Discussion:
The utility of miR-122 and K18 as exploratory DILI biomarkers has been demonstrated by several multicentre studies and this has resulted in FDA support for continued development to full qualification (Food and Drug Administration Centre for Drug Evaluation and Research, 2016). However, the properties of these biomarkers have not been robustly studied in patients with TB in Africa.
Given the potential involvement of the liver in TB infection it was important to determine if circulating concentrations of miR-122 and K18 differed from healthy individuals in the presence of infection, including active, latent and non-tuberculous mycobacterial infection and HIV-TB coinfection. For example, if either biomarker was elevated by mycobacterial infection per se then it would be de-prioritised as a biomarker in this important context of use. In this study we have demonstrated that circulating miR-122 and K18 in healthy volunteers and patients with active TB, latent TB and non-tuberculous mycobacterial infection are not substantially different. This suggests that the presence of mycobacterial infection does not affect circulating miR-122 and K18. Furthermore, circulating K18 in HIV-TB coinfected African patients was similar to the other groups, which were predominantly Caucasian. This suggests that the healthy reference interval for K18 in an African and Caucasian population is likely to be similar. In addition, we have demonstrated that, in the absence of DILI, neither miR-122 nor K18 change substantially upon commencing treatment. Both miR-122 and K18 correlate with ALT, indicating these biomarkers may have diagnostic utility. In this pilot study, miR-122 distinguished those patients with an elevated ALT with greater sensitivity and specificity than K18 but this should be interpreted with caution as there were only two cases of DILI as predefined in our study protocol. In these DILI cases the elevations in ALT temporarily correlated with a rise in both miR-122 and K18. In addition, in one of these patients, miR-122 and K18 rose before ALT, indicating a potential for these novel biomarkers to predict the development of DILI earlier than ALT. The results of this study provide initial evidence for the potential use of both miR-122 and K18 as biomarkers of TB medicine associated DILI.
Further work should focus on determining the diagnostic value of the biomarkers, whether they correlate with rises in ALT and so can diagnose DILI within this population. A clear definition of the dynamic range, sensitivity and specificity of miR-122 and K18 within this population is needed before they can be used as a biomarker of DILI. Furthermore, given that evidence suggests miR-122 and K18 both rise earlier than ALT in paracetamol DILI, it is important to determine if they have the same predictive value in patients with mycobacterial infections. This predictive ability of these novel biomarkers may enable early identification of patients at risk of DILI, leading to prevention of liver injury through halting or altering treatment regimens before significant liver injury develops. Specifically, the biomarkers could be a useful early indicator of the development of DILI in patients being reintroduced to essential anti-TB medications, a group at elevated risk of DILI recurrence.
Our study had a limited number of cases of anti-TB DILI. Historical data suggested approximately 2-5% of patients receiving anti-TB treatment in the UK will develop DILI. However, within the ALISTER study only 2 patients developed DILI, 1.4% of the patients recruited. Larger multi-centre studies are required to recruit enough patients to determine the diagnostic power of miR-122 and K18 in anti-TB DILI. The majority of values for circulating miR-122 concentrations in the patient groups fell within the published upper limit of the healthy reference interval of 45 fM generated from the SAFE-T dataset (Church et al. , 2018). However, there were two patients with miR-122 increased above this healthy reference interval (miR-122 = 77 & 77 fM) when ALT was still normal (ALT = 25 & 26 IU/L). This may reflect a limitation of miR-122, namely that it has been reported to have relatively high variability (Church et al. , 2018). In our study, the variability of the novel biomarkers was higher than ALT, with miR-122 having higher inter-subject variability than K18. Although the previously published healthy reference interval provides a valuable comparison, the circulating concentration of miR-122 in healthy volunteers in this study fell between 0.21-8.75 fM, considerably lower than the published ULN of 45 fM which was generated from the SAFE-T dataset (Church et al. , 2018). This healthy reference interval was developed using a larger sample size than that included in this study. However, it was determined using different quantification and normalisation methods, therefore a direct comparison is challenging. There were two patients who had substantially elevated K18 in the absence of elevated ALT. Firstly, in ALISTER (K18 = 4207 U/L, ALT = 43 IU/L). Secondly, in the SAEFRIF trial, where a patient had K18 ranged from 10,000-20,000 U/L, but not substantially elevated ALT (22-52 IU/L). The reason for these two outliners is unknown and requires further study with larger patient numbers.
In summary, the presence of mycobacterial infection does not alter miR-122 or K18 concentrations in the absence of DILI. African HIV-TB coinfected patients had similar K18 concentrations to healthy volunteers and Caucasian TB patients. Patients who experienced elevations in ALT also demonstrated rises in both miR-122 and K18 indicating the diagnostic potential of these biomarkers. Future trials of miR-122 and K18 as biomarkers of anti-TB DILI could be performed using the data presented in this paper to inform the study design.