Point-of-care diagnosis of tuberculosis: Past, present and future

Authors

  • KEERTAN DHEDA,

    Corresponding author
    1. Institute of Infectious Diseases and Molecular Medicine, University of Cape Town, Cape Town, South Africa
    • Lung Infection and Immunity Unit, Division of Pulmonology and UCT Lung Institute, Department of Medicine, University of Cape Town, Cape Town, South Africa
    Search for more papers by this author
  • MORTEN RUHWALD,

    1. Clinical Research Centre, Copenhagen University Hospital, Hvidovre, Denmark
    Search for more papers by this author
  • GRANT THERON,

    1. Lung Infection and Immunity Unit, Division of Pulmonology and UCT Lung Institute, Department of Medicine, University of Cape Town, Cape Town, South Africa
    Search for more papers by this author
  • JONATHAN PETER,

    1. Lung Infection and Immunity Unit, Division of Pulmonology and UCT Lung Institute, Department of Medicine, University of Cape Town, Cape Town, South Africa
    Search for more papers by this author
  • WING CHEONG YAM

    1. Department of Microbiology, Queen Mary Hospital, The University of Hong Kong, Hong Kong
    Search for more papers by this author

  • The Authors: Keertan Dheda is a Professor of Respiratory Medicine and the Director of the Lung Infection and Immunity Unit within the Division of Pulmonology, Department of Medicine, at the University of Cape Town. His main research interests are the study of the immunopathogenesis, epidemiology and diagnosis of tuberculosis (TB), including drug-resistant TB. Morten Ruhwald, MD, PhD, heads the TB Immunology Group at Clinical Research Centre, Copenhagen University Hospital, Hvidovre. Research interests include studies of adaptive immunity and development of simple and accurate immunodiagnostic tests for latent TB. Grant Theron is a research fellow in the Lung Infection and Immunity Unit at the University of Cape Town. He is funded by the National Research Foundation and the Claude Leon Foundation. His background is in basic science, and his main research interests relate to new and improved diagnostics for TB and drug-resistance, as well as novel approaches for understanding and disrupting TB transmission. Jonathan Peter is a PhD fellow and consultant physician at the Lung Infection and Immunity Unit, Department of Medicine, at the University of Cape Town. His major research interest is the development and validation of field-friendly TB diagnostics, such as urine lipoarabinomannan, suitable for use in resource-limited, high human immunodeficiency virus (HIV)-prevalent settings. Wing Cheong Yam is a Clinical Bacteriologist/Honorary Associate Professor in the Department of Microbiology, The University of Hong Kong. He is in charge of the development and daily running of molecular microbiology service for Queen Mary Hospital, including rapid diagnosis of emerging infections such as TB and drug-resistant HIV-1.
  • SERIES EDITORS: CHI CHIU LEUNG, CHRISTOPH LANGE AND YING ZHANG

Correspondence: Keertan Dheda, Groote Schuur Hospital, University of Cape Town—Lung Infection and Immunity Unit, H46.41, H Floor, Old Main Building, Cape Town, Western Cape 7925, South Africa. Email: keertan.dheda@uct.ac.za

Abstract

Diagnosis represents only one aspect of tuberculosis (TB) control but is perhaps one of the most challenging. The drawbacks of current tools highlight several unmet needs in TB diagnosis, that is, necessity for accuracy, rapidity of diagnosis, affordability, simplicity and the ability to generate same-day results at point-of-care (POC). When a return visit is required to access test results, time to treatment is prolonged, and default rates are significant. However, a good diagnostic tool is also critically dependent on obtaining an adequate biological sample. Here, we review the accuracy and potential impact of established and newer potential POC diagnostic tests for TB, including smear microscopy, the Xpert MTB/RIF assay (Cepheid) and the Determine TB lipoarabinomannan antigen test (Alere). Novel experimental approaches and detection technologies for POC diagnosis of active TB, including nucleic acid amplification tests, detection of volatile organic compounds or metabolites, mass spectroscopy, microfluidics, surface-enhanced Raman spectroscopy, electrochemical approaches, and aptamers among others, are discussed. We also discuss future applications, including the potential POC diagnosis of drug-resistant TB and presumed latent TB infection. Challenges to the development and roll-out of POC tests for TB are also reviewed.

Abbreviations
HIV

human immunodeficiency virus

LAM

lipoarabinomannan

LED

light-emitting diode

LTBI

latent tuberculosis infection

MDR

multidrug-resistant

MTB/RIF

Xpert MTB/RIF assay

NAAT

nucleic acid amplification test

POC

point-of-care

TB

tuberculosis

WHO

World Health Organization

Introduction

Tuberculosis (TB) remains a global public health malady that claims almost 1.5 million lives annually.[1] While upstream factors such as alleviation of poverty, improvement in housing and transport, and reduction in the rates of human immunodeficiency virus (HIV) co-infection, and downstream factors such as stigmatization, availability of effective TB drugs, adherence and active case finding are clearly important, diagnosis represents another vital link in the chain of TB control. Thus, while rapid and accurate diagnosis represents only one facet of TB control, it is a quintessential one. Before TB can be treated, a diagnosis needs to be made in an efficient and timely manner, preferably at point-of-care (POC), and using accurate and field-friendly tools. The definition of POC is contentious, but, at the least, it implies the ability to make a diagnosis at the point where patient consultation and presentation occurs, and the ability to translate the result into same day treatment, if appropriate (see Textbox 1 for key prerequisites for a POC test). The World Health Organization (WHO) and Stop TB Partnership have earmarked 2015 as a target date for developing a simple POC test for TB.

