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)

Systematic surveillance and tracking of drug-resistant TB helps in understanding the overall burden of the disease and can inform research and practice in diagnosis, treatment, and infection control. Speakers at the workshop described various approaches being taken to advance the tracking of drug-resistant TB in South Africa. This chapter summarizes those presentations. The first section reviews the use of genetic fingerprinting methodologies to understand the genotype and physiology of the various drug-resistant TB strains found in South Africa. The second section describes a clinical screening tool that has been developed to intensify TB case finding. The final section addresses the need for information systems to increase laboratory capacity.

In South Africa and most of the world, three standardized methods are being adopted to classify TB strains by genotype. Warren and his colleagues have been applying these methods in numerous settings throughout South Africa to elucidate the mechanisms driving the drug-resistant TB epidemic in different regions.

An early study revealed a complex population structure of drug- resistant strains in the Western Cape Province of South Africa, with some strain genotypes being highly dominant (i.e., Beijing, LAM, and Low Copy Clade) (Streicher et al., 2001). A more detailed examination of this data set demonstrated the evolutionary history of an outbreak strain (DRF 150) (Victor et al., 2007). The study found that isoniazid resistance occurred first, followed by streptomycin resistance. With the subsequent introduction of rifampicin, many of these strains acquired rifampicin resistance on different occasions. Most important, the strains that had acquired additional mutations conferring resistance to ethambutol and pyrazinamide were the most dominant.

At about the same time, a preliminary study found that the Beijing/R220 strain was widespread in the Western Cape (Johnson et al., 2006). Warren and his colleagues conducted a follow-up study of the drug-resistant strains in the Western Cape to determine whether there had been a change in the drug-resistant TB epidemic from 2002–2003 to 2005–2006 ( Johnson et al., 2010). The number of identified cases was found to be doubling approximately every 8 years. Of significant concern was their observation that 90 percent of all drug-resistant cases tested were smear positive and thus indicative of being highly infectious, leading to extensive transmission. Stratification of the data by drug resistance pattern showed that the observed increase was strongly driven by MDR TB. The Beijing/R220 strain contributed to 42 percent of this increase, with a doubling time of 2.4 years (Johnson et al., 2010).

A third outbreak strain was identified in an urban setting in Cape Town, South Africa (van Rie et al., 1999). This strain (strain U) is a member of the Beijing genotype and showed a doubling time of approximately 5 years. Thus, as in KwaZulu-Natal (see Chapter 2), the increase in drug-resistant TB in the Western Cape Province was driven by a small number of specific strains. The implication is that the current TB control program is unable to prevent ongoing transmission of these highly resistant strains.

Studies from the Eastern Cape are limited. An initial study showed the spread of an atypical Beijing strain (Strauss et al., 2008). This finding was

surprising, since it had been assumed that the strain was not spreading (as was observed in the Western Cape). According to Warren, the collation of genotyping data supports the notion that the population structure of drug-resistant strains differs among the provinces of South Africa. However, insufficient data are available from the Northern Cape and Limpopo Provinces, as well as Namibia and Botswana, with which to make accurate assessments of the population structure of drug-resistant TB in southern Africa.

Since the outbreak at Tugela Ferry was disclosed, XDR TB has been identified in all the provinces of South Africa, particularly in the Western Cape, which has a very high incidence of drug-resistant TB. DNA finger-printing of MDR and XDR TB strains has shown that the main outbreak strains seen in the Western Cape (discussed above) are those that evolve into the pre-XDR and XDR group. At first glance, the DNA fingerprinting data suggest ongoing transmission of pre-XDR² and XDR TB—a situation analogous to the Tugela Ferry outbreak. Warren pointed out, however, that genetic fingerprinting has certain limitations, and not all interpretations are necessarily correct. The discriminatory power of the DNA fingerprinting method can be improved by sequencing of the target genes conferring drug resistance. Using this approach, Warren and his colleagues showed that MDR TB strains with associated pyrazinamide and ethambutol resistance are spreading and that most XDR TB is acquired. According to Warren, these findings clearly indicate that patients with MDR TB are not being managed aggressively enough. This conclusion contrasts with the case of the Tugela Ferry outbreak, for which ongoing transmission due to poor infection control (hospital and community) was identified.

