Genotypic patterns of multiple isolates of M. tuberculosis from tuberculous HIV patients


correspondence Dr L. S. Fonseca, Instituto de Microbiologia, UFRJ, Centro de Ciências da Saúde, Bloco I, Ilha do Fundão, 21.941–590, Rio de Janeiro, Brazil; E-mail:


Summary We investigated whether the recurrence of tuberculosis in HIV-infected patients is due to an exogenous reinfection or relapses after antituberculosis chemotherapy. We reviewed clinical information on 32 patients at a Rio de Janeiro hospital from whom multiple Mycobacterium tuberculosis isolates were taken. All isolates were analysed by DRE-PCR fingerprinting technique, and those with identical DRE-PCR patterns were analysed by the RFLP method. Twenty patients had M. tuberculosis simultaneously isolated from different organs. These patients and nine others with sequential positive cultures after 2 months of therapy showed stable DRE-PCR and RFLP patterns. One patient's isolate became resistant to isoniazid, but the molecular pattern remained unchanged despite the development of drug resistance. In three patients, the DRE-PCR patterns of the isolates changed dramatically. Clinical and microbiological evidence was consistent with active tuberculosis caused by a new strain of M. tuberculosis. The exogenous reinfection of the three patients was not due to an outbreak, but the isolates from each patient showed unique patterns.


Tuberculosis (TB) is an important manifestation of acquired immunodeficiency syndrome (AIDS) ( Pitchenik 1990). However, the response of HIV-infected patients to antituberculosis therapy is similar to that of non-HIV infected patients ( Small et al. 1991 ; Perriens et al. 1995 ). Regardless of the efficacy of antituberculosis therapy, recurrence of pulmonary tuberculosis in HIV-infected patients is frequent in developing countries with high incidence of tuberculosis ( Perriens et al. 1991 ). Recurrence can be a consequence of either exogenous reinfection or relapses after inadequate therapy. Using molecular methods of characterizing strains of tubercle bacilli, it is possible to determine whether the disease in a patient is due to a new infection or the reactivation of an old one ( Daley et al. 1992 ; Small et al. 1993 ; Alland et al. 1994 ; van Rie et al. 1999 ). This report describes the use of a double-repetitive-element PCR (DRE-PCR) fingerprint, a simple and rapid technique to determine genotypic variability of M. tuberculosis multi-isolates from a single patient and for differentiating between reactivation and exogenous reinfection in HIV-infected patients.

Materials and methods


Evandro Chagas Hospital is a 30-bed hospital affiliated to the Oswaldo Cruz Foundation, a research institution providing secondary care to HIV patients. From 1990 to 1994, 192 patients in this hospital were found to have microbiologically confirmed tuberculosis. Clinical and demographic data from the patients were collected by reviewing medical charts.

Laboratory methods

Specimen registers of the bacteriological laboratory were reviewed and all patients who had multiple isolates of M. tuberculosis were enrolled. The isolates were maintained in Middlebrook 7H9 broth (Difco Laboratories, USA) at −20 °C, then subcultured in Loewenstein-Jensen medium for 4 weeks at 37 °C. Their susceptibility to isoniazid (0.2 μg/ml), rifampin (100 μg/ml), streptomycin (4 μg/mL), pyrazinamide (40 μg/ml), ethambutol (2 μg/ml) and ethionamide (20 μg/ml) was determined by the proportion method ( Canetti et al. 1963 ). All M. tuberculosis isolates were typed by a DRE-PCR fingerprinting screening technique based on the IS6110 and GC-rich repetitive sequence as described by Friedman et al. (1995) . Isolates with identical DRE-PCR fingerprints were analysed by the restriction fragment length polymorphism technique (RFLP) according to the standardized method of van Embden et al. (1993) .


From 1990 to 1994, all 32 patients with multiple isolates of M. tuberculosis were selected for this study. Twenty patients had M. tuberculosis isolated simultaneously from two or more noncontiguous organs, confirming the diagnosis of disseminated tuberculosis ( Table 1). The DRE-PCR and RFLP fingerprints showed the same pattern for the isolates from each patient, confirming that they were infected by only one strain of M. tuberculosis.

