Fast track — Articles
Microscopic-observation drug susceptibility and thin layer agar assays for the detection of drug resistant tuberculosis: a systematic review and meta-analysis

https://doi.org/10.1016/S1473-3099(10)70165-1Get rights and content

Summary

Background

Simple, rapid, and affordable tests are needed to detect drug resistance in Mycobacterium tuberculosis. We did a systematic review and meta-analysis to investigate the accuracy of microscopic-observation drug susceptibility (MODS) and thin layer agar (TLA) assays for rapid screening of patients at risk of drug-resistant tuberculosis.

Methods

In accordance with protocols and methods recommended by the Cochrane Diagnostic Test Accuracy Working Group, we systematically searched PubMed, Embase, and Biosis for reports published between January, 1990, and February, 2009. We included studies investigating detection of drug resistance in M tuberculosis with the MODS or TLA assay, and in which an accepted reference standard was used. Data extracted from the studies were combined by use of bivariate random-effects regression models and hierarchical summary receiver operating characteristic curves to estimate sensitivity and specificity for detection of resistance to specific drugs.

Findings

We identified 12 studies, of which nine investigated the MODS assay and three investigated the TLA assay. For the MODS assay of rifampicin resistance, pooled estimates were 98·0% (95% CI 94·5–99·3) for sensitivity and 99·4% (95·7–99·9) for specificity. For the MODS assay of isoniazid resistance with a 0·1 μg/mL cutoff, pooled sensitivity was 97·7% (94·4–99·1) and pooled specificity was 95·8% (88·1–98·6), but with a 0·4 μg/mL cutoff, sensitivity decreased to 90·0% (84·5–93·7) and specificity increased to 98·6% (96·9–99·4). All assessments of rifampicin and isoniazid resistance with the TLA assay yielded 100% accuracy. Mean turnaround time was 9·9 days (95% CI 4·1–15·8) for the MODS assay and 11·1 days (10·1–12·0) for the TLA assay.

Interpretation

MODS and TLA assays are inexpensive, rapid alternatives to conventional methods for drug susceptibility testing of M tuberculosis. Our data and expert opinion informed WHO's recommendation for use of selected non-commercial drug susceptibility tests, including MODS, as an interim solution until capacity for genotypic or automated liquid culture drug susceptibility testing is developed.

Funding

Stop TB Department of WHO.

Introduction

The growing problem of drug resistance in Mycobacterium tuberculosis is accompanied by increasing demand for quick, cheap, and easy techniques to detect resistance.1 Patients infected with multidrug resistant (MDR) or extensively drug resistant (XDR) strains of M tuberculosis need treatment regimens that include second-line drugs in addition to extended treatment duration.2 Even when treated appropriately, MDR and XDR tuberculosis have substantially worse outcomes3 and longer infectious periods than does drug-susceptible tuberculosis.4 Therefore, prevention of the spread of MDR and XDR tuberculosis is important, for which timely identification of such cases is the first and most crucial step.

Several methods to detect drug resistance are available, but none clearly satisfies the demands of quick, cheap, and easy. Traditional agar-based methods can take months for results. Commercial drug susceptibility testing with liquid culture decreases turnaround times but requires expensive equipment. Molecular detection of gene mutations associated with drug resistance has also been developed, with variable sensitivity reported especially for in-house methods.5 Commercial versions of line-probe assays have high accuracy6, 7 and were recently endorsed by WHO for rapid screening of MDR tuberculosis.8 However, the expertise needed for a laboratory to offer molecular diagnostics is too great for these tests to be implemented in many resource-constrained settings with poor laboratory infrastructure.

Although liquid cultures and line-probe assays have been endorsed by WHO and phased implementation is underway in many countries, interim measures are needed to meet the needs of low-income settings with high rates of MDR and XDR tuberculosis. Non-commercial techniques to test drug susceptibility with inexpensive and widely available laboratory equipment and supplies—such as microscopic-observation drug susceptibilty (MODS) and thin layer agar (TLA) assays—could be implemented in these settings with minimum cost and training. In both MODS and TLA testing, drug-free and drug-containing media (liquid for MODS, solid for TLA) are inoculated with specimens from patients, and cultures are microscopically examined for early growth or microcolonies.9, 10 Growth of M tuberculosis in or on drug-free media indicates a positive culture, whereas growth of M tuberculosis in or on both drug-free and drug-containing media indicates resistance.