Textbox 1. Optimal characteristics of a future point-of-care TB diagnostic test (detailed minimum requirements have previously been outlined[2])

Test performance characteristics

    Test performance characteristics
  • Sensitive (>95% for smear-positive and >60% smear-negative culture-positive samples) and specific (>95% compared with culture)
  • User-friendly (ease of sample collection and processing, minimal technical and training requirements)
  • Results available preferably within a single patient-health-care contact (less than ∼3 h) that enables treatment initiation (if available and if appropriate)
  • Robust (shelf-life >24 months, stable in high temperature and humidity, can be battery powered, easy and environmentally friendly waste disposal, and minimal maintenance requirement)

Other important considerations

    Other important considerations
  • Affordable (cost <USD10/test) and accessible in high burden countries
  • Targeted to one or more specific health-care levels, for example home, community, clinic, peripheral lab, hospital; each level has specific user and device requirements
  • Throughput of at least 20 tests per day
  • Ability to process different types of biological samples
  • Easy to read (clear yes/no readout)
  • Can be carried by any health-care worker with minimal training
  • Should ideally screen for drug resistance

Although accuracy, simplicity, affordability and technical robustness are important for sustained uptake, the critical advantage of a POC test is its potential ability to enable rapid treatment initiation at the point of presentation, thus circumventing the problem of patient dropout. Indeed, a significant proportion of patients in high burden settings fail to return to collect their smear microscopy results.[3] It is assumed that providing a rapid result and initiating treatment while the patient is still within the confines of the health-care facility will translate into continued adherence and completion of treatment. However, it is critical to bear in mind that any (POC) diagnostic tool relies on obtaining a suitable and representative biological sample, such as sputum, urine, gastric aspirate, tracheal aspirate or one from an extrapulmonary site.

It has also become apparent that good test accuracy may not necessarily translate into impact and making a diagnosis may not translate into better TB control if other associated factors remain unaddressed. Thus, TB diagnosis cannot be seen in isolation but as a package of care that must be delivered to have optimal impact on TB control. Empirical treatment of TB, which is common in high burden settings, may mask the overall incremental impact of a new diagnostic, but at the least, accurate diagnosis may reduce unnecessary treatment and hence exposure to toxic drugs and significant cost to the health-care system. Thus, it remains unclear if more accurate diagnostic aids will increase the total number of patients initiated on treatment.

In this paper, we will review and discuss the newer POC testing platforms, including the Xpert MTB/RIF assay (Xpert; Cepheid, Sunnyvale, CA, USA), the lipoarabinomannan (LAM) lateral flow assay (Alere, Waltham, MA, USA) as well as revisiting aspects of traditional POC tests, such as smear microscopy. Other technologies such as the line probe assays and loop-mediated isothermal amplification (Eiken Pharmaceuticals, Japan) are briefly discussed, as they are relevant to the future development of POC platforms. The global resurgence of multidrug-resistant (MDR) and extensively drug-resistant TB, and untreatable forms of TB,[1, 4-6] of which less than 10% were diagnosed in 2011 and for which treatment costs are exorbitant and unsustainable, makes the need to develop POC tests with drug-resistant TB readouts a high priority. Presumed latent TB infection (LTBI) is an important facet of TB control in low burden settings, and ultimately diagnosis and treatment of LTBI will be required in all settings to eliminate TB. Thus, we also discuss potential POC diagnosis of drug-resistant TB and LTBI. An update is provided on material covered in recently published reviews about POC tests for TB.[7, 8]

Obtaining a Biological Sample and Other Considerations

Approximately 85% of the burden of TB is due to pulmonary TB. Diagnosis of pulmonary TB, particularly at primary care level, depends on obtaining an adequate expectorated sputum sample. However, in up to a third of TB cases, an adequate biological sample is not readily available or has a very low concentration of TB bacilli rendering the sample smear-negative (cases of extrapulmonary TB requiring sampling at secondary care level, sputum scarce (unable to produce sputum), and smear-negative patients]. The latter is particularly relevant in children TB-HIV co-infection where up to 50% of persons are smear-negative.[9] Thus, alternative techniques such as sputum induction, gastric aspiration, bronchoscopy, and organ aspiration or biopsy may be required to obtain an adequate sample. However, the availability of these techniques is severely limited in high TB burden settings. Attention has therefore been focused on alternative biological samples, such as exhaled breath and urine, which are more readily available even in children. Urine as a biological fluid for diagnostic testing is particularly attractive because it is sterile, less complex than other fluids such as sputum and serum, is readily available and TB-specific proteins and DNA may be found in the urine of patients with TB.[10, 11] Even though a biological sample may be successfully obtained, other characteristics including sample volume (e.g. Xpert MTB/RIF requires ≥1 mL) and time-to-testing have the potential to impact results. It should be borne in mind that the reference standard for TB, that is, culture, is a suboptimal gold standard (prone to bacterial overgrowth, excessive decontamination, cross-contamination, etc.) and appropriate analytical strategies and methods may have to be employed to deal with this when evaluating new POC tests. Finally, more consideration should be given in combining tests and developing testing algorithms to rule-in TB,[12] and to screening tests, including chest X-ray and computer-assisted diagnosis, to rule-out TB. These measures would help decrease the number of patients that require more expensive and complex tests, thus reducing burden on the patients as well as cost.

The Century Old POC Test: Smear Microscopy

Direct microscopy of Ziehl–Neelsen-stained sputum smears remains the mainstay of POC diagnosis in most TB endemic countries. The method is relatively rapid, inexpensive and has high specificity. However, direct Ziehl–Neelsen microscopy has low sensitivity (∼50–60%) and is less sensitive in children, in HIV co-infected patients and in patients with extrapulmonary TB.[13, 14] Decontamination using chemicals, including bleach and NaOH, and concentration of acid fast bacilli by centrifugation slightly improves the sensitivity.[15, 16]

An alternative to Ziehl–Neelsen -based direct microscopy is staining with a fluorescent molecule such as auramine O and visualization using a microscope with a mercury vapour bulb. This method is faster and improves sensitivity by ∼10% without a compromise in specificity, but its use has been limited by its higher cost, maintenance and darkroom requirements.[17, 18] More recently, light-emitting diode (LED) microscopy was introduced. This low-cost method offers the benefit of fluorescence microscopy without the associated operational requirements, including a dark room and special microscope. LED has a lifespan of up to 50 000 h and may even be battery-operated. LED microscopy is endorsed by the WHO and also for use in resource-limited settings.[19] However, there are limited data about performance of LED microscopy in HIV-infected persons. A recent large study using samples from TB-HIV co-infected persons, LED microscopy was cheaper, faster and performed, as well as Ziehl–Neelsen and fluorescence microscopy independent of the staining and processing methods used.[20]

