Published ahead of print on December 30, 2003, doi:10.1164/rccm.200305-714OC
© 2004 American Thoracic Society Patients with Active Tuberculosis often Have Different Strains in the Same Sputum SpecimenMedical Research Council Centre for Molecular and Cellular Biology, Department of Medical Biochemistry, and Centre for Tuberculosis Research and Education, Department of Child Health and Paediatrics, Faculty of Health Sciences, Stellenbosch University, Tygerberg, South Africa Correspondence and requests for reprints should be addressed to Robin M. Warren, M.D., MRC Centre for Molecular and Cellular Biology, Department of Medical Biochemistry, Stellenbosch University, P.O. Box 19063, Tygerberg, South Africa 7505. E-mail: rw1{at}sun.ac.za
It is generally accepted that tuberculosis results from a single infection with a single Mycobacterium tuberculosis strain. Such infections are thought to confer protective immunity against exogenous reinfection. In this study, a novel polymerase chain reaction method was developed to specifically identify M. tuberculosis strains belonging to the Beijing and non-Beijing evolutionary lineages in sputum specimens collected from tuberculosis patients resident in an epidemiologic field site in Cape Town, South Africa. The sensitivity and specificity of the polymerase chain reaction–based strain classification method were 100% (95% confidence interval, 85–100%) when compared with DNA fingerprinting and spacer oligotyping (spoligotyping). Application of this method showed that 19% of all patients were simultaneously infected with Beijing and non-Beijing strains, and 57% of patients infected with a Beijing strain were also infected with a non-Beijing strain. Multiple infections were more frequent in retreatment cases (23%) as compared with new cases (17%), but were not associated with sex, age, or smear grading. These results suggest that multiple infections are frequent, implying high reinfection rates and the absence of efficient protective immunity conferred by the initial infection. This finding could influence our understanding of the epidemiology of disease in high-incidence regions and our understanding for vaccine development.
Key Words: multiple infection • Mycobacterium tuberculosis • reinfection Active tuberculosis is thought to develop as a continuation of the primary infection (primary tuberculosis), or after endogenous reactivation of the primary infection or exogenous reinfection with a second Mycobacterium tuberculosis strain (1). Understanding the relative contribution of each of these mechanisms will have important implications for prevention of the development of new cases (2), evaluation of new drugs (3), interpretation of molecular epidemiologic data (4), as well as for the design and evaluation of protective and therapeutic vaccines (3, 5). The relative quantification of exogenous reinfection and endogenous reactivation depends on our ability to accurately differentiate between strains of M. tuberculosis. The development of an internationally standardized DNA fingerprinting method (6) has enabled the genotypic classification of M. tuberculosis strains with a high level of sensitivity and specificity. Using this methodology, molecular epidemiologic studies have shown the presence of a single strain in most cultures collected from patients with tuberculosis (7, 8), thereby suggesting that disease is caused by a single strain (infection). In contrast, we and others have shown that both human immunodeficiency virus (HIV)–negative and HIV-positive individuals can be infected with more than one strain (1) during the same episode (multiple infection) (4, 8–10), (2) in different lesions (multiple infection) (11, 12), or (3) during successive episodes (reinfection) (3, 9, 13–15). However, multiple infections are rarely found when using the DNA fingerprinting method (7, 8) and therefore their significance remains unknown. This study aimed to determine the extent of multiple M. tuberculosis infection in sputum specimens collected from new and retreatment tuberculosis cases, resident in a community with a high incidence of disease (16), using a polymerase chain reaction (PCR) method based on comparative genomic data. The results are discussed in the context of the bacterial population structure within the human host and how multiple infection could influence the interpretation of molecular epidemiologic data. These findings could have important implications for the future design of antituberculosis vaccines, as well as vaccine and drug trials. This work has been presented at the Keystone Symposium, Taos (2003) (30) and in Prague (2003) (31).
