Sarah Lofgren



Nominated From: University of Minnesota

Research Site: Uganda

Research Area: Infectious Disease

Primary Mentor: David Boulware

Research Project

Xpert MTB/RIF Ultra detection of TB meningitis

Tuberculosis meningitis (TBM) has high morbidity and mortality. Diagnosis is very difficult owing to less than ideal laboratory diagnosis techniques. Early initiation of therapy leads to better outcomes and so delay is costly to the patient – particularly in terms of neurologic outcomes. Presently, AFB smear microscopy of the CSF is seldom positive. TB culture is a gold standard but often takes 4-6 weeks to ‘become positive’ due to the slow growth of M. tuberculosis and initial paucity of organisms. A more rapid, accurate test would allow more rapid initiation of therapy and likely better outcomes.

The GeneXpert system (Cepheid, Sunnyvale, CA) uses the GeneXpert instrument and a single use diagnostic cartridge (Xpert MTB/RIF test). Typically sputum or a processed sputum pellet is treated with the sample reagent (SR) that has bactericidal and liquefaction activity prior to loading in the cartridge – the instrument then automatically controls nucleic acid isolation and real-time PCR detection to identify M. tuberculosis. This test has been endorsed by the World Health Organization for use in respiratory samples (1-3) and is commercially available in a number of countries – including Uganda. A new GeneXpert System the Ultra from the same manufactures is now out and has been found to be 10 times more sensitive than the original GeneXpert system (4).

The Xpert MTB/RIF test has been tested extensively on sputum samples, but less frequently on CSF (5-7). All of these reports have used the sample processing protocol designed for sputum. Xpert MTB/RIF detects between 71 and 84% of smear-negative, culture positive cases of pulmonary TB, suggesting that the test may have sufficient sensitivity to be used in the diagnosis of TBM (1,2). This was recently shown in a csf study of TBM with a modified protocol which had a 72% sensitivity (8).

Current methods for diagnosis of TBM are inadequate. A rapid, sensitive test that resulted in prompt and actionable results would likely improve patient outcomes. Xpert MTB/RIF is a rapid and accurate test with some experience on CSF however the GeneXpert Ultra promises to have sensitivities consistent with culture (4).

Specific Aims:

This primary project aim is to compare the diagnostic yield of the Xpert assay with improved CSF processing vs. the Xpert after conventional CSF processing vs. mycobacterial culture.


Research Significance

Research Cohort
The proposed plan would be conducted as an IRB-approved nested sub-study to be conducted by Dr. David Boulware and Dr. David Meya and colleagues in Kampala, Uganda as part of the screening process for the “Neurology Outcomes on Antiretroviral Therapy” (NOAT) study sponsored by the National Institute of Neurologic Diseases and Stroke (NINDS). All participants provide written informed consent for meningitis screening.

Research Outline:
The NOAT study is prospectively enrolling HIV infected persons with CNS infections in Uganda. This study would utilize CSF collected during routine implementation of the NOAT study. The proposed study would only alter lab techniques use and would not affect enrollment, methods of specimen collection or treatment.

For the TB diagnostic sub-study, laboratory technologists performing investigational Xpert MTB/RIF Ultra testing on CSF specimens will be blinded as to the results of mycobacterial cultures and smear microscopy. Coded CSF specimens will be submitted to the study laboratory for testing. For each sample 5 ml (or greater) CSF, tested for presence of M. tuberculosis using each of 4 methods:

  • AFB smear microscopy, Endpoints are yield and sensitivity.
  • Mycobacterial Culture, Endpoints are yield, % failed tests, and time to culture result
  • Xpert MTB/RIF using IMPROVED CSF processing, Endpoints are yield, sensitivity, specificity, and % failed tests
  • Xpert MTB/RIF Ultra using IMPROVED CSF processing, End points are yield, sensitivity, specificity, and % failed tests

For specimens with greater than 2 mL remaining after performance of the above tests, the specimen will be centrifuged and sedimented material used for Xpert MTB/RIF testing. This will be the initial protocol, and may be modified based on interim analysis at 3 and 6 months.

Results of the above tests will be linked with a limited set of data from the NOAT study. This dataset will include age or birth year, sex, HIV status, CD4, clinical information about symptoms, clinical information about TB treatment, results of other laboratory testing performed on CSF. Hospital outcomes (e.g. survival) will be collected and assessed via diagnostic results, including Xpert cycle count and time to CSF culture positivity.

Proportions will be compared using the chi squared or Fisher’s exact test, as appropriate. Descriptive statistics will be used to describe diagnostic yield, sensitivity, specificity, test failure rate and proportion of participants with culture-proven TBM.





View on PubMed