TRANSMISSION OF DRUG-RESISTANT TB IN CHINA¹

Among the approximately 1.3 million new cases of TB in China each year, 35.2 percent are drug-resistant, 5.7 percent are MDR TB, and about 0.5 percent are XDR TB, according to a 2007–2008 nationwide survey. The most important question to ask about this drug resistance, said Gao, is where it came from. Was it primary, in that a drug-resistant strain was transmitted from one person to another? Or was it acquired, in that a strain developed drug resistance during treatment because of noncompliance with a treatment regimen, poor-quality drugs, or treatment with a single drug? The significance of the distinction is that primary drug resistance indicates a need for better TB control to interrupt transmission, whereas acquired drug resistance indicates a need for better patient management to prevent the evolution of resistance.

A simple way to make this distinction is by referring to the history of care. If a patient has not been treated before, drug resistance is often assumed to be primary and the result of transmission. If a patient has been treated previously, drug resistance is often assumed to be acquired. WHO data indicate that 6–10 percent of new TB cases and 13–60 percent of cases in previously treated patients are MDR. One possible conclusion that could be drawn from these data, said Gao, is that previously treated patients should be the focus of improved patient care. In most parts of the world, however, new cases make up more than half of MDR TB incidence. In Shanghai, for example, from 2002 to 2006, 59 percent of the MDR TB cases were among new TB patients, indicating the transmission of drug resistance. The same observation can be made for XDR TB patients, although the total numbers are much smaller. In Shanghai from 2004 to 2007, 6 of 11 XDR TB patients represented new cases.

Another way to track transmission is by genotyping different strains. If two people have strains with the same genotype, transmission can be assumed. A search of the biomedical literature reveals several cases in which drug-resistant strains were responsible for clusters of MDR and XDR TB. In a 2009 study conducted in Samara, Russia, for example, 63 of 189 patients with XDR TB fell into two large clusters of similar genotypes. Gao observed that this means one of the patients could transmit, on average, to 31 or 32 patients.

In an ongoing study in China's Shandong, Shanghai, and Sichuan provinces, Gao and his colleagues found different clustering rates (i.e., clusters of TB with similar genotypes) in different geographic areas. According to preliminary results, 39 percent of TB patients in Shandong Province fell into clusters of similar genotypes, indicating primary transmission from one person to another. In the other two provinces, Shanghai and Sichuan, the rates at which clusters of patients with TB had similar genotypes were 10.8 and 13.2 percent.

Another study in which Gao was involved found that some previously treated patients were being reinfected by drug-resistant strains rather than developing drug resistance in ongoing infections. The researchers genotyped TB strains of 32 TB patients sampled before and after treatment. If the bacterial genotype changed, the drug resistance likely arose from reinfection; if patients had a mix of genotypes in an isolate, they had multiple infections. The results indicated that a significant fraction of the treated patients had been reinfected with a more drug-resistant strain. Thus, even some of the patients in whom the evolution of drug resistance was assumed may have developed primary resistance through reinfection.

The bottom line, said Gao, is that more than 50 percent of MDR and XDR TB patients represent new cases. The evidence is strong that MDR and XDR TB are being transmitted among individuals, and even some previously treated patients show primary drug resistance. The conclusion to be drawn, said Gao, is that new strategies to block the transmission of drug resistance are urgently needed.

TRANSMISSION-BASED GENETIC ANALYSIS IN SOUTH AFRICA²

According to WHO, South Africa, a country of around 48 million people, saw 461,000 new cases of TB in 2007—about 950 cases per 100,000 population (WHO, 2009). About three-quarters of all TB cases occurred in people who were HIV-positive, and about 112,000 people died of TB in 2007. Among those coinfected with HIV and TB, 90 percent die within months if not treated. South Africa has more HIV-positive TB cases than any other country in the world, said Gey van Pittius.

