The Changing Natural History of Tuberculosis and HIV Coinfection in an Urban Area of Hyperendemicity

Abstract

Tuberculosis (TB) has proven to be difficult to control in regions with a high prevalence of human immunodeficiency virus (HIV) infection. We previously described high prevalence of HIV infection among adults (23%) and rapidly escalating TB notification rates in a peri-urban township, Site-M in Cape Town, South Africa. The combination of delineated boundaries, a well-characterized population, centralized TB record keeping, and high levels of HIV testing make this population uniquely suited for TB epidemiologic and transmission studies. The driver of the HIV and TB coepidemic appears to be a high annual risk of Mycobacterium tuberculosis infection in this community. A high annual risk of M. tuberculosis infection may be the result of unrecognized infections coupled with intense social interaction and crowding. New non-facilitybased interventions will be required, with emphasis on community-based case finding and contact tracing to decrease the infective TB pool. There is a need for better understanding of the transmission dynamics of TB and the intensity of social interactions, which have exacerbated an HIV and TB epidemic in this community of hyperendemicity.

Tuberculosis (TB) remains a challenge to global public health, is a major cause of mortality, and has proven to be particularly difficult to control in regions with a high prevalence of human immunodeficiency virus (HIV) infection. An estimated 1.3 million deaths due to TB occur annually among HIV-uninfected individuals, and an additional 0.5 million deaths occur among HIV-infected persons. Of the estimated global burden of 9.3 million new TB cases in 2007, 1.37 million (14.8%) were associated with HIV infection and accounted for almost 25% of global AIDS-related mortality [1].

Sub-Saharan Africa has borne the brunt of the HIV and TB coepidemics, accounting for 79% of the global burden of HIV infection-associated TB cases in 2007. The 9 countries of the southern African region with hyperendemicity have TB case notification rates that are much higher than those for the rest of the African continent; these 9 countries have generalized HIV epidemics and report a prevalence of HIV infection of ⩾50% among persons with newly diagnosed TB (Figure 1). In 2007, the estimated rate of TB case notifications in Africa was 161 cases per 100,000 population; however, in this subregion of hyperendemicity, incidence rates of TB in South Africa and Swaziland increased to 948 cases per 100,000 population and 1198 cases per 100,000 population, respectively, with 73% and 80% of new TB cases, respectively, involving HIV coinfection.

The Millennium Development Goals for global TB control are to halt and start to reverse the increasing incidence of TB and to halve the 1990 prevalence and death rates by 2015 [2]. In countries where TB is hyperendemic, such as South Africa and Swaziland, achievement of Millenium Development Goals for TB is unlikely, because it would require a reversal of present TB incidence trends and a 6-fold reduction in TB incidence during the next 6 years.

Since the World Health Organization (WHO) declaration in 1993 that TB was a global emergency, the directly observed therapy short-course (DOTS) strategy has been the key public health intervention that has been widely used to affect global TB control [3]. The strategy focuses on TB case management of sputum smear-positive cases with use of short-course rifampicin- containing chemotherapy. Case finding is passive and facility based, with emphasis placed on case retention and the achievement of a high cure rate. Although DOTS has been effective in most regions of the world, contributing to the sustained downward trend in global TB prevalence, it has been comparatively ineffective in countries with a high prevalence of HIV infection [1, 4–7]. During 2002–2004, the WHO and the Stop-TB Partnership published guidelines [8], a strategic framework [9], and an interim policy for TB and HIV infection [10] to address the specific challenge of HIV infection-associated TB. These interventions aim to reduce the burden of TB in HIV-infected persons through use of TB prevention strategies, including isoniazid preventive therapy (IPT), intensified case finding, and infection control in conjunction with antiretroviral therapy (ART)—the so-called “3 I's.”

However, mathematical modeling suggests that a combination of very high levels of ART coverage and early ART initiation at high CD4 cell counts may be required to significantly affect population TB control, especially in settings where TB and HIV infection are hyperendemic [11]. Similarly, IPT is an intervention that reduces the risk of active TB in already HIVinfected individuals with latent TB infection rather than a primary strategy to control the public health burden of TB [12]. Although IPT is effective in decreasing the individual risk of progression to TB [13], the modeled population impact of IPT in areas of hyperendemicity is predicted to be small [14]. Therefore, there is an urgent need to understand the epidemiological factors driving the coepidemics in regions of hyperendemicity to inform TB-control strategies.

