Improving the management of hospitalized presumptive Xpert MTB/RIF-negative tuberculosis cases in Ethiopia
Ethiopia is one of the 30 countries in the world with a high burden of tuberculosis (TB). While the country has made notable progress in TB management over the past decade, only 61% of pulmonary cases are bacteriologically confirmed. Consequently, empiric TB treatment was initiated in 39% of cases. While empiric treatment is beneficial for patients who truly have TB, it has been shown that it does not reduce overall mortality among people presenting with signs and symptoms of TB and may even increase the risk of death when other diagnoses are overlooked. To improve the management of presumptive TB cases, defined as persons presenting with signs or symptoms of TB in whom the presence of Mycobacterium tuberculosis could not be detected, it is important to understand the ability of the diagnostic algorithm to detect TB disease and ascertain the influence of empiric therapy on survival in people with a negative Xpert MTB/RIF result. It is also essential to accurately discriminate pulmonary TB from other pulmonary infections in ill, hospitalized cases in settings such as Ethiopia, where the health infrastructure is limited. Therefore, the overall goal of this PhD study was to improve the management of hospitalized presumptive Xpert MTB/RIF-negative TB cases in Ethiopia by evaluating the role of empiric TB treatment in settings where Xpert MTB/RIF is used as a diagnostic test for people with presumptive pulmonary TB, the value of switching from the Xpert MTB/RIF assay to the more sensitive Xpert MTB/RIF ultra (Xpert Ultra) assay to diagnose pulmonary TB, and the role of chest X-ray and Lateral flow urine lipoarabinomannan (LF-LAM) in settings where Xpert Ultra is available. Moreover, in patients with presumptive TB in whom the presence of TB was excluded, we aimed to better understand the potential bacterial pathogen most likely to be responsible for the patients’ lower respiratory tract infections (LRTIs) to develop optimal diagnostic and treatment strategies. To achieve this goal, we enrolled a cohort of 250 people presenting with symptoms or signs of TB who had a negative Xpert MTB/RIF result, the initial diagnostic used in Ethiopia. Of these 250 participants, 125 were consecutive hospitalized patients initiated on empiric TB treatment and 125 participants consecutive hospitalized patients who were not started on TB treatment. We first assessed the ability of diagnostic algorithms to detect TB disease in people who screened negative on the Xpert MTB/RIF assay and found that the clinical algorithm consisting of clinical response to an antibiotic trial and chest X-ray findings led to identification of 15.6% of microbiologically confirmed TB cases. Even though confirmed TB was more prevalent in those patients in whom the physician decided to start empiric TB treatment (24.8% vs 6.4%), empiric TB treatment led to substantial overtreatment of TB and had no effect on 6-month survival (aOR 0.74, 95% CI: 0.1-2.7). The use of the more sensitive Xpert Ultra assay diagnosed 89% of all bacteriologically confirmed cases of TB in our study population. The odds of confirmed TB were two to five times as higher in patients with chest X-ray findings suggestive of TB, but chest X-ray had poor sensitivity (69% when read by clinicians and 79.5% when read by a radiologist) and the addition of chest X-ray to a simplified TB risk prediction tool based on clinical symptoms did not improve its performance (AUC: 84.5% to 85.6% vs 84% to 86%). Among the HIV positive (n = 52) and severely ill (n = 16) patients, the LF-LAM assay failed to detect any case that was not detected by Xpert Ultra. Among patients with presumptive TB in whom the presence of TB was excluded by the Xpert Ultra assay, 16S rRNA gene amplicon sequencing identified the pathogen responsible for the patients' LRTI in 6.0% (13/215) of cases. Mycoplasma pneumoniae (n = 7), Bordetella pertussis (n = 2), Acinetobacter baumanii (n = 2), and Pseudomonas aeruginosa (n = 2) appeared to be possible causes of the TB-like symptoms. In other patients, putative pathogens were present but the findings were similar in sputum samples from patients diagnosed with active TB (Xpert Ultra positive) and samples from patients without TB (Xpert Ultra negative). The presence of Streptococcus (pseudo) pneumoniae was associated with higher odds of radiological abnormalities (aOR 2.5, 95% CI 1.12–6.16), and the presence of Streptococcus (pseudo) pneumoniae (aOR 5.31, 95% CI 1.29–26.6) and Moraxella catarrhalis/nonliquefaciens (aOR 12.1, 95% CI 2.67–72.8) with higher odds of six-month mortality, suggesting that these pathogens may have clinical relevance. In conclusion, even though the currently used clinical algorithm identified most of the bacteriologically confirmed cases, it has suboptimal performance when used in the context where Xpert MTB/RIF is the first diagnostic. The positive predictive value of the algorithm was low (24.8%), indicating a substantial number of false-positive results; it failed to detect 6.4% of people with active TB, leading to delayed or missed treatment initiation; and 75% of patients who started empiric TB treatment were culture-negative, suggesting overtreatment. Adding chest X-ray readings to the decision algorithm did not significantly improve clinicians' ability to predict bacteriologically confirmed TB beyond what is possible through clinical judgment. The use of LF-LAM did not yield any benefit when used in combination with Xpert Ultra in HIV-positive presumptive pulmonary TB cases. Nevertheless, even in settings with access to Xpert Ultra, patients who are unable to produce sputum may still benefit from the use of the LF-LAM assay. Countries with high TB and HIV burdens, like Ethiopia, should urgently replace Xpert MTB/RIF with the highly sensitive Xpert Ultra assay for the diagnosis of TB. The use of LF-LAM in hospitalized HIV-positive patients with suspected TB should be avoided in settings where Xpert Ultra is available, but it may still be beneficial for those unable to produce sputum. To improve clinical judgment skills, the National TB Program (NTP) should provide professional development training and keep healthcare providers updated on other bacterial pathogens causing LRTIs. The routine practice of administering an empiric TB treatment to patients negative for the Xpert Ultra test should be re-evaluated. Future research using tools with higher discriminatory power than 16S rRNA sequencing is needed to identify with greater accuracy the bacterial, viral, and fungal pathogens that may cause LRTI in Xpert Ultra-negative patients.
Antwerp : University of Antwerp, Faculty of Medicine and Health Sciences , 2024
171 p.
Supervisor: Van Rie, Annelies [Supervisor]
Supervisor: Abebe, Gemeda [Supervisor]
Supervisor: Gudina, Esayas Kebede [Supervisor]
Full text (open access)
Research group
Publication type
Publications with a UAntwerp address
External links
Creation 15.02.2024
Last edited 22.02.2024
To cite this reference