Another recent WHO-endorsed approach to smear-based diagnostic work-up is front-loaded microscopy. Front-loaded microscopy addresses the problem of high dropout rates with focused collection of two or more sputum specimens during one clinic visit, and immediate referral and treatment of patients with positive smears.[21, 22] Front-loaded microscopy leads to a minor reduction in diagnostic sensitivity for the individual patient but is expected to improve case findings through enhanced quality of service and reduced dropout rates.[23, 24] Other novel approaches to smear microscopy include filtration techniques and magnetic beads[25] to concentrate samples, and an automated slide-reading prototype that captures images and uses computerized algorithms to count acid fast bacilli (Signature Mapping Medical Sciences, Inc., a subsidiary of Applied Visual Sciences Incorporated, Herndon, VA, USA).[26]

Nucleic Acid Amplification Tests

A key advantage of near-patient rapid testing is that it may allow for the initiation of treatment within a very short time frame. New phenotypic methods such as commercial liquid culture drug susceptibility testing,[27, 28] microscopic observation drug susceptibility,[28, 29] colorimetric redox indicator methods[30] and the nitrate reductase assay,[31] although approved by the WHO cannot provide results within a single clinic visit and also require extensive operator training, infrastructure needs and standardization before implementation.[32] By contrast, nucleic acid amplification tests (NAAT), which can rapidly detect small quantities of DNA through several different amplification methods, including the polymerase chain reaction, represent one of the most accurate known methods of detecting TB. With their improved simplification and automation in recent years, NAAT is becoming increasingly attractive candidate for use at the POC.

Overview of NAAT technologies

Several NAAT are commercially available for the laboratory-based diagnosis of TB. These include: (i) the Amplicor MTB, Cobas Amplicor and LightCycler Mycobacterium Detection kits (all produced by Roche, San Diego, CA, USA), the Amplified M. tuberculosis Direct Test (Genprobe, San Diego, CA, USA), the BD-ProbeTec-ET (Becton Dickinson Diagnostic Systems, Sparks, MD, USA); (ii) the Genotype MTBDRplus and GenotypeMTBDRsl (Hain Lifescience, Nehren, Germany), and INNO-LiPA Rif.TB (Innogenetics, Ghent, Belgium) line probe assays; (iii) the Capilia TB-Neo (Taunus, Numazu, Japan) rapid detection and speciation assay; and (iv) the Xpert MTB/RIF assay (Cepheid).[33-37] The second-generation Genotype MTBDRplus assay (version 2.0 format) is now available and has similar sensitivity to the Gene Xpert assay for the detection of smear-negative TB (∼70%),[38] and the assay is also available in the MTBDRsl format, which is the only commercially available NAAT that interrogates for second-line drug resistance (fluoroquinolones and aminoglycosides/capreomycin).[39] Although highly sensitive and specific for the detection of TB, and rifampicin and isoniazid resistance,[40] the line probe assays are functionally ‘open’ systems and therefore more prone to intralaboratory DNA contamination. Nevertheless, as the accuracy of several NAAT on respiratory specimens is generally similar,[33, 41] the main factors influencing their utility at the POC is their ease-of-use, rapidity of sample preparation and test completion, infrastructure requirements, and cost-efficacy. Although few comparative studies exist, Xpert MTB/RIF, which is accurate, relatively simple and recently endorsed by the WHO,[42, 43] arguably represents, out of all commercially available NAAT, the most promising for potential TB diagnosis at the POC.

The Xpert MTB/RIF assay

Xpert MTB/RIF is a largely automated real-time polymerase chain reaction assay able to detect M. tuberculosis complex DNA and resistance to rifampicin.[44] It performs optimally on expectorated sputum specimens, using a disposable single-use cartridge, and the test may be completed within 2 h, including a 15-min sample preparation step where sputum is homogenized using sterilizing sample buffer (Fig. 1).[35]

Figure 1.

Xpert MTB/RIF assay overview. Panel (a) shows the single-use, disposable assay cartridge, a Gene Xpert IV machine, and an example of a tuberculosis- and rifampicin resistance-positive result. Panel (b) shows an overview of the assay procedure. Images courtesy of Cepheid.

Accuracy and impact of Xpert MTB/RIF

Several large-scale trials have assessed the accuracy of Xpert MTB/RIF,[34, 35, 45-48] where its sensitivity for TB detection in smear-positive and smear-negative patients was found to be ∼98% and ∼75%, respectively,[41] although some studies from high HIV-prevalent settings have reported sensitivities in latter group to be as low as ∼50%.[34, 46, 49] The specificity of the assay for TB detection is ∼98%. For the detection of rifampicin resistance in regions with high disease prevalence, the sensitivity and specificity are ∼94% and ∼97%, respectively.[41] Importantly, a recent large study[45] has shown that the improved accuracy of Xpert MTB/RIF over that of smear microscopy (the most widespread diagnostic test for TB, including at the POC) can translate into an improvement in the time-specific proportion of TB patients initiating TB treatment, where about 90% of TB patients could initiate treatment based on their Xpert MTB/RIF result on the same day they provided a sample. In contrast, only about 67% of TB patients were diagnosed by smear microscopy and able to initiate treatment rapidly, as this usually happened the day after a sample was provided for testing. However, whether this advantage is sustained and whether earlier diagnosis translates into reduction in morbidity and mortality remains unclear. It is critical that MDR-TB treatment capacity be scaled up in parallel to the roll-out of the MTB/RIF assay.

Assay procedure

An overview of the Xpert MTB/RIF procedure is given in Figure 1b. Briefly, a twofold volume of sample buffer is added to sputum within the original collection container. The mixture is then manually agitated by shaking for about 20 s, before incubation at room temperature for ∼15 min. The mixture is agitated a second time approximately halfway through this incubation. Two millilitres of the resultant mixture are then pipetted into a single-use Xpert MTB/RIF cartridge, which is then loaded into the GeneXpert machine.[50] The purpose of the sample buffer is twofold: first, to homogenize the sputum sample and make it amenable for DNA extraction, and second, to render any potentially infectious TB bacilli non-viable.