Patients and Specimens During the period March 2000–June 2002, pretreatment sputum specimens were collected only from adult patients (  15 years of age) diagnosed with smear-positive tuberculosis (17). All patients were resident in an epidemiologic field site in Cape Town, South Africa (16). Patient demographics including sex, age, and previous history of tuberculosis were recorded at diagnosis and stored in a database. HIV testing was not routinely done, although a recent survey of 366 new adult smear-positive tuberculosis patients at the field site showed that 10% were HIV coinfected. This study was approved by the Ethics Committee of Stellenbosch University (Tygerberg, South Africa).
Classification of M. tuberculosis Isolates To identify a strain belonging to the Beijing evolutionary lineage, the DNA was subjected to PCR amplification using HotStar Taq polymerase (Qiagen, Hilden, Germany) and overlapping primer sets complementary to the 3' end of the IS6110 element and Rv2820 (primer set 1, TTCAACCATCGCCGCCTCTAC and CACCCTCTACTCTGCGCTTTG; primer set 2, ACCGAGCTGATCAAACCCG and ATGGCACGGCCGACCTGAATGAACC) (see online supplement for PCR amplification conditions and fractionation of products). A positive amplification product of 393 base pairs (bp) and 239 bp, respectively, indicated the presence of an IS6110 insertion in Rv2820 that is unique to the Beijing evolutionary lineage (18) (Figure 1 and Figures 2A and 2B) .
To determine the presence of M. tuberculosis strains other than those belonging to the Beijing evolutionary lineage, the DNA was PCR amplified with overlapping primer sets complementary to Rv2819 (primer set 3, GATCGCTTGTTCTCAGTGCAG and CGAAGGAGTACCACGTGGAG; primer set 4, GGTGCGAGATTGAGGTTCCC and TCTACCTGCAGTCGCTTGTGC). A positive amplification product of 569 and 308 bp, respectively, indicated the presence of M. tuberculosis strain(s) belonging to non-Beijing evolutionary lineages, as the region spanning the genes Rv2816 to Rv2819 (including part of Rv2820) is deleted in all Beijing strains (18, 19) (Figure 1 and Figures 2C and 2D). The sensitivity and specificity of amplification for the different primer sets was determined by PCR amplification of DNA templates from genetically distinct M. tuberculosis strains characterized by IS6110 DNA fingerprinting (6) and spacer oligotyping (spoligotyping) (20). In addition, DNA templates from different mycobacterial species were subjected to PCR amplification to further determine the specificity of the primer sets. Smear-positive sputum specimens that showed positive amplification products only with primer sets 1 and 2 were classified as single Beijing infections (Figure 2, lane 2). Similarly, specimens that showed amplification products only with primer sets 3 and 4 were classified as single non-Beijing infections (Figure 2, lane 3). Specimens that showed positive amplification products with primer sets 1, 2, 3, and 4, were classified as multiple infections (Figure 2, lane 1). Specimens that showed discordant amplification were assigned as cross-contamination and excluded from the study. To assess the extent of possible laboratory cross-contamination, control sputum specimens were concurrently collected from individuals who were smear and culture negative and who were resident in the same community. These specimens were incubated in BACTEC medium for 7 days and crude DNA preparations were obtained after boiling (see online supplement). In addition, aliquots of BACTEC medium were coprocessed as a second group of negative controls (see online supplement).
Spoligotyping
Statistical Methods
Description of Patients During the period March 2000 to June 2002, 407 adults ( 15 years of age) resident within the epidemiologic field site (16) were diagnosed with smear-positive tuberculosis. The average age of these patients was 36.3 years and included 249 (61.2%) male patients and 158 (38.8%) female patients. According to the National Tuberculosis Control Program in line with the Directly Observed Therapy Short-course strategy, diagnosis of tuberculosis is made by sputum smear microscopy in new cases, and by smear microscopy and culture in retreatment cases. We endeavored to recover these sputum specimens from the National Health Laboratory after routine processing; however, some sputum specimens were not identified and were not available for research purposes. Pretreatment sputum specimens were available for analysis from 200 of the 407 patients. Their average age was 35.7 years and included 132 (66%) males and 68 (34%) females. This group of patients did not differ according to the demographic measures of age, sex, and previous history of tuberculosis when compared with patients for whom a sputum specimen was not available.