Strains with isoniazid resistance and associated cross-resistance to ethionamide have become the dominant population in the XDR TB group. This may be explained by the fact that patients with MDR TB received a standardized treatment regimen that included ethambutol, pyrazinamide, ethionamide, ofloxacin, and amikacin or kanamycin. Only two of these drugs would have been active, given that MDR TB was found to be strongly associated with resistance to ethambutol and pyrazinamide and that cross-resistance to ethionamide was common in the Western and Eastern Cape Provinces. Warren suggested that cross-resistance between isoniazid and

ethionamide would be easily identified using the line probe assay (LPA) diagnostic test and that this information (which already exists) should be made available to expert MDR TB clinicians.

Warren concluded by noting that XDR TB is a major component of the drug-resistant TB epidemic in the Eastern Cape Province. In the MDR TB group, the atypical Beijing strain is seen in only 26 percent of the population, and is strongly selected in the pre-XDR and XDR TB population (accounting for approximately 90 percent of cases). This selection is driven by isoniazid/ethionamide cross-resistance and the associated standardized treatment regimen (see above). Of further concern is the significant association between the atypical Beijing strain and aminoglycocide resistance in pre-XDR and XDR TB strains in the Eastern Cape. In these strains, aminoglycoside resistance occurs through a mutation in the rrs gene. This mutation also leads to capreomycin resistance, thereby potentially compromising the XDR TB treatment regimen. Finally, Warren showed that pre-XDR and XDR TB strains are moving between provinces, emphasizing the need for rigorous TB control in both the Eastern and Western Capes.

According to Verkuijl, the South African national guidelines recommend a three-tiered approach to establishing collaborative TB–HIV activities. First is to establish coordinating bodies to organize surveillance, planning, monitoring, and evaluation. Second is to decrease the burden of TB in people living with HIV/AIDS through the internationally recommended “three I’s” strategy: intensified TB case finding, isoniazid preventive therapy, and infection control. Third is to decrease the burden of HIV in TB patients through HIV counseling and testing for TB patients, HIV prevention, cotrimoxazole preventive therapy, enrollment of coinfected patients into HIV care and support, and access to antiretroviral therapy for coinfected patients.

Verkuijl’s presentation focused on the first of the “three I’s”: intensified TB case finding through a method of TB symptom screening that was developed in the Eastern Cape. The identified need to shift from an acute to a continuing care model led to the development of a comprehensive HIV adult clinical record (ACR) through a partnership between the Department of Health and several nongovernmental organizations. The objective was to achieve uniformity in recording patient data and to help site staff track patients’ clinical progress and social support needs. A four-page version was piloted, and a revised eight-page record is now being implemented. Staff are being trained and mentored regularly to ensure the correct use of the

ACR. A six-question TB screening questionnaire is integrated into the ACR in such a way that clinicians can record information on TB symptoms at every visit. The six questions concern

• weight loss (unintended),

• cough for more than 2 weeks (including haemoptysis),

• night sweats for more than 2 weeks,

• fever for more than 2 weeks,

• swollen lymph nodes, and

• respiratory symptoms and/or chest pains.

These questions should be asked at each visit, and if the conditions are present, boxes in the ACR should be ticked accordingly. Patients presenting with any of the TB symptoms should be further investigated to confirm or rule out active pulmonary or extrapulmonary TB. Patients without symptoms are eligible for isoniazid preventive therapy. The ACR also includes room for findings from the physical examination and sputum results (two smears and one culture), as well as other results, including those of x-rays, needle aspirate biopsies, and drug susceptibility testing.

Currently, the ACR is being revised to accommodate the new guidelines for antiretroviral therapy. The record complements existing monitoring and evaluation systems; in addition, an electronic database, mirroring the ACR, is currently being piloted in East London. This database will provide a link with the District Health Information System, the national monitoring and evaluation system.

