The reviewer commends the study for its unique population-level insights and important policy implications. Nonetheless, they noted limitations such as potential data quality issues, adherence challenges, and residual confounding factors.
RR\ID Evidence Scale rating by reviewer:
Reliable. The main study claims are generally justified by its methods and data. The results and conclusions are likely to be similar to the hypothetical ideal study. There are some minor caveats or limitations, but they would/do not change the major claims of the study. The study provides sufficient strength of evidence on its own that its main claims should be considered actionable, with some room for future revision.
***************************************
Review: This is a well-written and interesting study that uses routinely collected data on HIV and TB from Ukraine to assess the efficacy of isoniazid (INH) preventive therapy in a setting with high rates of MDR TB. The study addresses an important policy question; in areas where MDR is highly prevalent, does it make sense to adhere to the broad recommendation to provide people living with HIV with preventive therapy with a drug (INH) to which the Mtb strains they are exposed to have a reasonable chance of being resistant? Several previous studies (in full disclosure, including one of ours: Huang CC, Becerra MC, Calderon R, Contreras C, Galea J, Grandjean L, Lecca L, Yataco R, Zhang Z, Murray M. Isoniazid Preventive Therapy in Contacts of Multidrug-Resistant Tuberculosis. Am J Respir Crit Care Med. 2020 Oct 15;202(8):1159-1168.) have shown that INH preventive therapy protects individuals exposed to MDR strains in households, and some investigators have suggested that the mechanism by which INH prevents TB progression may differ from its anti-microbial activity (Khan SR, Manialawy Y, Siraki AG. Isoniazid and host immune system interactions: A proposal for a novel comprehensive mode of action. Br J Pharmacol. 2019 Dec;176(24):4599-4608. doi: 10.1111/bph.14867. Epub 2019 Nov 12. PMID: 31517993; PMCID: PMC6965675.).
This study conducted in the real-world setting of Ukraine where 25% of TB is MDR finds that individuals enrolled in the national HIV treatment program who received INH preventive therapy were significantly less likely to develop TB than those who did not receive INH. To my knowledge, this is the first study that has examined this issue on a population level and it thus provides a unique perspective on a critically important public policy question. These results provide support for the idea that INH preventive therapy can be a useful strategy even in areas with high levels of MDR. Furthermore, this additional evidence of the efficacy of INH against MDR TB may help focus future research on the potential alternative mechanisms by which this drug might act.
The study has several limitations which inform my decision to classify this as “reliable” rather than “strong.”
First, the study merges data that were routinely collected in two different programs – the national HIV and TB programs which use different electronic systems to record patient data. Fortunately, these data were collected between January 2018 and Feb 2022 – ie. before the war (which has likely disrupted both HIV and TB public health systems) but it remains difficult to assess the quality of the routine data. Ideally, some form of validation would reassure readers that the data are relatively accurate – this might include comparing the rates of TB in HIV program participants with some external assessment of TB rates in the HIV-positive Ukrainian population (such as the WHO estimates for those years which are admittedly hard to find). Given the technical challenges of merging the eTB system data with the national HIV database, I suspect these data are not as reliable as one would expect in an ideal cohort study with prospective data collection and follow-up.
In addition, the surveillance data identifies people who received INH but not necessarily people who adhered to the medication they received. Often, patients who drink alcohol do not reliably adhere to INH due to the side effects. Notably, the study reports alcohol “misuse” (not well defined) in only 2.7% of non-IPT users and 4.3% of IPT users while a cross-national 2020 study of the prevalence of alcohol use found that 10% of randomly sampled Ukraine residents reported “alcohol abuse: and 13% were diagnosed with an alcohol use disorder (Glantz, M.D., Bharat, C., Degenhardt, L., Sampson, N.A., Scott, K.M., Lim, C.C., Al-Hamzawi, A., Alonso, J., Andrade, L.H., Cardoso, G. and De Girolamo, G., 2020. The epidemiology of alcohol use disorders cross-nationally: Findings from the World Mental Health Surveys. Addictive behaviors, 102, p.106128). This suggests to me that some of the routinely collected data is incomplete and that adherence might not have been well captured by data on who received their medication doses.
Similarly, there is no information on how often participants were evaluated for TB at local clinics and what the compliance rate was with these planned visits.
Secondly, the main issue in any comparison of people who got and did not get an intervention in the “real-world” context is why the “control group” did not get the intervention. In this case, the authors clarify that the INH preventive therapy program started in 2018 and initially focused on people who were recently diagnosed with HIV. They note the following differences between the control (no IPT) group and the IPT group: the no IPT group had much longer follow-up, more viral suppression, were younger, had higher CD4 counts, less alcohol use and more history of incarceration. The final efficacy assessments are adjusted for these factors but not for other factors that might be related to both TB outcome and IPT use – socio-economic status, nutritional status, occupation, drug use, education level etc. the authors note but do not quantify “a large amount of missingness on smoking status, nutritional status and homelessness.” There is thus the real risk of residual confounding by some of these factors and given the distribution of known risk factors, it is difficult to speculate about what kinds of bias this might have introduced.
Lastly, the efficacy of IPT in this study was about 50% (non-IPT users were twice as likely to develop TB as IPT users.) This efficacy is far less than what is observed for IPT in clinical trials of IPT in HIV patients (on the order of 90% efficacy or above) and it is possible that IPT worked for the majority of people who were exposed to DS strains and not for those exposed to INH resistant or MDR strains – although the finding that MDR rates were similar in the IPT and non-IPT group militates against this interpretation. Notably, this efficacy is very similar to that reported on a 2020 multi-center prospective study of IPT in India which also reported 50% efficacy among HIV-infected individuals followed prospectively between 2013 and 2016 (Padmapriyadarsini, C., Sekar, L., Reddy, D., Chitra, A., Poornagangadevi, N., Selvaraj, M., Bhavani, P.K., Mothi, S.N., Nandagopal, K., Vennila, S. and Tamizhselvan, M., 2020. Effectiveness of isoniazid preventive therapy on incidence of tuberculosis among HIV-infected adults in programme setting. Indian Journal of Medical Research, 152(6), pp.648-655). That study made no mention of the background rates of MDR in the specific sites in which the study was carried out but MDR rates in India in 2014 were much lower than in Ukraine during this period, again suggesting that the lower efficacy rate has less to do with drug resistance and more to do with sub-optimal adherence in the “real world.”
All in all, I find this an important contribution to the sparse literature on IPT in the setting of high MDR rates and my enthusiasm is dampened only by the serious challenge of using routine data to address such a complex question.
Creative Commons Attribution 4.0 International License (CC-BY 4.0)
