The Emerging Threat of Drug-Resistant Tuberculosis in Southern Africa: Global and Local Challenges and Solutions: Summary of a Joint Workshop by the Institute of Medicine and the Academy of Science of South Africa (2011)

TB patients progress from exposure to cure in discrete steps: infection, symptoms, clinic visit, diagnosis, treatment onset, conversion from an active infection to a noncontagious state, and cure. To decrease transmission, the time between each of these stages needs to be reduced. The workshop addressed challenges in the capacity of the health care system in a high-burden country such as South Africa to meet the need for rapid and effective diagnosis of TB. Diagnostics not only can help reduce the interval between presenting at a clinic and diagnosis, but also can be used within a comprehensive program to address and perhaps reduce the earlier and some of the later intervals in the progression of TB. Thus diagnostics can play a key role in curbing the epidemic if they are dramatically improved and

used proactively. This chapter summarizes the workshop presentations and discussions addressing the need for rapid diagnostics, progress on point-of-care diagnostics, the challenges of laboratory capacity in South Africa, and the use of biomarkers to diagnose TB.

Although some have suggested that greater than 70 percent detection and cure rates for TB will constitute success, van Helden suggested that in fact, this will not be enough. If only 70 percent of infected individuals are detected and only 70 percent of that group is cured, the overall cure rate will be the product of those two numbers, or 49 percent. Furthermore, even infected people whose disease resolves will spread the organism before they are diagnosed. A mathematical model suggests that if diagnosis takes 40 days, a rising epidemic will continue, whereas diagnosis within 6 days or less will result in the epidemic’s decreasing.

The need for patients to return to the clinic is a factor. Studies of the current system for diagnosis indicate that 20 percent of patients fail to return after their first visit to a clinic and thus are lost to follow-up. A point-of-care diagnostic device (see the discussion below) could prevent those patients from being lost to follow-up at that early stage of diagnosis and treatment of their TB. The algorithm used suggests that typically when patients present to a clinic or TB facility, a point-of-care test that may be 40–50 percent sensitive is administered. Positive patients are absorbed into therapy, and it may not be necessary that they provide another sputum sample. Negative patients that are highly clinically suspicious should be managed according to the normal National Health Laboratory Service algorithm for that country, suggested van Helden.

In considering what types of diagnostic tests should be developed and adopted within a high-burden country, a number of criteria come into play. An ideal needs assessment may have to be evaluated and judged against what is realistic.

To illustrate this point, van Helden used the example of the Ziehl-Neelsen stain test,² which is used to evaluate sputum. Many argue that this test is appropriate because it is effective, relatively quick, and inexpensive. Others argue that it is insufficient, and a test that is better on all criteria

(more effective, simpler, and even less expensive) should be adopted. Still others argue that a test that is more effective and more rapid than Ziehl-Neelsen should be adopted even if it is more expensive. And, as discussed in this chapter, there are strong arguments that anything less than the ability to make a point-of-care diagnosis of drug-resistant TB ultimately will not succeed against the disease.

A variety of criteria can be applied in considering the desirability of a diagnostic test. Speed, cost, sensitivity, and specificity are among the most important; one issue is whether speed, sensitivity, or specificity should be paramount. Current tests generally are 50 percent sensitive and usually are not performed at the point of care. On the other hand, rapid point-of-care diagnosis could result in very few TB cases in 5 years’ time; ultimately, such lowering of case burden would cost less overall. Indeed, the development of new diagnostics will be essential to stem the TB epidemic; saving money on cheap, ineffective diagnostics will result in the epidemic’s continuing. In considering implementation and efficacy, issues include number of manipulations; repeats (including client returns); invasiveness of sample taking; skills, instruments, and personnel required; and central versus decentralized laboratory capacity and logistics.

The problem with many diagnostic methods currently in use is that the answer they provide can be both yes and no, said van Helden (Box 5-1 describes some of these methods). For example, genotype information can reveal that an inhA mutation is present, but in this case isoniazid should not be excluded and is still a potentially usable drug. However, when a KatG mutation is present, treatment with isoniazid is not recommended. Genotyping and phenotyping will give different results, and even phenotypes are not yet fully understood.

Results from testing in miners reveal some of the negative consequences of reliance on simple diagnostics. The diagnoses were unreliable, and regimens were implemented in a formulaic way, resulting in the development of MDR and XDR TB. The epidemic was amplified by transmission of MDR TB, after which some individuals progressed to XDR TB.