The ACR addresses all the steps that need to be taken during the TB screening process, from symptom screening to management of patients with TB. If a patient has positive symptoms upon screening, the ACR indicates that investigations need to be done. If TB is diagnosed, the ACR indicates that TB treatment and cotrimoxazole prophylactic treatment need to be started.

The ACR captures information from each clinical visit. For example, it includes entries for the clinician to mark if a patient is on TB treatment at the time of that visit. This feature helps prompt the clinician to monitor the clinical response to TB treatment and can assist in identifying patients not responding to first-line treatment. This integrated screening tool provides for quality and continuity of care through

• screening for active TB at every visit for patients enrolled in HIV care (who are not yet eligible for antiretroviral therapy) and for patients on antiretroviral therapy,

• early diagnosis and treatment of active TB,

• identification of TB-associated IRIS,

• monitoring of TB investigation results and TB treatment progress and outcomes,

• improved practical integration of HIV and TB programs, and

• identification of patients eligible for isoniazid preventive therapy.

In addition, the tool benefits prevention and early diagnosis of drug-resistant TB by monitoring the clinical progress of patients on first-line TB treatment, assisting with the identification of TB patients with suspected drug resistance, and monitoring breakthrough TB disease in patients on isoniazid preventive therapy. It allows for the monitoring and evaluation of the entire TB screening process using standard pre-antiretroviral therapy and antiretroviral therapy registers. The indicators that can be collected from the ACR include, for example:

• proportion of patients screened (among those enrolled in HIV or antiretroviral therapy care),

• proportion of patients who screened positive (among those screened),

• proportion of patients with investigations performed (among those screening positive),

• proportion of patients diagnosed with active TB (among those screening positive),

• proportion of patients starting on TB treatment (among those diagnosed with active TB), and

• proportion of patients starting on cotrimoxazole (among those diagnosed with active TB).

To illustrate the effectiveness of the tool, the proportion of patients in the Qaukeni Local Service Area (O.R. Tambo District) who were screened improved from less than 50 percent to more than 76 percent after use of the tool began. The proportion of new TB cases diagnosed among patients newly enrolling in HIV care gradually increased from 1.6 to 5.8 percent. Similarly, the proportion of patients with TB among the cohorts of patients starting HIV care, which includes patients already on TB treatment upon enrollment in HIV care, steadily increased from 4.5 to just over 20 percent.

According to Verkuijl, lessons learned include the following:

• A TB screening questionnaire integrated into a comprehensive HIV clinical record reminds clinicians to “think TB” at all times.

• The ACR facilitates effective monitoring of TB screening among HIV patients at enrollment and subsequent visits.

• The ACR facilitates proactive TB screening, leading to increased TB case finding.

• The TB screening questionnaire can be used to monitor the clinical progress of TB patients on first-line regimens and identify patients with suspected drug-resistant TB.

• Routine TB screening is critical in identifying breakthrough TB disease in patients on isoniazid preventive therapy.

• The ACR can be linked with the District Health Information System through an electronic database.

Verkuijl offered the following recommendations:

• TB and HIV should be managed by a comprehensively trained health care provider (“under one ceiling”).

• TB and HIV clinical records should be kept together for improved clinical management.

• The electronic ACR database should be finalized based on the results of the current pilots for adoption in the Eastern Cape to ensure that the link with the District Health Information System is available in all facilities.

The World Health Organization (WHO) Global Laboratory Initiative (GLI) has identified the need for an “urgent and massive scale-up of laboratory services” (WHO, 2010b). More specifically, it has stated that “the critical lack of TB laboratory capacity constitutes a global crisis, requiring a paradigm shift in providing laboratory policy guidance, quality assurance, and knowledge creation within a global and integrated laboratory network.”

Nordenberg identified some of the issues related to the urgent information needs of TB laboratories:

• Laboratory capacity is desperately insufficient.

• Laboratory capacity building efforts rarely take into account the development of data and information management capabilities.

• Laboratory services are microbe- or specimen-focused, in contrast to clinical and population health programs.

• The sensitivity and specificity profiles of emerging diagnostics may differ from those of existing diagnostics, rendering surveillance trend estimates difficult to interpret.

