were collected at first clinic visit and subsequent clinic visits. Once collected, blood was processed by centrifugation and the supernatant was aliquoted and stored at −80 C.

| Quantification of miR-122
Serum samples were stored at −80 C before analysis. Freeze-thaw cycles were avoided to preserve miRNA integrity. MiR-122 was quan-

| Quantification of K18
Samples were stored at −80 C before analysis. K18  was used to determine formal causality between anti-TB medication and liver injury. 33 The pattern of liver injury was determined using the R ratio, considering ALT and alkaline phosphatase activity. Further factors considered include time to onset, course of injury, risk factors (age and alcohol), concomitant drugs, the exclusion of nondrug causes of injury and previous information on drug hepatotoxicity.   Arab 0 (0%) 1 (1%) 0 (0%) 0 (0%)  Table 1. There was no significant difference in miR-122 (P = .09) between the groups. There was a statistically significant but clinically minor difference in K18 between groups (P = .03; Table 2,   When all samples from all time points were included, there was a significant correlation between ALT and miR-122 (n = 251, Spearman rank r = 0.52, 95%CI = 0.42-0.61, P < .0001; Figure 4A). There was a significant, but less tight, correlation between ALT and K18 (n = 491, Spearman rank r = 0.42, 95% CI = 0.34-0.49, P < .0001; Figure 4B).

MiR
The interindividual variability in miR-122, K18 and ALT was compared in the different patient groups ( In this study there were 2 cases of DILI (as predefined in study protocols), both cases were for patients receiving isoniazid alone for the treatment of latent TB within the ALISTER study. The first case was of a 51-year-old white British male patient, who experienced peak ALT activity of 431 IU/L, (Figures 6A,E). Before starting treatment, he had normal ALT (18 IU/L), miR-122 was 4 fM and his K18