Emerging information on Xpert MTB/RIF's suitability for POC use

Although no studies have evaluated the performance and feasibility of Xpert MTB/RIF when carried out in a primary care clinic, there is a growing body of evidence, which suggests it may hold considerable promise in this context (Table 1). First, the early landmark large-scale evaluation of Xpert MTB/RIF showed that when performed in reference laboratories by a laboratory technician, the test was accurate, did not have significant problems with cross-contamination or error rates, and required only minimal biosafety infrastructure.[35] Second, the procedure has, as measured using two different types of air samplers, been shown to not generate any aerosolized viable bacilli. Only when sample preparation was performed incorrectly, with the sample buffer incubation step omitted, were aerosolized bacilli detected; however, these levels were still lower than those generated by smear microscopy.[52] Of note is that in India and many countries in Africa, the preparation of sputum microscopy slides is often performed in a primary care clinic on a bench top and without a biosafety cabinet. Although test operators should wear a N95 mask for safety, these data suggest that Xpert MTB/RIF is likely a safer procedure than smear microscopy and may be performed in the absence of a biosafety cabinet. Third, two studies have demonstrated that Xpert MTB/RIF sample buffer provides an almost 8-log reduction in viability within 15 min of incubation, thereby rendering even highly smear-positive sputa effectively non-infectious.[44, 52] This also suggests that the cartridge waste generated by the procedure will not represent an infectious risk.

Table 1. Summary of important Xpert MTB/RIF studies relevant to its use at the point-of-care
Study descriptionContribution to the potential use of Xpert MTB/RIF at the point-of-careReference
  1. MTB/RIF, Mycobacterium tuberculosis/resistance to rifampicin; TB, tuberculosis.
This by Boehme et al. was first large-scale evaluation of the accuracy of Xpert MTB/RIF for the detection of TB.

Although the test was conducted in reference laboratories by a technician, this study demonstrated that the assay:

  1. — Required only minimal training
  2. — Was not prone to cross-contamination
  3. — Required minimal biosafety facilities
[35, 51]
This study tested if the Xpert MTB/RIF assay procedure (including sample preparation steps) resulted in the generation of aerosolized biohazards. This was compared with those generated by the preparation of smear-microscopy slides. The bactericidal activity of the assay sample buffer was also measured.
  1. — The Xpert MTB/RIF assay (including sample preparation) does not result in the production of culturable aerosol.
  2. — Only when sample preparation was performed incorrectly was culturable aerosol detected; however, this was still less than that produced by smear preparation.
  3. — A 15-min incubation in sample buffer rendered the TB in ∼98% of smear-positive samples non-viable.
[52]
This follow-up study by Boehme and colleagues examined the accuracy, feasibility and impact of Xpert MTB/RIF when performed in performed at a district and subdistrict health-care centre.
  1. — The accuracy of Xpert MTB/RIF performed in resource-scarce environments at a district and subdistrict level is comparable with that seen at a centralized reference laboratory
  2. — The assay may be performed with ease by a minimally trained technician.
  3. — Implementation of the assay can dramatically improve the proportion of TB patients placed onto early TB treatment.
[45]
This study examined the utility and cost of different diagnostics combined with Xpert MTB/RIF (either performed as a prescreen or downstream of Xpert MTB/RIF)

Although using data from gathered from Xpert MTB/RIF performed at a research laboratory, this study demonstrated that:

  1. — Smear microscopy followed by Xpert MTB/RIF had the lowest cost of diagnosis
  2. — A normal chest radiograph performed prior to Xpert MTB/RIF testing was able to rule-out TB in ∼1/5 patients
  3. — A normal chest radiograph performed in Xpert MTB/RIF-negative patients was able to rule-out TB in ∼1/4 TB patients
[12]
The study examined the utility of Xpert MTB/RIF for predicting the smear-status of TB patients. This is important for infection control and monitoring and predicting transmission.
  1. — Xpert MTB/RIF quantitative information (cycle threshold values) can be used with good rule-in accuracy to detect is someone would be smear-positive.
  2. — This is relevant because it allows patients who are potentially the most infectious to be targeted at diagnosis with interventions designed to reduce their infectiousness.
[57]

Fourth, as part of a follow-up study in their earlier large-scale evaluation of Xpert MTB/RIF performance, Boehme and colleagues showed that when performed at district or subdistrict level, assay performance was comparable with that seen at a reference laboratory, without any difference in feasibility or error rates, even though laboratory staff had little experience with methods other than smear microscopy.[45] This work suggests that, if the potential of the superior analytical performance of Xpert MTB/RIF is to be fully realized, the test needs to be placed in a context where the result can impact upon patient care as quickly as possible. Fifth, as Xpert MTB/RIF is more costly than smear microscopy (although it is more cost-effective[53]) and its negative predictive value is significantly diminished in HIV-infected patients,[46, 54] it might be useful in combination with other diagnostic tests. Recent work has shown that performing Xpert MTB/RIF only in smear-negative individuals can substantially reduce the overall cost of diagnosis,[12, 53, 55] whereas chest radiography can be used either as a prescreen for Xpert MTB/RIF testing or for the further downstream investigation of Xpert MTB/RIF-negative individuals.[12] As in some settings, these alternative tests are already accessible at a primary care level, it appears that where necessary, Xpert MTB/RIF will be able to complement the existing primary care test infrastructure. Finally, it is important to note that as in certain settings as much as 80% of drug-resistant TB is transmitted via person-to-person spread,[56] the disruption of transmission is key to limiting the spread of disease. Recently, research has shown that the quantitative information generated by Xpert MTB/RIF can be shown to identify, with modestly good rule-in accuracy, individuals who potentially represent the greatest infectious risk.[57, 58] Having this diagnostic available at POC means that such individuals can potentially be rapidly targeted for infection control interventions. There is also a good correlation between Xpert MTB/RIF Ct values and bacterial load,[40] suggesting that Xpert MTB/RIF may hold promise as a marker of disease activity or to monitor the efficacy of new immunotherapeutic interventions.