Lessons from Epidemiologic Studies in an Urban Community Where TB is Hyperendemic

Although the global burden of HIV infection-associated TB is concentrated in the southern African subregion, there are large differences in disease burden even in the subregion. Specific subpopulations, such as South African mine workers, have been well documented to have a high TB incidence, in part because of the multiplicative effect of HIV infection and mine work-associated pulmonary silicosis [15]. In addition, rapid growth of urban areas is occurring in the context of generally declining economic performance, and the growth of urban areas includes huge numbers of persons with low-income status [16, 17]. It is estimated that ∼61% of South Africans are urbanized, and 57% of these persons live in slum conditions [18] where TB and HIV burdens are greatest [19].

We previously described an epidemiologic study in South Africa that found a high prevalence of HIV infection among adults (23%) [5] and rapidly increasing TB notification rates [20]. Specifically, annual TB notifications have now reached 2000 cases per 100,000 population in this peri-urban township in Cape Town, designated by our study as “Site-M” [5, 20]. Regular household censuses have been performed that have shown that the community has undergone rapid population growth from 5000 residents in 1996 to 15,000 residents in 2008. This population growth has occurred within well-circumscribed boundaries (Figure 2). The community is socially deprived, living in overcrowded, largely informal dwellings located on demarcated plots serviced with water and sanitation. There is a single health care facility that provides primary medical care to community residents, and there is a primary and secondary school. Increases in TB notification rate have occurred despite a well-implemented national TB-control program based on the WHO DOTS strategy [21] at the single community clinic that manages all resident TB cases. Routine HIV testing (with consent) of patients with incident TB was introduced in 2002. The combination of delineated boundaries, a well-characterized population, centralized TB record keeping, and high levels of HIV testing make this population uniquely suited for studies on TB epidemiology and transmission.

Impact of HIV Infection on TB Control

Almost 2 decades ago, before the development of effective combination ART, Styblo [6] reported that existing TB-control strategies would be significantly undermined by HIV infection, particularly in Africa. It was postulated that the impact of HIV infection on the epidemiological situation of TB would depend primarily on the following parameters: (1) the prevalence of HIV infection in a community, (2) the prevalence of TB in the general population aged 15–49 years, (3) the progression from latent TB to active disease, (4) the level and trend in the annual risk of (new) TB, and (5) the detection rate of new and relapse cases of TB and cure rate among persons with smear-positive cases.

Other more recently recognized factors include the observation that combination ART can significantly decrease TB incidence [22] and the observation that TB incidence is very dependent on current CD4 cell counts [23]. Taking these factors into consideration, studies have focused on measuring the following likely drivers of increasing TB incidence in the study community: prevalence of HIV infection, prevalence of underlying latent TB, rates of progression from latent TB to active TB, annual risk of TB, and case detection rates.

Prevalence of HIV Infection

Since 1990, the South African Department of Health has performed annual national surveys on the prevalence of HIV infection among women attending antenatal services [24]. The TB notification rates in South Africa from 1990 through 2005 [1] and the national antenatal seroprevalence of HIV infection are shown in Figure 3 [24]. The corresponding adult TB notification rates and prevalence of HIV infection among adults at Site-M from 1996 through 2005 are also shown in Figure 3. During the these periods, the seroprevalence of HIV infection increased markedly, reaching levels of 30% and 23% among national antenatal attendees and adults at Site-M, respectively. TB notification rates have increased logarithmically for linear increases in prevalence of HIV infection; an even stronger positive relationship was shown in the high-burden township. Possible explanations for this nonlinear relationship could include changes in CD4 cell count distribution in the HIV-infected population during the rapid-growth phase of the HIV epidemic or increasing TB transmission between HIV-infected individuals as the HIV epidemic grows rapidly.

Age-Specific TB and HIV Infection

Over the past decade the number of TB notifications has increased markedly at Site-M, with the increased burden of TB disease predominantly affecting persons aged 15–45 years (Figure 4A). The numbers of TB presentations at any age are a function of the number of individuals in each age strata and the TB rate specific to that age group. There have been significant changes to age-specific TB rates over the past decade that have been associated with increasing prevalence of HIV infection (Figure 4B). TB rates appear to have increased in all age groups; however, the most marked increases are among persons aged 15–44 years, the age group most at risk of acquisition of HIV infection.