Other methods need to be considered to yield informative diagnostics. Some of the antibody tests used in the 1990s had a sensitivity of 30 to 70 percent and were simple to use. The sensitivity of lateral flow devices may be 77 percent, with a specificity of 93 percent. According to van Helden,

more time and effort should have been invested in testing and developing these methods, which now need to be reconsidered.

The presence of antibodies implies the presence of antigens and could be useful in detecting infection. The Foundation for Innovative New Diagnostics (FIND), together with a number of researchers, is investing in the advancement of lipoarabinomannan (LAM) tests. Lipids and mycolic acids also are potentially interesting antigens because they drop very quickly as

the bacterial load disappears. Studies of urine-based tests for LAM show that sensitivity is lower in smear-negative patients but increases as CD4 count decreases,³ making these tests potentially highly useful in the South African context.

According to Jacobs, intensified case finding (see Chapter 3) is essential in South Africa to reach MDR TB cases that are not yet being treated. Such case finding would be dramatically enhanced by tools enabling point-of-care diagnoses, but important questions about such tools remain unanswered.

Jacobs remarked that the ideal point-of-care diagnostic test will produce a diagnosis in 1 to 24 hours, be inexpensive, distinguish drug- susceptible TB from MDR and XDR TB, be accurate and easy to perform, and give fractional drug resistance. One simple approach for a drug susceptibility assay relies on the luciferase reporter phage (LRP), also called the “turn on the light” assay. A TB sample is divided into five tubes, and different drugs are added to four of the tubes, with the fifth serving as a control. If a drug works, it kills the cell and the light does not come on; if the cell is drug resistant, the light comes on. Despite skepticism from reviewers, this has proven to be a simple and inexpensive test. Important questions for this assay, according to Jacobs, are

• What are the accuracy and speed of reporter phages for susceptibility testing of clinical M.tb. isolates?

• What is the feasibility of performing a phage-based susceptibility assay in developing countries?

One evaluation of the use of LRPs in Mexico and South Africa found that they provided susceptibility results with an overall accuracy of 99 percent and had a median turnaround time of 3 days, making them the fastest phenotypic method available. The future direction of LRPs is toward a 1-hour test, extended for XDR TB, as well as the development of reporter phage assays for susceptibility testing with second-line drugs.

Recently, a reporter phage test based on green fluorescent proteins (GFPs) was developed by Jacobs and his collaborator Graham Hatfull from

the University of Pittsburgh. This test has been shown to have a number of potential advantages, including

• expected low cost,

• monitoring of individual cells for drug resistance,

• fixation of cells for enhanced safety,

• analysis of mixed cultures,

• potential multicolor internal controls, and

• alternative detection systems.

The initial GFP test took about 24 hours to provide a response. It could be used to assess drug susceptibility and detect rifampicin and streptomycin resistance. Further work is needed to develop assays that can produce results in cells in 1 hour. Jacobs stated that GFP tests using fluoromyco-bacteriophages have considerable potential. They can provide a simple, inexpensive, quick, and reliable method for drug susceptibility testing and offer multiple platforms for detection. Efficient recovery of M.tb. from sputum is needed, as are second-generation phages with enhanced fluorescence and specificity. Current studies evaluating the efficacy of these tests are under way in Durban, South Africa.

Coetzee observed that South Africa’s response to MDR TB has been limited by the unsatisfactory performance of its laboratory services and inadequate human resources. In 2009 nearly 1 million cultures were performed in South Africa, most of which were followed up with drug susceptibility testing, and these tests identified 9,000 new MDR TB cases. Yet many communities still are not receiving adequate laboratory services. The National Health Laboratory Service now has well over 100 MGIT (mycobacteria growth indicator tube) machines, but little progress is being made in identifying all MDR TB cases.

In 2008, the World Health Organization (WHO) endorsed the use of line probe assays (LPAs) capable of detecting resistance to rifampicin and isoniazid (MDR TB is defined by resistance to both of these drugs). The new WHO TB guidelines, which became operational on April 1, 2010, have led to a doubling of laboratory investigations. The roll-out of this test in South Africa is currently being organized. As of the middle of 2009, 5 laboratories were able to conduct a substantial volume of LPA tests. It was decided that by the end of 2010, 20 more secondary (not academic)

laboratories would be rolled out; 11 of these are already functioning. The objectives of the roll-out include

• achieving rapid diagnosis of MDR TB;

• providing effective treatment to patients at the appropriate time;

• preventing further resistance to anti-TB drugs;

• preventing further TB transmission;

• decreasing the cost of treating TB by reducing unnecessary transmission through earlier diagnosis, as well as by preventing the development of drug resistance, which is more expensive to treat; and

• realizing financial savings by eliminating drug susceptibility tests from the diagnostic process (it is expected that these savings will fund the implementation of LPAs).