• There is a critical need to integrate laboratory information systems with those of clinical and population health programs.

• There is a critical need as well for information systems to track operational activities focused on drug-susceptible and drug- resistant TB, such as infection control programs and drug supply chains, as well as to support performance improvement programs.

• Information services and products must be thought of as a supply chain that moves information in a highly structured way. Building information supply chains is work-intensive, but their public health impact could be enormous.

These issues represent a large-scale problem that requires a large-scale solution. Unfortunately, according to Nordenberg, solutions that support capacity building for laboratory information management across many countries and thousands of laboratories have not been available.

The GLI addresses data and information needs in terms of TB diagnostics (WHO, 2010b). Information derived from diagnostics needs to be delivered to the right places at the right times. Nordenberg noted that, although WHO’s New Diagnostics Working Group of the Stop TB Partnership has developed an effective workflow model for the development of new diagnostic tests to support TB control programs globally, this model lacks reference to the necessary data and information supply chains that are critical to optimize the public health impact of emerging TB diagnostics (WHO, 2009b). This omission represents a major information chasm that requires urgent attention, according to Nordenberg.

Several years ago, the Association of Public Health Laboratories and the Public Health Informatics Institute identified 16 essential business processes for U.S. public health laboratories (Association of Public Health Laboratories and Public Health Informatics Institute, 2003). These unique functions highlight the different workflows that exist in the laboratory as opposed to the clinic or public health program office. Therefore, laboratory information systems that support these unique workflows will be most effective in supporting the laboratory’s public health mission. Given that the laboratory is the source for much of the information used to control TB and MDR and XDR TB, as well as other diseases, laboratory information management systems are central to any health care system. Regardless of what scientific or diagnostic work is being done in laboratories, the data and information they produce must be linked to clinical and public health programs.

Laboratories need to manage data and information more systematically, said Nordenberg. Public health programs require at least three critical supply chains to be successful: people, products, and information. The

information supply chain is complex. Information must be viewed as an intervention, not a technology, as it drives the right therapy to the right patient, prevents the emergence of drug-resistant strains, enables patients to be treated more cost-effectively, and improves operations. A disciplined approach to data and information provisioning will facilitate measurement of the quality and impact of the data and information. Such an approach will also enable performance improvement of the information supply chain to optimize public health impact.

Analyzing information gaps can be a means to achieve several important objectives:

• designing the model information supply chain for control of TB/drug-resistant TB;

• assessing gaps within one or more communities;

• quantifying the impact of the gaps in terms of treatment timeliness, treatment appropriateness, the drug supply chain, and so on;

• calculating the cost of the gaps in terms of the spread of disease, the emergence of resistance, morbidity and mortality, and other negative outcomes; and

• developing information supply chains for drug-sensitive and drug-resistant TB at the community, district, provincial, and national levels.

The building of laboratory information management capacity globally is being undertaken through participation in public–private partnerships. An example is the Laboratory Information for Public Health Excellence (LIPHE) program based at the CDC Foundation. This program is a growing collaboration that includes the Centers for Disease Control and Prevention (CDC), Eli Lilly, Fondation Merieux, the GLI, and WHO. LIPHE has created a set of processes and a framework designed to enable laboratories to work together to define best practices. The platform may vary, but standardization of the information is encouraged to facilitate sharing across communities, regions, and countries. Currently, however, no efforts are under way to develop an information supply chain model for TB.

According to Nordenberg, all the components needed to develop a robust model for information flow in TB control programs are available. Information products can be identified based on the questions that need to be answered to support TB control programs. Information supply chains can then be designed to deliver these information products. Once TB information supply chains have been designed and implemented, it will be possible to develop models that can quantify the public health impact of suboptimal information supply chains (e.g., data on missed diagnoses,

delayed treatments, propagation of drug-resistant TB, increased hospitalizations, increased cost of drugs, and increased mortality). These data should help make the development of effective TB information supply chains a priority along with the development of new therapeutics, new diagnostics, improved infection control practices, and other critical components of an effective TB control program.