From the authors:
We appreciate the response from R.F. Luo and co-workers on our meta-analysis on the accuracy of Xpert MTB/RIF for extrapulmonary tuberculosis [1], as well as the data they present on various approaches to cerebrospinal fluid (CSF) processing for Xpert MTB/RIF testing.
In a recent guidance document on Xpert, the World Health Organization (WHO) recommend that Xpert should be used as a first-line test over conventional microscopy and culture in patients with suspected tuberculous meningitis [2, 3]. This recommendation was based on a systematic review of the evidence and expert consensus [4]. However, our systematic review noted the highly variable sample processing methods used across and within studies, and was unable to identify the best approach for sample processing. The latter is largely due to the lack of recommendations from both the manufacturer and WHO on how to process nonrespiratory samples.
WHO has recognised the need for such guidance and has published an Xpert MTB/RIF implementation manual with recommendations on the technical and operational “how to”, which includes standard operating procedures for processing of CSF, lymph node samples and other tissues [5]. While this is a good step forward, the recommendations are based on expert opinion and limited experimental data on the optimisation of sample preparation comparing different protocols on the same clinical samples or spiked samples in a controlled laboratory setting.
The data by R.F. Luo and co-workers addresses this knowledge gap. In a controlled laboratory environment comparing different protocols on CSF, they were not able to reproduce the finding in our systematic review of an increased sensitivity of Xpert on CSF with a centrifugation step prior to inoculation with the sample reagent. Their findings further suggest that for the paucibacillary type of samples, such as CSF, less sample reagent might be necessary.
These findings need to be confirmed independently and should be combined with testing of the tuberculocidal effect at lower sample-to-sample-reagent ratios before they can be recommended for clinical use. We hope that other groups will follow the example of R.F. Luo and co-workers and evaluate different protocols for different sample types, so that Xpert testing on nonrespiratory samples can be optimised.
Acknowledgments
We would like to acknowledge that we have collaborated with R.F. Luo and N. Banaei (both Clinical Microbiology Laboratory, Stanford University School of Medicine, Palo Alto, CA, USA) on other projects.
Footnotes
Conflict of interest: Disclosures can be found alongside the online version of this article at erj.ersjournals.com
- Received June 11, 2014.
- Accepted June 12, 2014.
- ©ERS 2014