WHO has estimated that MDR TB accounts for 1.8 percent of all new TB cases, although estimates presented at the workshop were closer to 5 percent. Gey van Pittius reported that 68 percent of the budget for the South Africa National Tuberculosis Program is devoted to MDR TB, as opposed to only about 4 percent for first-line treatment.

The first documented case of XDR TB in South Africa was detected in 1997. In 2008 a study of about 700 MDR TB cases showed that 14 percent were pre-XDR—that is, MDR with one of the markers for XDR—and 6 percent were already XDR (Mlambo et al., 2008). It took 13 years from the introduction of rifampicin for TB treatment in 1972 for the first MDR TB case to arise in 1985 in South Africa, said Gey van Pittius, and another 12 years for the first XDR TB case to appear in 1997. Now TDR TB is starting to appear after another 12 to 13 years.

Cape Town, where the Centre of Excellence for Biomedical Tuberculosis Research is located, has one of the highest incidences of TB in the world. The Centre has been working in many areas of TB research, including pharmacology, genetics, bacteriology, immunology, and molecular epidemiology. At the Centre, the mycobacterial molecular epidemiology and genomics group has been looking at the evolution of mycobacteria, pathogenicity, genomic variation, non-tuberculous mycobacteria, mycobacterial diagnostics, and drug resistance. Over the years, the group has acquired a large mycobacterial sample bank from study sites throughout Southern Africa.

In the Cape Town suburbs of Ravensmead and Uitsig, Gey van Pittius and his colleagues have collected more than 8,500 isolates and have used restriction fragment length polymorphisms to identify more than 875 different strains of M.tb. These strains can be grouped into 40 different strain families, which in turn fit into the strain families found throughout the world. However, two families make up the majority of their strains: the F11 strains within the LAM group and the typical Beijing strains within the Beijing group.

The strain populations within this community have changed over time. Many have declined in numbers, while some have increased. Most alarming, the Beijing strains have increased with a doubling time of 3.9 years (van der Spuy et al., 2009). This doubling phenomenon is not due to drug resistance, as only drug-sensitive strains were included in this study. In particular, a recently evolved sublineage of the Beijing strain family is associated with the increased ability to spread and cause disease within this community (Hanekom et al., 2007).

Different MDR TB sublineages are associated with specific host populations in South Africa. In the Eastern Cape Province, for example, the majority sublineage is atypical Beijing, whereas in KwaZulu-Natal, the predominant MDR TB strain is F15. In Johannesburg, the strains are more mixed. In the Western Cape Province, four major strains are responsible for much of the MDR TB—the typical Beijing strain R220, the LAM strain F11, the S-family strain F28, and the low-copy clade strain DRF150 (Streicher et al., 2004). For all of the drug-resistant TB in the Western Cape, the doubling time is 8.19 years, whereas for MDR TB, the doubling time is about 4 years. More alarming, a single Beijing strain, the R220 sublineage, is responsible for 42 percent of the total drug resistance epidemic and has a doubling time of only 2.38 years. Furthermore, 90 percent of patients infected with MDR TB strains are smear positive, which indicates that they have the potential to spread these resistant strains rapidly.

Even where the TB control program is strong and patients are adherent to therapy, XDR TB has a tendency to emerge. In a study of a mining community in northwestern South Africa, for example, with a more than 85 percent cure rate and semiactive case finding, TB patients were put on first-line treatment before it had been determined whether they had MDR TB (Calver et al., 2010). Pyrazinamide resistance was acquired very rapidly. Since these patients in effect received only ethambutol monotherapy (because of the resistance of the bacterium to the other three constituents of the first-line therapy—isoniazid, rifampicin, and pyrazinamide), ethambutol resistance also evolved rapidly. Once the patients had been identified as having MDR TB, they were given second-line treatment. Only ofloxacin, kanamycin, and ethionamide were active (resistance to pyrazinamide and ethambutol already having developed). Then second-line resistance emerged, followed by XDR TB. This progression was especially alarming, said Gey van Pittius, because there was effective control over the patients, and adherence to therapy was very good. The overall transmission rate in this setting was 71 percent.