The need for further research

Placement of Xpert MTB/RIF at a primary care level will be more costly than a centralized approach simply because there are many more clinics than reference laboratories.[55] Many clinics would require their infrastructure (in terms of electricity supply and security, for example) and supply lines to be upgraded; however, a targeted and rational roll-out of Xpert MTB/RIF to those with the least access to centralized facilities should also be considered. Indeed, many of these aspects are presently been evaluated compared with smear microscopy in ∼1500 patients in an EDCTP (European and Developing Countries Clinical Trial Partnership)-funded, randomized, control trial being performed across five different African sites (Clinicaltrials.gov #NCT01554384), where the accuracy, feasibility and impact of Xpert MTB/RIF testing performed by a minimally trained nurse in a primary care clinic. Finally, it should be noted that this test by no means represents an ideal POC NAAT, and it may prove unfeasible to adopt it at POC in certain settings.[59, 60] Other drawbacks include limited affordability in high burden settings (capital outlay of ∼ USD17 000 and ∼USD10 per cartridge), lack of an isoniazid-resistance readout and thus the risk of large numbers of patients remaining inappropriately treated and at risk of acquiring MDR-TB, and suboptimal positive predictive value for rifampicin resistance in settings where the MDR-TB prevalence is less than ∼20%. Thus, in lower MDR-TB prevalence settings (including resource-rich and intermediate burden settings but also in high burden settings like South Africa), a rifampicin-resistant result may require confirmation using a second method, for example, culture or line probe assay. However, further data are required to get a more precise estimate about the positive predictive value of the MTB/RIF assay in such settings. Alternative technologies and platforms to enable potential POC testing for drug-resistant TB are outlined later.

Antigen Detection-Based POC Tests for Active TB

The search for suitable TB-specific diagnostic antigens is ongoing and has been extensively studied in a variety of biological samples (e.g. sputum, blood, body cavity fluids and urine).[61] A recent meta-analysis evaluated 47 studies using 12 single or combinations of TB antigens in different clinical specimens for pulmonary and extrapulmonary TB.[61] With the exception of LAM, TB antigen test sensitivity was as low as 2%, and specificity was suboptimal. However, antigen detection tests appear to offer a number of advantages over conventional diagnostics and have great potential for use as simple bedside tools. Antigen, as compared with whole M. tuberculosis organisms or TB-specific genetic material, is more likely to be detectable remote from the disease site in easily accessible biological fluids like urine. Antigen detection platforms, such as the lateral flow immunochromatographic assay (otherwise known as a strip test), require little or no sample processing to yield a rapid result. Unfortunately, despite the promise that antigen detection holds for POC diagnosis, available technologies have not yet delivered clinically useful results. Diagnostic accuracy measures vary widely, and with the exception of LAM, there are limited data about antigen-specific tests and none are currently in routine clinical use.[61]

LAM, a 17.3-kDa immunogenic glycolipid component of the mycobacterial cell wall, has been the most extensively studied antigen and offers potential clinical utility in HIV-infected patients with advanced immunosuppression in both inpatient or outpatient (antiretroviral clinic) setting.[62, 63] In HIV-infected patients, urinary LAM using an enzyme-linked immunosorbent assay kit (TB LAM enzyme-linked immunosorbent assay, Alere) had an overall sensitivity of ∼50%, increasing to 67% and 85% in HIV-infected patients with CD4 count <50 cells/mL from outpatient and inpatient settings, respectively[62, 63], and an overall specificity of 83–100%.[62, 64, 65] In addition, urine LAM correlated with bacterial burden[66] and may have prognostic utility by identifying TB HIV co-infected patients with the highest mortality.[67] Performance of the TB LAM enzyme-linked immunosorbent assay using sputum or induced sputum samples has also been evaluated. Although sensitivity improved to over 80%, the specificity dropped to under 50% likely due to cross-reactivity with Candida spp and normal oral flora containing LAM-like molecules.[62, 68] The enzyme-linked immunosorbent assay kit has now been superseded by the POC Determine TB LAM Ag strip test (Alere) (see Fig. 2), which is the first bedside TB test, provides a result within 25 min and will have a likely landing cost under USD3.5 in the first quarter of 2013.[62] Two initial evaluations in HIV-infected outpatients and inpatients showed similar diagnostic accuracy to the preceding TB LAM enzyme-linked immunosorbent assay and improved sensitivity (over the LAM strip test) when combined with sputum smear microscopy.[11, 69] However, specificity and interreader agreement decreased when using the manufacturer's suggested grade-1 cut point (see Fig. 2). Thus, we recommend the grade 2 cut point at the expense of a lower sensitivity but with a higher specificity. In addition, the test either alone or combined with urine-based Xpert MTB/RIF testing was useful in sputum-scarce diagnostically challenging patients.[10] Further study is ongoing to clarify cut-point selection and the impact on patient-important outcomes, for example mortality, when LAM is used to guide the early initiation of treatment.

Figure 2.

Urine lipoarabinomannan (LAM) strip test and reference scale card. The reference scale card, provided with each 100-strip packet, illustrates six cut-off points (visual grades 0–5) categorized by different band intensities appearing in the patient window[11] (Figure provided with permission of the European Respiratory Journal).

A number of antigens have also been evaluated in various compartments using non-sputum or urine samples, for example pleural fluid, cerebrospinal fluid, etc.[61] Urinary LAM had poor sensitivity and specificity in pleural and pericardial fluid, and in cerebrospinal fluid.[61, 70, 71] A limited number of studies with small numbers of patients have evaluated alternative diagnostic antigens using mainly ‘in-house’ assays with widely variable diagnostic accuracy.[61] These data reflect that antigen concentration in different body compartments are modulated by several factors including molecular weight, structure, and host degradation and processing. Thus, antigen performance may be highly variable between body compartments and sample specific. A possible solution may be to use combinations of antigens to improve overall diagnostic accuracy.[61]

The availability, low cost, rapid format and modest performance of urine LAM in HIV-infected patients, although not ideal, gives us hope that antigen detection may still provide a broadly applicable and effective POC test. For this to be successful, candidate antigens or combinations of antigens will need to be specific for M. tuberculosis, be produced in abundance, be excreted into the extracellular environment and be resistant to rapid degradation associated with the host inflammatory response.