Population Prevalence of TB

Prevalence of underlying latent TB at any age is influenced by both the prevailing TB transmission rate and transmission rates during the preceding years of life; therefore, prevalence of TB increases with increasing age because of accumulated exposure. The traditional way to measure latent TB is to measure reaction to tuberculin antigens by tuberculin skin testing. Population tuberculin skin testing surveys have been infrequently performed in recent decades; however, a recent tuberculin skin testing survey at Site-M township primary school reported TB prevalences that increased from 8% at school entry to 53% by the age of 15 years [25]. Moreover, prevalence of latent TB appeared to continue to increase throughout adolescence. In 2006, the HIV-uninfected control population at a similar nearby township in Cape Town had a TB prevalence of 77% by the age of 28 years [26]. Other similar township populations in Cape Town have also shown equally high prevalence of adult TB infection [27].

Rate of Progression to Active TB Disease

The temporal association between infection and risk of progression to active disease has been well recognized [28]. Progression to active disease is particularly rapid in children and has been a marker of ongoing transmission; however, the resultant TB disease is frequently sputum smear negative [29]. Childhood TB is conventionally reported internationally as a <15 years smear positive rate [1]. In 2007, South Africa reported a high smear positive childhood rate of 30 cases per 100,000 population. However, the high burden of childhood disease is not adequately reflected by the <15 years smear positive rate. In 2007, although the <15 year smear positive rate for Site-M was 81 cases per 100,000 population, the TB notification rate was 54 cases per 100,000 population among children <15 years of age and reached 1390 cases per 100,000 population among children <5 years of age.

There has been a marked change in the adult age of TB disease presentation. During 1996–1997, a period of relatively low prevalence of HIV infection, the incidence of TB increased progressively with advancing age, with no case notifications for adolescents (age, 10–19 years); however, TB notification rates increasing steadily to 1700 cases per 100,000 population in the fifth decade of life [5]. During 2003–2004, when the prevalence of HIV infection among adults exceeded 20%, TB notifications predominantly were for adolescents and young adults.

The estimated incidence of TB among HIV-uninfected and HIV-infected adult community members in 2005 was 953 cases per 100,000 population and 5140 cases per 100,000 population, respectively [20], indicating a 5-fold increased risk among HIVinfected individuals. As a consequence of these high TB incidence rates, the lifetime cumulative risk of TB is extremely high for both HIV-uninfected and HIV-infected individuals in this community. The very high cumulative lifetime TB risk for HIVuninfected individuals is also much higher than the conventional estimated lifetime risk of latent infection progressing to TB disease of 10%–20% [30]. The increased lifetime risk may result from increased rates of progression because of poor nutrition, exogenous reinfection, or exposure to a high initial amount of infective TB [31]. The majority of individuals who are coinfected with TB and HIV live in sub-Saharan Africa, an area where hunger and malnutrition were already pressing concerns before the onset of the HIV and TB epidemics.

Level and Trend of Annual Risk of TB

Population density varies markedly among and within countries. Both the nature of the dwelling and crowding within the dwelling will have an impact on the number of individuals exposed to an infected person. Site-M has had an increasing population density, reaching 15,700 persons/km2 in 2008. The annual risk of TB among primary school children in 2008 was estimated to be 3.8%–4.8% [25], which is unprecedented in the current TB chemotherapeutic era. The annual risk of TB is similar to that found in several large scale surveys performed in western, eastern, and southern Africa from 1995 through 1960 [32]. In the prechemotherapy era, mean annual rates of infection as high as 13% per annum were reported among Parisian children in 1910 [33]. Lower rates of infection of 3% per annum were recorded among children in post-World War II Denmark [34].

A 77% prevalence of TB infection by age 28 years (during a period of increasing TB notifications) would indicate a high and ongoing mean risk of TB infection of 5.5% during an individual's preceding years of life. Childhood infection and TB disease in Site-M have been shown to be strongly associated with exposure to adult smear-positive TB in combined family groups that are resident on each serviced plot [35].

In summary, the annual risk of TB in this community is extremely high and appears to be maintained or to increase throughout childhood and adolescence. Trends in the annual risk of TB over time in any specific age group in this community are less certain; however, there is little evidence for decreasing transmission.

Case Detection and Treatment

Efficient case management of infective TB is the cornerstone of the DOTS strategy [3], to which other supplementary control strategies may be added [8–10]. The single TB facility in Site- M implements DOTS-based, short-course, rifampicin-containing chemotherapy, administered in accordance with national guidelines [36]. TB-associated mortality during 2002–2004, before availability of ART, among HIV-infected and HIV-uninfected persons with TB was 13% and 3%, respectively [20]. Treatment completion rates of persons surviving to 6 months of age were 84% among HIV-infected persons and 86.6% among HIV-uninfected persons [20]. In 2005, a cross-sectional survey of a randomly selected subset of the general population found that the existing facility-based smear-positive case finding was higher for HIV-uninfected community members than for HIV-infected community members (rates, 0.67 [95% confidence interval, 0.25–0.53] and 0.37 [95% confidence interval, 0.41–1.0], respectively) [20].