MDR TB is evolving into XDR TB in common strains in a number of places in South Africa. MDR TB in these settings is largely transmitted, whereas there is currently less transmission of XDR TB and more development of XDR TB from the large pool of MDR TB cases.

Specific mutations can provide insight into treatments for MDR and XDR TB. For example, a mutation in the inhA gene or promoter region generally confers low-level isoniazid resistance, creating the possibility of treating patients with high doses of isoniazid. This mutation is widespread in the Western Cape Province. However, this mutation also provides high-level ethionamide resistance, so that ethionamide treatment should be discontinued in these individuals. Similarly, a mutation in the KatG gene indicates treatment not with isoniazid but with ethionamide if the strain is found to be susceptible to ethionamide by drug susceptibility testing, as mutations in KatG do not confer resistance to ethionamide.

The conventional wisdom is that resistance to ethambutol is rare, so second-line treatment includes ethambutol, pyrazinamide, ofloxacin, amikacin, and ethionamide. However, a 2006 study showed that 20 percent of drug-resistant MDR TB isolates were already resistant to ethambutol (Johnson et al., 2006). Likewise, Gey van Pittius cited unpublished results from a 2008 review of MDR TB isolates indicating that approximately 50 percent had mutations in the embB gene conferring resistance to ethambutol.

Although the extent of pyrazinamide resistance is currently unknown, a 2006 study found that 53.5 percent of drug-resistant isolates were already pyrazinamide-resistant (Louw et al., 2006). An association also was found between pyrazinamide resistance and MDR TB. A later study found that 52 percent of MDR TB isolates showed resistance to the drug (Mphahlele et al., 2008). For these reasons, Gey van Pittius and colleagues recommended that ethambutol and pyrazinamide not be regarded as effective second-line drugs and that drug susceptibility testing be implemented for all TB cases to ensure appropriate treatment and thereby limit amplification of resistance (Hoek et al., 2009).

Gey van Pittius noted that during the time it took to deliver his talk, four people in South Africa had died of TB—one person every four and a half minutes on average. “We are losing the battle against resistant TB,” he suggested. And although transmission of XDR TB is relatively uncommon at present, Gey van Pittius believes that, given the rapid spread of MDR TB, XDR TB is likely to take hold among susceptible populations. Molecular methods will be needed to determine what types of treatment will be effective.

In response to a question about multiple infections in a single patient, Gey van Pittius said infection pressures are so high in some communities that patients are frequently infected with multiple TB strains. Sometimes the strains compete, but sometimes infection with a new strain makes a previous strain more active. A workshop participant pointed out that different strains can be found in different parts of the body—for example, in sputum and in TB lesions. In addition, Gey van Pittius noted the association between particular strains and population groups. The relationship is complicated by the varying virulence of different strains. According to Gey van Pittius, interplay between the host genetics and the bacteria is involved. Further, M.tb. is not known to recombine, a fact that Gey van Pittius termed “one of the lucky things in the mycobacterial research field.”

A workshop participant pointed out that children often acquire TB strains different from those of their presumed adult contacts. As a result, a child may have drug-susceptible TB when a parent has drug-resistant TB, meaning that the child can be treated with less toxic and more effective first-line drugs rather than more toxic and expensive second-line drugs.

REDUCING NOSOCOMIAL DRUG-RESISTANT TB TRANSMISSION³

Every case of MDR and XDR TB that is prevented represents one less patient who will require 18–24 months of difficult and expensive treatment, said Nardell. Effective prevention requires consideration of the priorities in infection control. For example, if reinfection is driving the epidemic, how effective can isoniazid be for prevention? If reinfection is as common as appears to be the case, what is the potential for a new vaccine? Is there a vaccine that is better than natural infection? And how can community-based care get people out of hospitals where they are transmitting drug-resistant strains among each other?