A procedure was developed for the early detection of MDR TB. Every new smear-positive patient in South Africa will receive an LPA test to rule out MDR TB. A 7-day sputum will be taken at the facility, and an MGIT culture will be performed on patients who are still symptomatic. Positive patients will again receive an LPA test.

Because the existing infrastructure could not accommodate new laboratory space, modular park home units were specially designed to be suitable for use as laboratories, and specifically for polymerase chain reaction (PCR) testing. Staffing these facilities has presented the biggest obstacle to rolling out the 20 additional laboratories. A new category of laboratory staff called “TB technician” has been approved, and a 2-year training program for these staff is due to commence. Sixty new laboratory staff were employed during 2009, and a further 60 will join the National Health Laboratory Service during 2010.

One of the concerns with regard to the roll-out of additional laboratories is quality control. Currently, no international quality assurance program is in place for LPAs, and only interim measures have been taken to accommodate quality control in laboratories. The biggest technical problem has been the variability of LPAs; to address this problem, a scanner was developed that automatically reads and quantifies the results and actual mutations.

One workshop participant cited another concern with the implementation of the LPA test: to the extent that many more undiagnosed cases of MDR TB will be diagnosed through use of this test, it will be necessary to consider how the health care system will treat those additional cases.⁶

Community-based care beyond present practice may be a potential solution (see Chapter 6). Current policy is that all MDR TB cases must be admitted until culture conversion is seen. In KwaZulu-Natal Province and the Western Cape, however, pilot studies are under way on community-based care that will be decentralized by strengthening existing hospitals and other TB facilities and by creating satellite centers in communities, with teams that travel throughout an area. Integration of MDR TB and HIV care will take place at the primary care level.

Parida reported that sub-Saharan Africa experiences 12 times more deaths from TB every hour than Europe. More needs to be known about the outcomes associated with exposure to M.tb. to devise interventions, he suggested, and one way to learn more is to reduce the gap between the research laboratories and clinics. There are important differences in immune response between individuals who are exposed to TB and protected from the disease and those who develop active disease. Particularly for people coinfected with M.tb. and HIV, the design and testing of new TB vaccines, drugs, and diagnostics will be critical.

The project Biomarkers of Protective Immunity against TB in the Context of HIV/AIDS in Africa, which involves 15 partner institutions from Africa, Europe, and the United States, has been seeking to identify correlates of protective immunity and host biomarkers of TB that have prognostic potential for delineating disease susceptibility and protection.⁸ The project’s objectives are to

• investigate differences in immune response between individuals who are exposed to TB and protected from the disease and those who develop active disease; and

• coordinate to promote the design and testing of new TB vaccines, drugs, and diagnostics, especially in areas with high HIV infection rates.

Particular attention is being paid to people coinfected with both M.tb. and HIV, whether or not they are receiving antiretroviral therapy.

The different platforms for biomarkers are immunologic, transcriptomic, proteomic, and metabolomic. Most likely, a combination of these markers will prove to be useful. Specific biomarker needs in the context of TB are

• surrogate markers of immune protection for assessing potential vaccine candidates,

• a surrogate marker of bacterial clearance (the clinical end point) for assessing potential drug candidates,

• markers of relapse,

• markers of treatment failure (drug resistance),

• diagnostic markers,

• markers for infection, and

• prognostic markers for reactivation of disease.

The project has been structured around five different cohorts, or work packages (WPs). WP1 is focused on pathogens; WP2 is focused on host responses to infection; and WPs 3, 4, and 5 are prospective cohorts in African field sites used to study the natural history of TB. WP1 and WP2 are based on cross-cutting methods carried out in the northern laboratories in collaboration with the field sites to narrow down the biomarkers to 25–50. As noted, the focus is on pathogen and host, respectively, which then can be assessed in longitudinal prospective studies in the field sites. WP5 is being used to assess the immune responses to BCG vaccination and the effects on disease progression. About 17,000 individuals from different age groups with a varied spectrum of TB with or without HIV in five African countries are being followed. The group characteristics are as follows:

• 898 HIV-negative TB index patients,

• 3,935 HIV-negative subjects with latent TB infection free of any disease symptoms and signs,

• 6,363 adolescents with latent TB infection free of any disease symptoms and signs,

• 894 HIV-positive subjects with latent TB infection and 304 HIV-positive and TB-positive patients, and

• 5,663 neonates plus 200 children following BCG vaccination.