Validation of the PCR Method
Detection of Multiple Infections Using primer sets 1 and 2, specimens from 61 of the 186 (32.8%) patients showed positive amplification products, demonstrating infections with strain(s) belonging to the Beijing evolutionary lineage (Figures 2A and 2B). PCR amplification using primer sets 3 and 4 demonstrated positive amplification products in 160 (86%) of the patient specimens (Figures 2C and 2D). Comparison between the results obtained with primer sets 1 and 2 and primer sets 3 and 4, showed that 35 (19%) of the specimens were positive for all the primer sets, demonstrating multiple M. tuberculosis infection (Table 1) . This represents multiple infection in 57% of patients infected with strains belonging to the Beijing lineage. The remaining 26 patient specimens (14%) were positive for amplification with primer sets 1 and 2 only, demonstrating single infection with strains belonging to the Beijing lineage.
Of the 35 specimens assigned as multiple infection by PCR amplification, 9 (26%) showed the presence of both Beijing and non-Beijing spoligotype patterns, confirming multiple infection. Eleven (31%) of these specimens showed only a Beijing spoligotype and 15 (43%) showed only a non-Bejing spoligotype. Of the 26 specimens that showed a single Beijing infection by PCR amplification, all were confirmed by spoligotyping and no underlying non-Beijing spoligotypes were observed. All of the 125 specimens assigned as non-Beijing infections by PCR amplification showed non-Beijing spoligotypes and no underlying Beijing spoligotype could be detected. The demographics of the 35 patients coinfected with strains belonging to the Beijing and non-Beijing lineages were similar to that of the 151 patients assigned as having single infections (Table 1). Multiple infections were more frequently observed in retreatment cases (23%) as compared with new cases (17%); however, this trend was not significant (Fisher exact odds ratio, 1.4; 95% confidence interval, 0.7–3.1). No significant association could be demonstrated between multiple infection and sex, age, or smear grading (Table 1).
Detection of Laboratory Cross-contamination
Over the past decade, molecular genotyping methods have highlighted extensive genotypic heterogeneity among clinical isolates of M. tuberculosis, making accurate classification of the different disease-causing strains a complex science (6). Many of these chromosomal polymorphisms have been analyzed, including strain-specific IS6110 insertion polymorphisms (23, 24) and chromosomal deletions (25–27), to identify possible correlates between genotype and phenotype. In this study we have used these genetic data to describe a method to classify clinical isolates according to the presence or absence of chromosomal markers unique to defined evolutionary lineages. Using this method, it was possible to investigate the epidemiologic phenomenon of multiple infections in sputum cultures originating from new and retreatment tuberculosis cases resident in an epidemiologic field site in Cape Town, South Africa (16). This study demonstrates that 19% of the patients included in the study were simultaneously infected with strain(s) belonging to the Beijing and non-Beijing evolutionary lineages. However, among patients infected with a Beijing strain, 57% were also infected with a strain(s) belonging to the non-Beijing evolutionary lineage. This PCR-based estimate of multiple infection was substantially higher than the 4.8% detected by spoligotyping (this study) and the 2.3% reported using DNA fingerprinting (8). Extensive evaluation of a series of control samples showed that laboratory cross-contamination is an unlikely explanation for the high frequency of multiple infections, although we acknowledge that the observed level of cross-contamination may lead to an overestimate of multiple infection. However, our estimate is also limited by the inability of the method to identify multiple infections with different Beijing strains or different non-Beijing strains. Therefore we conclude that this study represents a conservative estimate of the frequency of multiple infections in the study setting. We acknowledge that the frequency of multiple infection may be vastly different in different settings, as well as in patients with either extrapulmonary or smear-negative tuberculosis. The high number of multiple infection cases seen in this setting is unlikely to be due to HIV-induced immune deficiency given the low prevalence of HIV and tuberculosis coinfection in the study setting. However, in the absence of HIV testing of all patients, we cannot exclude HIV as a mechanism enabling multiple infection. In addition, we cannot exclude other factors that may alter the immunity of the patients, such as alcoholism and malnutrition. The occurrence of multiple infections in 17% of the new tuberculosis cases implies that reinfection