Even in places such as Peru that have good DOTS programs and effective nationwide MDR TB treatment, the number of MDR TB cases continued to rise until recently. There are probably several reasons for this, said Nardell. Standardized treatment may have contributed to the increase in drug resistance. Echoing Gao, however, Nardell noted that another major factor is that many previously treated individuals are being reinfected, even though their cases are being misclassified as acquired (i.e., secondary to poor case management). Thus while the global TB community has emphasized the need for excellent treatment adherence, transmission is in fact very important to stemming the tide of drug-resistant TB.

In places such as Tomsk, data showing a downward trend in MDR TB are encouraging, said Nardell. Yet he also cited a study based on the hypothesis that an association between substance abuse and nonadherence increased MDR TB in Tomsk; however, the major driving factor behind MDR TB appeared instead to be hospitalization (Gelmanova et al., 2007). Indeed, adherent patients hospitalized in the course of treatment were six times more likely to develop MDR TB. Nardell said this finding should be a strong message that the hospital is not a place to treat MDR TB in the long run. (See Box 4-1 for a case example of control of nosocomial transmission of TB in the Russian Federation.)

BOX 4-1

Control of Nosocomial TB Infection in a TB Dispensary in Vladimir Oblast, Russian Federation. Elina Sevastyanova, Central TB Research Institute (CTRI) of the Russian Academy of Medical Sciences, presented on control of nosocomial TB infection in the TB (more...)

Nardell noted that the most well-known outbreak of XDR TB took place in KwaZulu-Natal Province in South Africa, where many patients at the Church of Scotland Hospital and other hospitals in the region were infected. Mortality was very high and very rapid. The majority of these patients had undergone no previous TB treatment, all were infected with HIV, and most had the same KwaZulu-Natal strain.

Health care workers also are at risk of infection. A 9-year study in Samara Oblast showed a TB risk of 742 per 100,000, 10 times that of the general population (Dimitrova et al., 2005). In inpatient TB facilities, the incidence ratio was almost 18 times that for general health care workers. This additional risk to health care workers has serious implications in places like Africa where it is very difficult to get people to work in TB hospitals. Transmission among health care workers also is a way of monitoring the spread of TB. “It is clear from this kind of data that hospitals provide a unique focus for transmission,” suggested Nardell.

As early as 1986, Nardell published a study of TB reinfection among non-HIV-infected people in a homeless shelter (Nardell et al., 1986). Although such studies are difficult, they have important implications. If reinfection is occurring in institutions, people need to be treated in their homes and communities. Such treatment is occurring in a growing number of places around the world, reducing opportunities for institutional transmission.

Prompt and effective treatment stops transmission, Nardell pointed out. In a series of experiments conducted since the 1950s, researchers have exposed guinea pigs, which are highly susceptible to TB, to the air from patients in TB wards. In all of these experiments, just a few patients accounted for most of the infections in the guinea pigs, while the other patients were much less infectious. In almost all cases, the infectious patients were the ones with unrecognized or inadequately treated drug-resistant TB. For example, in a recent set of experiments in South Africa, 360 guinea pigs were exposed to 26 patients who were strongly smear positive, had cavitary TB, and had recently started on therapy. Among the infected guinea pigs from which spoligotypes could be obtained, all were infected by three patients later found to have had XDR TB and not to be on effective treatment. “Treatment has a very profound and very rapid effect,” said Nardell. “We act in hospitals now as if all smear-positive patients are infectious. In fact, if they are on effective treatment, they are not.”

Conventional thinking about TB in low-prevalence areas is that people are infected and have dormant TB that is later reactivated. But in high-prevalence parts of the world, people may have a certain amount of immunity to TB by virtue of prior exposure, or some of the strains may be somewhat attenuated because of drug resistance. In such cases, reinfection may be driving the epidemic more than the original infection. According to Nardell, “You get a critical infection in some part of the lung or with some strain that allows infection to progress. That's a very different pathogenesis, where reinfection is absolutely part and parcel and not an unusual phenomenon at all.”