We did a systematic review of published reports and a meta-analysis of studies examining the diagnostic accuracy, performance (including contamination rate and turnaround time), and outcomes important to patients of MODS and TLA assays for the detection of drug resistance in M tuberculosis. We followed a standard protocol for systematic reviews and meta-analyses,11 and used methods recommended by the Cochrane Diagnostic Test Accuracy Working Group.12

Section snippets

Search strategy and study selection

We searched PubMed, Embase, and Biosis for reports published between January, 1990, and February, 2009 (inclusive). The first search used the terms tuberculosis[mesh] OR mycobacter*[ti] OR acid-fast[ti] OR tuberculous[ti] and the second search used the terms MODS[tw] OR “microscopic-observation”[tiab] OR “drug-susceptibility”[ti] OR (drug[ti] AND susceptibility[ti]) OR microcolony[tiab] OR (liquid[ti] AND culture[ti]) OR thin-layer[tiab] OR (thin[tiab] AND layer[tiab] AND agar[tiab]) OR

Results

We identified 2072 citations, of which 1406 unique articles remained after exclusion of duplicate articles, and 54 remained after screening of titles and abstracts (figure 1). These articles were retrieved for full-text review, of which 12 were eligible for inclusion in the primary analysis: nine of MODS testing21, 22, 23, 24, 25, 26, 27, 28, 29 and three of the TLA assay30, 31, 32 (table 1; see webappendix for primary accuracy data from included studies). We identified two additional studies

Discussion

From pooled estimates of nine studies, MODS testing had high accuracy for detection of rifampicin resistance, but showed slightly lower sensitivity for detection of isoniazid resistance. We identified only three studies assessing the TLA assay; however, in all studies, the assay had 100% concordance with the reference standards for detection of rifampicin or isoniazid resistance.

For detection of isoniazid restistance with the MODS assay, studies in which a cutoff of 0·1 μg/mL was used had

References (43)

  • Molecular line probe assays for rapid screening of patients at risk of multidrug resistant tuberculosis (MDR-TB)

  • J Coronel et al.

    MODS: a user guide (v12.1). Microscopic observation drug susceptibility assay

    (2008)
  • A Martin et al.

    Procedure manual: thin layer agar (TLA) microcolony detection, version 3

    (2009)
  • M Pai et al.

    Systematic reviews and meta-analyses: an illustrated, step-by-step guide

    Natl Med J India

    (2004)
  • MM Leeflang et al.

    Systematic reviews of diagnostic test accuracy

    Ann Intern Med

    (2008)
  • P Whiting et al.

    The development of QUADAS: a tool for the quality assessment of studies of diagnostic accuracy included in systematic reviews

    BMC Med Res Methodol

    (2003)
  • J Zamora et al.

    Meta-DiSc: a software for meta-analysis of test accuracy data

    BMC Med Res Methodol

    (2006)
  • M Pai et al.

    Systematic reviews of diagnostic test evaluations: what's behind the scenes?

    ACP J Club

    (2004)
  • B Littenberg et al.

    Estimating diagnostic accuracy from multiple conflicting reports: a new meta-analytic method

    Med Decis Making

    (1993)
  • RM Harbord et al.

    metandi: meta-analysis of diagnostic accuracy using hierarchical logistic regression

  • JP Higgins et al.

    Measuring inconsistency in meta-analyses

    BMJ

    (2003)
  • Cited by (113)

    • Methods and reporting of systematic reviews of comparative accuracy were deficient: a methodological survey and proposed guidance

      2020, Journal of Clinical Epidemiology
      Citation Excerpt :

      In 61 (48%) reviews, a flow diagram was presented without the number of studies per test, whereas 25 (20%) reviews presented a comprehensive flow diagram with the number of studies per test. Of these 25 reviews, the flow diagrams in five reviews [14,18–21] were notable examples. These flow diagrams clearly showed the number of studies included in the analysis of each test, and also indicated the number of comparative studies available.

    View all citing articles on Scopus
    View full text