Antibody Detection and Microfluidic Technologies

Antibody detection tests based on lateral flow or other immunochromatic formats are attractive POC candidates. These tests monitor the humoral antibody immune responses to antigens and have proven to be rapid and accurate in the context of HIV diagnosis.[72] A number of commercial antibody-based rapid TB tests are on sale,[73] but significant clinical validation is absent and diagnostic accuracy is, at best, poor. In a recent updated meta-analysis including 67 studies of commercial serological tests, Steingart et al. showed that study quality was generally poor, and estimates of sensitivity and specificity were inconsistent and imprecise.[18] These findings lead the WHO to proclaim a negative recommendation (its first) against the use of TB serological tests.[74]

The failure to develop antibody-based TB tests that meet clinical needs does not imply that such an approach should be abandoned. However, the heterogeneity in antibody responses from patient to patient suggests that a more complex multiplex approach is required.[75-78] Several novel promising antigenic targets have been identified.[75, 79, 80] A POC platform targeting several antigens, and co-developed by FIND (Geneva, Switzerland) and MBio Diagnostics Inc. (Boulder, CO, USA), using multiplex serology on a dot matrix readout will soon enter field evaluation studies. Microfluidics technology permits manipulation of fluids on a sub-millimetre scale enabling portability, affordability, easy disposal, user-friendliness, rapidity, multiplexing and feasibility with limited sample (reviewed in Lee et al.[81] and Wadhwa et al.[82]). A microfluidic platform seems well suited to a future multiplexed serological test.

POC Approaches for the Diagnosis of LTBI

In presumed LTBI, mycobacteria are not directly detectable, and therefore, diagnostic tests rely on measuring the presence of an adaptive immune response against M. tuberculosis.[83] The major drawback of this approach is that a detectable response may represent exposure without infection or infection that has been cleared. The tuberculin skin test has been mainstay of LTBI diagnosis for a century. This test is cheap and simple to apply, but there are several drawbacks including the need for a second test reading visit, subjective interpretation, cross-reactivity in persons BCG vaccinated after birth and no assessment of immune anergy. The discovery of regions of differentiation, that is, parts of the M. tuberculosis genome absent from most non-TB mycobacteria and BCG,[84, 85] facilitated the development of specific immunodiagnostic tests—the interferon-γ release assays (IGRA). IGRA are in vitro assays detecting interferon-γ secretion from RD1 (ESAT6 and CFP10)-specific T cells. Two IGRA are commercially available—the QuantiFERON-TB Gold In-Tube assay (QFT, Qiagen, Hilden, Germany) and the T-SPOT.TB (Oxford Immunotec, Oxford, UK). IGRA addresses several of the limitations of the tuberculin skin test; they have excellent specificity, but sensitivity (assessed in cases with active TB) is only 80%, and they are expensive and require specialized equipment and overnight incubation.[86, 87] Although data are variable,[88, 89] a recent meta-analysis showed that ability to predict short-term progression to active TB was similar to the tuberculin skin test ∼1–2%.[86]

Although several biomarkers have been studied,[90-94] the most promising interferon-γ alternative is the chemokine inducible protein-10 (IP-10), which has comparable diagnostic accuracy and higher sensitivity in HIV-infected persons[95, 96] (reviewed in Ruhwald et al.[94]). An IP-10 lateral flow platform can deliver quantitative results within minutes[97] and is stable in dried blood spots on filter paper allowing for letter-based sample transport for centralized analysis.[95] Another interesting approach to LTBI diagnosis, similar to that of measuring interferon-γ messenger RNA levels and transcriptional profiles,[98] is IP-10 and MIG detection at messenger RNA level.[99] Advances in microfluidics and lab-on-a-chip technology could enable a novel generation IGRA-like test devices that combine incubation of small volumes blood (e.g. from a finger prick) and detection in a disposable device.[100, 101] A novel skin test using recombinant ESAT6 and CFP10 antigens, C-Tb (Statens Serum Institute, Copenhagen, Denmark) recently entered phase III clinical trials in South Africa and elsewhere, and performs comparably with the QFT in unexposed volunteers as well as in HIV-positive and -negative adults with confirmed TB (Hoff ST, pers. comm.; see Fig. 3). Diaskintest (Pharmstandard, Ufa, Russia) is a similar product, but accuracy data are unavailable at the time of publication.

Figure 3.

C-Tb is a novel intradermal skin test whose administration is similar to a tuberculin skin test, containing recombinant ESAT6 and CFP10 antigens developed by the Statens Serum Institute, Copenhagen, Denmark.

Future and Novel Technologies for POC Testing and Developmental Challenges

Newer NAAT and other novel detection platforms and technologies

Besides conventional and real-time polymerase chain reaction, there are many different NAAT that use isothermal methods rather sequential heating and cooling cycles used in polymerase chain reaction. The isothermal NAAT reactions (reaction performed at a constant temperature) differs by enzyme type and number, restriction or nicking, whether they are RNA or DNA-based, the temperature at which the enzymes function, and the type of primer complexes used (see Textbox 2; reviewed in Liu et al.[101]). Several newer testing platforms have emerged.