If patients with unsuspected or inadequately treated drug resistance cause most transmission, general medical practices need to be rethought, said Nardell. Many hospitals throughout the world have open wards, which can allow unsuspected and untreated TB to spread widely. In one study done in Peru, 250 patients admitted to a female ward over the course of a year were screened for TB regardless of why they had entered the hospital (Willingham et al., 2001). Among that group, 40 patients had positive cultures; 26 of those 40 patients were smear positive; 13 of the 40 were unsuspected cases of TB; and 8 of the 40 had MDR TB, including 6 whose MDR TB was unsuspected.

Even in TB hospitals, undiagnosed drug-resistant TB is transmitted during the months it takes to receive the results of drug susceptibility testing. And in MDR TB wards, the patients who will be spreading TB are the undiagnosed, unsuspected, untreated XDR TB patients.

These observations have major implications for rapid diagnostics, said Nardell. “If we can have a tool that will quickly diagnose, we can triage, we can treat, and we can probably do a much better job than our traditional approaches with ventilation and isolation rooms,” he said. Simple approaches can work. In the triage scheme that Farmer and his colleagues in Haiti have used for more than a decade, most patients are treated in the community and not in the hospital. Those who are in the hospital and are diagnosed are placed on treatment, after which they quickly become noninfectious. If they are smear negative, they can be put in the general medical ward, even if it contains HIV-positive patients, because they are smear negative and on treatment. If they are smear positive and HIV-negative, they go to a special TB pavilion with better ventilation, UV air disinfection, and so on. If they are both smear positive and HIV-positive, they go to one of six isolation rooms.

Nardell acknowledged that this is not the optimal triage scheme for every setting. In the case of MDR TB, for example, the situation is more complex. However, having a rapid diagnostic—beyond a sputum smear and an HIV-positive test—would make it possible to develop a rational scheme for separation and treatment that could limit transmission in hospitals much more effectively than negative pressure ventilation and isolation rooms. Tests such as Gene Xpert TB and the line probe assay are rapidly becoming available in resource-limited settings. The frequency of nosocomial transmission also calls for changes in the design of hospitals and clinics, said Nardell. In climates where doors and corridors can open to the outside, constructing hospitals with many small rooms connected to each other increases the possibility of transmission. Simple negative pressure ventilation systems and the use of respirators can be implemented in all countries at reasonable cost. Germicidal UV light presents more complex issues. It is important in cold climates where ventilation is expensive, but must be used properly. Better UV fixture designs are needed, as are locally produced UV fixtures. Although the practice is common in Eastern Europe, Nardell observed that there is no need to irradiate a room once a patient leaves because people cannot be infected from surfaces. Once organisms land on a surface, it is almost impossible to resuspend them in a particle size that can be inhaled into the alveoli. Also, devices that filter air should be avoided because they generally are not moving enough air to be highly effective and provide a false sense of security.

Nardell cited a summer course at the Harvard School of Public Health for engineers, architects, administrators, and physicians on the design of safe buildings for infection control (see https://ccpe.sph.harvard.edu/request.cfm). He also highlighted the website Global Health Delivery Online (http://www.ghdonline.org), a free online resource on MDR TB and TB transmission control. This website contains discussions on topics monitored by international experts and free guidelines and documents.

During the discussion period, Nardell and Cassell discussed efforts by the U.S. Department of Defense to develop air samplers for infectious agents. Nardell noted that the main difficulty with such samplers is distinguishing living from dead organisms in the air. Infectious agents typically are highly diluted in the air, which is why guinea pigs must be exposed to TB patients for weeks or months to become infected.

Footnotes

1

This section is based on the presentation of Qian Gao, Fudan University.

2

This section is based on the presentation of Nico C. Gey van Pittius, Stellenbosch University.

3

This section is based on the presentation of Edward A. Nardell, Harvard Medical School and Brigham and Women's Hospital.