in the study setting is extremely high. These multiple infections may occur when both infecting strains present to a "naive" immune response and thereby escape killing (11). An alternative possibility is that of "superinfection." In such cases, we speculate that an ongoing tuberculosis infection may significantly divert the immune response, thereby increasing the overall susceptibility to reinfection. Alternatively, reinfection occurring some time after the initial infection may initiate disease progression and the endogenous reactivation of the primary infection (12). The latter will imply that the primary infection is unable to confer protection against a secondary infection. A similar conclusion was drawn from mouse model experiments after repeated infection with different M. tuberculosis strains (28). The higher proportion of multiple infections in retreatment cases supports previous observations of the importance of reinfection in recurrent tuberculosis (3, 14). However, interpretation of restriction fragment-length polymorphism data in the previous studies depended on the assumption that a patient is infected only with a single strain during each episode of disease. The identification of multiple infections in both new and retreatment cases implies that reinfection studies need to be reevaluated with methodologies that can accurately determine the strain population structure present during each episode. Such studies will allow for a more accurate quantification of the sterilization efficacy of current and new antituberculosis therapies. From the data presented in this study it is not possible to predict the order in which the different infections occurred. Therefore, it would be unwise to assume that the reinfecting strain will always be overrepresented during disease progression. As described above, reinfection may reactivate a latent infection, which in turn may then be responsible for disease progression. If this scenario is true, it will have important consequences for the interpretation of molecular epidemiologic data, as it is possible that such cases will present with strains that are different from their source cases even though contact existed (4). Alternatively, multiply infected source cases may infect contacts with an underlying strain (not detected by IS6110 DNA fingerprinting), thereby making the inference of contact difficult when using only restriction fragment-length polymorphism data. This study demonstrates that multiple infections are present in patients with active tuberculosis in a high-incidence setting. Most importantly, the initial infection is unable to provide protection against a subsequent infection in this population. This result will have important implications for the understanding of protective immunity and the development and testing of new vaccines and drugs for use in communities where the burden of disease is high. Furthermore, this study highlights the importance of preventing transmission to reduce the risk of exposure and reexposure of all people to active sources of tuberculosis.
The authors thank the National Health Laboratory Service regional laboratory for providing access to specimens, and Mrs. L. Pretorius and Mrs. A. Huysamen for excellent technical support and preparation of specimens. The authors also thank Dr. I. Toms (Department of Health, City of Cape Town) and are indebted to the residents of the epidemiologic field site.
Supported by the GlaxoSmithKline Action TB Program and the IAEA. This article has an online supplement, which is accessible from this issue's table of contents online at www.atsjournals.org Conflict of Interest Statement: R.M.W. has no declared conflict of interest; T.C.V. has no declared conflict of interest; E.M.S. has no declared conflict of interest; M.R. has been employed by GlaxoSmithKline, with a monthly salary, from January 1994 and during this period of employment the techniques and methodology and background knowledge used for her contribution to this paper was developed, but it does not hold direct financial benefit to herself or her institution and she has received sponsorship from GSK as well as from the IUATLD and CDC to attend conferences where aspects of the research projects in which she is involved were presented, again without any specific gain to herself or her institution; N.B. received 111,150 pounds in 2000 and 150,000 pounds per year for 2001–2003 from GlaxoSmithKline Action TB Program as research grants for developing and maintaining an epidemiological field site and for doing a study aimed at identifying surrogate markers for response to treatment in TB patients; N.C.G.v.F. has no declared conflict of interest; P.D.v.P. has received a research grant from Glaxo-SmithKline for TB research. Received in original form May 30, 2003; accepted in final form December 19, 2003
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
METHODS