TB Transmission Patterns

Over a 5-year period from 2001 through 2005, all acid-fast bacilli-positive sputum samples obtained at the single clinic in Site-M were cultured, and IS6110-based restriction fragmentlength polymorphism analysis [37] was performed [38]. A broad diversity of ∼200 distinct circulating M. tuberculosis strains were estimated to be circulating in this community—a finding consistent with other studies in sub-Saharan Africa [39, 40]. This study also found an association between W-Beijing family strains and HIV infection that may reflect ongoing transmission of TB among HIV-infected persons. W-Beijing strains have been associated with increased virulence [41] and the development of multidrug resistance [42]. The high degree of genotypic diversity in certain strains may indicate that they either are endemic in this population or may be emerging and diversifying in the community.

Another important finding was the high rate of strain clustering. In this study, approximately half of the strains were clustered, and there were close temporal associations, especially among the paired clusters, supporting the likelihood that a significant proportion of disease in the community is attributable to recent infections. No association was found between HIV infection and clustering; therefore, new infections may be occurring in both HIV-uninfected and HIV-infected patients.

Discussion

This review has focused in detail on a specific, well-demarcated population that is heavily burdened with the dual epidemics of HIV infection and TB. Detailed analysis of the HIV and TB epidemics in this community may reveal insight into the factors driving the HIV and TB regional emergency in southern Africa. The incidence of TB has increased logarithmically with growth of the HIV epidemic and has been associated with a changed age distribution, resulting in the TB burden transferring from the elderly to young adults. The HIV epidemic appears to have unmasked a previously unrecognized high prevalence of TB and an unprecedented rate of TB in this crowded township.

Modeling studies suggest that a combination of interventions will be required to regain TB control [12]. The present strategy for global TB control remains the identification and effective case management of infectious TB cases [33]. However, although the facility-based program in this community appears to have achieved standard targets for TB case management for HIV-uninfected community members, the program has not had an impact on the extremely high annual risk of TB.

Strategies to decrease progression from prior infection may include ART and IPT. However, ART will need to be introduced with high coverage and earlier in the course of HIV infection, at higher CD4 cell counts, to significantly have an impact on rates of TB in the population [12]. Although the effect of ART is greater with continuing length of therapy [34], the benefits of IPT for HIV-infected patients are time-limited, compared with those for HIV-uninfected persons [14].

The underlying driver of the explosive HIV and TB coepidemic appears to be an extremely high annual risk of TB in this community. A high annual risk of TB may be the result of unrecognized infectious cases in the community and intense social interaction and crowding. New interventions in addition to the present clinic-based model of TB case management will be required, with increased emphasis on active communitybased case finding and contact tracing to decrease the infective TB pool. There is also an urgent need for better understanding of the transmission dynamics of TB and intensity of social interactions that have exacerbated the HIV and TB coepidemic in this community of hyperendemicity.

Acknowledgments

Potential conflicts of interest. L.-G.B. and R.W.: no conflicts.

Supplement sponsorship. This article is part of a supplement entitled “Synergistic Pandemics: Confronting the Global HIV and Tuberculosis Epidemics,” which was sponsored by the Center for Global Health Policy, a project of the Infectious Diseases Society of America and the HIVMedicine Association, through a grant from the Bill & Melinda Gates Foundation.

Figures and Tables

Figure 1.

Estimated prevalence of HIV infection among persons with newly diagnosed cases of tuberculosis (TB), 2007. Reprinted with permission from the World Health Organization [1].

Figure 2.

An aerial photograph of Site-M (in Cape Town, South Africa), with superimposed property boundaries outlined. The figure was created using ArcGIS, version 9.2 (ESRI, 380 New York St, Redlands, CA 92373-8100).

Figure 3.

The relationship between tuberculosis (TB) notification rates and seroprevalence of HIV infection in the South African population (diamonds) and the population at Site-M (triangles), with exponential regression lines for South African data during 1990–2005 (R2, 0.8461) and for Site-M data during 1996–2005 (R2, 0.9376).

Figure 4.

A, Number of tuberculosis (TB) notifications, stratified by age, at Site-M over two 4-year periods: 1996–1999 (diamonds) and 2004–2007 (triangles). B, TB notification rates, stratified by age, at Site-M over two 4-year periods: 1996–1999 (diamonds) and 2004–2007 (triangles).