At the start of the project, the hypothesis was that different clinical strains of M.tb. would show differences in gene expression levels. At that time, this hypothesis faced considerable skepticism. After a series of discussions with experts, two of the most divergent strains in the phylogeny—one from Gambia (a strain of M. africanum that is prevalent in West Africa) and the other from Uganda (an M.tb. strain of the same phylogeny of Euro-American lineage to which the laboratory strain H37Rv and clinical

strains such as CDC1551 belong)—were compared with the H37Rv reference strain. The analysis found striking geographic variations among the clinical strains, with the implication that effective universal vaccines and diagnostics cannot be produced. Rather, the uniqueness of the strains must be considered in the development of vaccines and must be reflected in the treatment of drug-resistant TB.

The project addressed about 100 genes, focusing on latency and reactivation, and made 86 novel antigens that were studied in five sites. The antigens were initially screened in a 7-day whole-blood stimulated assay, with interferon-gamma (IFNg) as readout in the antigen-stimulated supernatants. Then the antigens were subjected to 42 cytokines in an effort to uncover combinations that would result in biomarkers. It was speculated that a combination of antigens and cytokines might lead to a possible biomarker profile. Host biomarkers in disease and protection were assessed using microarray analysis of total ribonucleic acid (RNA) from whole blood from TB patients, tuberculin skin test–positive (latently infected) subjects, and tuberculin skin test–negative (noninfected) subjects. An earlier study with Caucasian subjects had identified three host biomarkers—CD64, LTF, and Rab33A—that could differentiate between TB patients and healthy contacts.

The TB contact study (cohorts WP3 and 4) followed the subjects over a 2-year period, with a focus on those both infected and clinically healthy. These subjects were followed up clinically over the 2 years, with blood samples being taken at three time points—0, 6, and 18 months. The objective was to assess immune responses over time using all the platforms described earlier and to identify biomarkers of disease susceptibility and protection, comparing subjects who contracted disease during the course of this followup period with those who remained healthy without succumbing to disease progression. Based on studies performed earlier, it was expected that 5 to 10 percent of subjects would develop TB during the 2-year period. The current follow-up status shows only 63 secondary cases from among almost 4,000 subjects in the HIV-negative cohort. However, the reactivation rate is about 2 percent rather than the 5 to 10 percent expected (which, according to Parida, has been a convention in the field that has never been verified). The ramifications of the TB contact study are complex and time and resource intensive, involving multiple teams working in tandem. There currently are 22 (about 2.5 percent) secondary TB cases based on progressors in WP4 (the HIV-positive cohort).

The current status of the project is as follows. Recruitment of the cohorts has been completed, and follow-up will be concluded in October 2010 (Parida and Kaufmann, 2010). Additional recruitment has been initiated and is ongoing to obtain more secondary cases, assay qualification is being completed at all sites, and large-scale production of antigens has been achieved. There are plans to perform analysis centrally on all second-

ary cases and matched controls at the end of the follow-up period and to validate patterns showing an association with protection. A complementary analysis using longer-term assays to delineate soluble cytokine expression patterns associated with protection also is planned.

This biomarker study for drugs and treatment outcomes has looked at the normal patient therapy seen in clinics. The aim of the study has been to make it possible to differentiate among patients at the time they seek medical help so they can be assigned directly to the correct regimen. According to Parida, the objectives include

• bringing together basic scientists from the laboratories and clinicians from clinics in the communities and hospitals through effective communication and interaction;

• achieving a holistic understanding of resistance, susceptibility, and protective immunity, taking an open-minded, global approach;

• coordinating initiatives and finding synergies within the scientific community, including interdisciplinary approaches and cross-fertilization of ideas;

• establishing global, shared, and comprehensive biorepositories;

• aligning vaccine trials and intervention studies;

• creating an open access policy and enforcing it through the buy-in of all stakeholders;

• generating public–private–philanthropic partnerships; and

• enhancing functional collateral/symbiotic interactions between basic science and real lives in clinics to translate research knowledge into products of public health importance.

The following issues require further consideration:

• the natural history of infection in the context of drug-resistant TB and immune responses; and

• differential gene expression patterns of MDR and XDR TB strains, with virulence depending on the host environment.

Parida also noted that a number of conventionally accepted “ dogmas” in the field have not been systematically questioned. According to Parida, evidence-based studies are needed in areas that include microbial persistence, immune evasion mechanisms, strain diversity, limitations of animal models, extrapolations of experimental results, ways to bridge the experimental results with clinical observations, and integration of multiple observations rather than a reductionist approach. Ultimately, lessons learned in the clinical setting need to be brought back to the laboratory to close links and fill the gaps.

This page intentionally left blank.