Textbox 2. Types of isothermal NAAT reactions

Types of isothermal NAAT reactions

    Types of isothermal NAAT reactions
  • Transcription-mediated amplification (TMA)/nucleic acid sequence-based amplification (NASBA)
  • Signal mediated amplification of RNA technology (SMART)
  • Recombinase polymerase amplification (RPA)
  • Helicase-dependent amplification (HDA)
  • Rolling circle amplification (RCA)
  • Ramification amplification (RAM)
  • Loop mediated amplification (LAMP)
  • Cross priming amplification (CPA)
  • Smart amplification (SMART-AMP)
  • Strand displacement amplification (SDA)
  • Nicking enzyme amplification reaction (NEAR)/nicking enzyme mediated amplification (NEMA)
  • Isothermal chain amplification (ICA)
  • Exponential amplification reaction (EXPAR)

Loop-mediated isothermal amplification is a rapid (approximately 1 h) isothermal high throughput TB-specific NAAT using a fluorescent visual readout that has been evaluated using clinical samples.[102] Sensitivity in smear-negative TB is approximately 49%, large-scale evaluation studies have recently been completed, and a recommendation by the WHO is expected shortly. The recombinase polymerase amplification technology has been utilized for TB diagnosis by USTAR in China, with a user-friendly readout, and such technology may in the future also facilitate POC detection.[103] Simultaneous drug-resistant TB readouts are only possible with selected isothermal amplification methods, and although suited mainly to district hospital or reference laboratories, the technology could in the future be adapted for POC usage. By contrast, high-resolution melt technology, which relies on the discriminatory value of amplicon-specific DNA melting temperatures, may circumvent this drawback while also minimizing contamination and cost, and maintaining accuracy[104, 105] (see Fig. 4). However, disadvantages include the need for highly purified DNA, increasing costs and complexity, and varying DNA concentrations and pipetting errors impacting upon results.

Figure 4.

High-resolution melt analysis, a process whereby the amplicons are heated up slowly to denaturation with real-time monitoring of the decrease in fluorescence during denaturation. By comparing the melting profile of the sample with a reference, even a single nucleotide polymorphism can be detected, distinguishing a linezolid sensitive from a resistant isolate.[106]

A recent study described a new technology known as TB ID/R, which utilized blocked-primer-medicated helicase-dependent amplification followed by hybridization to probes arrayed on a modified silicon chip to detect mutations in the rpoB gene.[107] This technology permits rapid amplification and a potentially greater level of multiplexing for detection of MDR or extensively drug-resistant TB. The automated disposable cartridge format of the assay makes it amenable to a POC format.

Several new TB-specific and non-specific commercial platforms that lend themselves to POC NAAT-based detection are now available and in clinical trials, or in development (see Textbox 3 and Fig. 5a–d). Thus, there are several novel POC NAAT platforms that could, in the future, be applied to TB and other diseases in developing countries (Textbox 3). A challenge to developing NAAT-orientated microfluidic-based automated sample preparation platforms is the need for large input volumes to attain a low-enough level of detection.

Figure 5.

Newer automated devices that integrate DNA extraction and nucleic acid amplification test readouts include the Liat platform (a), Enigma (b) and the isothermal TwistDX platform (c). The Gendrive is a handheld sequencing device that provides simultaneous resistance profiling for tuberculosis (d). Handheld surface-enhanced Raman spectroscopy (SERS) sensors to detect explosives have been developed (e). SERS is a label-free way to detect unique signatures generated by photons of light striking a precoated metallic surface (f).

Textbox 3. Newer POC formats and novel approaches to facilitate POC diagnosis

Novel TB-specific POC NAAT platforms in development

    Novel TB-specific POC NAAT platforms in development
  • TrueNAAT Mycobacterium tuberculosis (MTB) test by Bigtec labs and Molbio Diagnostics (India) detects TB DNA by processing sputum on a semi-automated, battery-operated, portable device.[85]
  • Genedrive is a handheld sequence-analysis device (Epistem, UK) that analyses biological samples.[86]
  • B-SMART (Sequella, Inc., USA) rapidly detects the presence of TB with resistance to the four first-line anti-TB drugs directly from sputum.[87]

Newer non-specific POC NAAT platforms that have been developed

    Newer non-specific POC NAAT platforms that have been developed
  • The LIAT analyzer automates all the nucleic acid testing processes without further operator intervention. The test is completed within 20 min.[108]

Other similar but alternative platforms include:

  • MDx (Biocartis Molecular Diagnostics)[109]
  • FL/ML (Enigma)[110]
  • R.A.P.I.D. BioDetection System (Biofire Diagnostics Inc; formally the Idaho Rapid Tech System)[111]
  • Twista (TwistDX; uses RPA) with completion of DNA amplification in as little as 10 min using a portable system[112]
  • Genie II (OptiGene)[113]
  • SAMBA (Diagnostics for the Real World Ltd)
  • BART (Lumora, UK)[114]

Alternative detection technologies that may facilitate POC testing

    Alternative detection technologies that may facilitate POC testing
  • NAAT product-based lateral flow readouts incorporating antigenic labels (e.g. BESt Cassette detection platform),[116] and other simple NAAT readouts including turbidity, bioluminescence, electrochemical and calorimetric microparticle-based systems.[109]
  • Detection of SNP encoding rifampicin resistance using gold nanoparticles and a calorimetric readout.[110]
  • M-chip platform using an optical density detector.[112]
  • FLISA that allows detection of fluorescent signals three orders of magnitude below ELISA.[113]
  • Bio-sensors using electrochemical, piezoelectric quartz crystal, and magneto elastic readouts
  • SERS (surface enhance Raman spectroscopy) is a label-free way to detect unique signatures generated by photons of light striking a pre-coated metallic surface. Hand held SERS sensors to detect explosives have been developed (see Fig. 3).
  • Electrochemical detection systems using nanoparticle filaments or wires, or aptamers.

Aptamers are alternative target-specific detection molecules, which in contrast with conventional protein-based antibodies are nucleic acid or chemical ‘antibodies’ with higher sensitivity and specificity for target antigens (see Fig. 6). We have recently successfully raised aptamers to TB-specific antigens and detected these in clinical samples.[116] Several alternative detection platforms, some using label-free methods, and alone or in combination, are being explored for potential application to POC TB diagnosis, and these are listed in Textbox 3 (see also Fig. 6e,f). It remains to be seen whether these novel technologies will translate into affordable and impactful diagnostic tests. Design of testing platforms that will simultaneously detect several infections simultaneously, for example, TB, HIV, malaria, and pneumocystis and pneumococcal pneumonia will likely facilitate impact and algorithmic disease management.

Figure 6.

Model of an aptamer (three-dimensional intricate nucleic acid ‘anti-body-like’ structure) with high specificity complementary binding to human immunodeficiency virus-associated GP120 protein (left panel;[115] aptamers are generated by the SELEX procure where an oligonucleotide library is used to generate target-specific aptamers, which are amplified after successive rounds of elution (right panel). PCR, polymerase chain reaction; ssDNA, single-stranded DNA.

Detection of volatile organic compounds and mass spectroscopy approaches

M. tuberculosis metabolites can be detected as volatile organic compounds in breath using gas chromatography/mass spectrometry-based analysers.[117] Although several volatile organic compound-based fingerprints have been identified using POC prototype devices,[118, 119] specificity, suboptimal positive predictive value and low concentrations of metabolites (parts per trillion) are hurdles to development. Presently, the lack of antigenic targets that can be applied to available detection technologies represents a bottleneck to POC development. We have identified several novel TB-specific biomarkers in the urine of TB patients using a mass spectroscopy approach. Although commercially available mass spectrometry-based methods of diagnosis TB exist, such as the Ibis platform (Abbott, USA), there are limited data about their accuracy,[120, 121] and their infrastructure requirements generally exceed that available to most district laboratories. At the time of publication the product was temporarily withdrawn from the market.

Reporter enzymes—detection of markers amplified by M. tuberculosis enzymes

A promising alternative to smear microscopy is a detection of a metabolic signature of M. tuberculosis. Xie and colleagues recently developed a panel of fluorogenic substrates metabolized by BlaC, a TB-specific β-lactamase.[122] A prototype under development by Global Biodiagnostics, Temple, TX, USA (gbdbio.com) demonstrated high specificity and sensitivity (10 cfu in unprocessed sputum) and enabled detection in a homemade box containing a LED, filters and a mobile phone camera.[123] The urease breath test is used to detect Helicobacter pylori-associated stomach ulcers where the enzyme hydrolyses labelled 13C-urea to 13CO2 gas, which is readily detectable in the exhaled air with a portable infrared spectrometer (e.g. Breathtek, Otsuka). A similar approach has been explored in TB,[124] and the breath test correlated with bacterial load in a rodent model.[125]

Challenges for the development and roll-out of POC tests

Several of the earlier mentioned established and novel platforms may have limited accuracy and impact on TB diagnosis because of several technical and non-technical challenges (outlined in Textbox 4). Furthermore, for diagnostic tests to have an impact, robust health-care systems are needed with good supply chain management, and these challenges need to be better represented in medical and nursing teaching curricula. Finally, there is a considerable shortfall, amounting to over 6 billion dollars, to fund TB research for the next 5 years (2011–2015).[126]

Textbox 4. Challenges to the development and roll-out of point-of-care TB tests

Technical factors

    Technical factors
  • Antibody profiles in patients with active TB overlap with patients with LTBI, previous TB or those with significant exposure to NTM.
  • Antigens may be differentially expressed in different body compartments.
  • Test accuracy may vary in HIV-infected versus uninfected persons and those with pulmonary versus extrapulmonary (EP) TB.
  • HIV-infected persons and those with compromised immunity may be colonized with NTMs.
  • Sensitivity is often suboptimal when using immunoassays in lateral flow format.
  • Samples like sputum may not be readily available in ∼20–30% of cases because of EPTB or sputum scarce PTB, and in HIV-infected persons and children.
  • Lack of suitable or limited antigenic targets.

Non-technical factors

    Non-technical factors
  • Lack of information on target product profiles.[8]
  • Lack of private investment because of a perceived lack of return.
  • Lack of regulatory standards and their international harmonization.
  • Variable quality of diagnostic services and lack of appropriate study design and scientific rigor in evaluation studies.
  • Lack of funding for developmental and biomarker research.
  • Lack of dedicated funding and infrastructure for innovation, field studies, and taking products to market from charities and governments (particularly in developing countries).
  • Lack of networking and coordination between funding agencies, academic institutions and industry partners.
  • The ‘one size fits all’ approach to intellectual property and patent fees for diseases affecting developing countries.
  • Failure to appreciate that technologies applied to TB may be applied to diseases relevant to resource-rich countries, for example, COPD, pneumonia, ICU-related infections.
  • Red tape in the approvals process in various countries and lack of harmonization thereof.
  • Development of TB-related tests suited to resource-rich settings.

Conclusion

Although several factors converge to facilitate effective TB control, diagnosis represents an important unmet research need that can, like in the case of malaria and HIV, enhance control. Significant strides have been made in developing POC technologies for TB diagnosis. More recently, tests like the Gene Xpert and the LAM lateral flow assay hold promise to transform the diagnosis of TB as we know it. Xpert has the potential to be decentralized to clinic level and, in appropriate settings and if found to be feasible, may even be utilized for community-based intensive case finding. It is unlikely that a single test will have optimal performance in different settings and clinical categories of TB, and a selection of context-specific tests will likely evolve. However, new technology is likely to make future POC tests more affordable and user-friendly while retaining accuracy. There are several newer NAAT and other user-friendly detection platforms, including surface-enhanced Raman spectroscopy and electrochemical detection, that hold promise to transform POC testing. Lack of suitable antigenic targets and alternative user-friendly detection platforms remain an obstacle to the development of POC tests. Sustained funding and a robust developmental, evaluation and demonstration infrastructure, especially in several emerging economies including India, China, Brazil and South Africa, needs to be further developed to evaluate technologies that are in the pipeline. Promising new candidates will have to be evaluated in phases II and III studies, but tests will also need to be prioritized based on their impact on morbidity and mortality in high burden settings. In parallel entrepreneurial collaborative funding models that incentivize both industry and academic partners are required, and logistic, technical, regulatory, educational, advocacy-related and economic factors that may limit the uptake of new diagnostic tests will need to be addressed.[127] The impact of POC tests will be nullified if preventative and treatment-related factors, including reliable drug delivery, are not addressed in parallel. Finally, more funding will need to be devoted by both government and international agencies to drive innovation and bring new technologies to the market.

Ancillary