Table 1. M. tuberculosis strains simultaneously isolated from different specimens of HIV seropositive patients Thumbnail image of

Twelve patients with sequential positive cultures after at least 2 months of daily therapy with rifampin, isoniazid and pyrazynamide are the focus of this report. The isolates from nine patients showed stable patterns by DRE-PCR and RFLP techniques ( Figures 1 and 2), six patients with drug-sensitive M. tuberculosis and three with drug-resistant strains ( Table 2). In all but one patient, the isolates also showed stable susceptibility patterns. The patient whose isolates showed different susceptibility patterns (patient 5) had three positive cultures, one in 1990 and two one year later. The organism isolated from stool was resistant to isoniazid although the one from bronchoalveolar lavage isolated one month later remained sensitive to all antimycobacterial drugs tested. However, all three isolates showed identical DRE-PCR and RFLP patterns. All nine individuals were noncompliance patients and never considered cured.

Figure 1.

DRE-PCR patterns of the M. tuberculosis isolated in different periods from patients persistently infected. a) PM: molecular weight standard (100 bp ladder); lines A, B and C isolates from patient 5; lines D, E, F and G isolates from patient 14; lines H and I isolates from patient 15; CN negative control; CP positive control. b) PM: Molecular weight standard (100 bp ladder); lines A and B isolates from patient 2; lines C, D, E isolates from patient 3; lines F and G isolates from patient 6; lines H, I, and J isolates from patient 16; lines K and L isolates from patient 27; lines M and N isolates from patient 31.

Figure 2.

RFLP patterns of the M. tuberculosis isolated from patients persistently infected. M: DNA molecular weight marker (VII Dig Labeled Boehringer); lines 2a and 2b isolates from patient 2; lines 3a, 3b and 3c isolates from patient 3; lines 5a, 5b and 5c isolates from patient 5; lines 6a and 6b isolates from patient 6; lines 15a and 15b isolates from patient 15; lines 16a, 16b and 16c isolates from patient 16.

Table 2.  Epidemiologic characteristics of the HIV patients with persistent M. tuberculosis isolates Thumbnail image of

In three patients the DRE-PCR and RFLP patterns of the original strains differed dramatically from the patterns of the strains obtained subsequently ( Table 3, Figures 3 and 4). Those patients were treated with standard therapy for tuberculosis (isoniazid, rifampin and pyrazinamide daily for 2 months followed by isoniazid and rifampin daily for 7 months). They were considered cured by radiological, clinical and bacteriological criteria. Their clinical course is described below.

Table 3.  Epidemiologic characteristics of the HIV patients with exogenous reinfection by M. tuberculosisThumbnail image of
Figure 3.

DRE-PCR patterns of the M. tuberculosis isolated from patients with exogenous reinfection. PM: Molecular weight standard (100 bp ladder); lines A, B and C isolates from patient 7; lines D, E, F and G isolates from patient 11; lines H and I isolates from patient 32.

Figure 4.

Dendogram of the RFLP patterns of the M. tuberculosis isolated from patients with exogenous reinfection. Lines 7a, 7b and 7c isolates from patient 7; lines 11a and 11d isolates from patient 11; lines 32a and 32b isolates from patient 32.

Patient 7 had culture-confirmed tuberculosis and received standard therapy for 9 months. Two years later he was readmitted to the hospital with clinical and radiological evidence of tuberculosis and a new strain of M. tuberculosis was isolated. Three months after the end of the second period of treatment, he again showed clinical signs of TB due to reinfection with a new strain of M. tuberculosis. Once more, he received standard therapy and showed negative culture. All strains were fully sensitive to antimycobacterial drugs. The DRE-PCR and RFLP from the isolates showed different patterns ( Figures 3 and 4), demonstrating that the patient had three episodes of tuberculosis due to exogenous reinfection with different strains over a period of 4 years. The first isolate showed a 12-band-RFLP pattern, the second strain isolated 1 year later a 17-band pattern and the third an 8-band-RFLP pattern.

Patient 11, from whom a fully sensitive strain was originally isolated, showed negative culture after 9 months of therapy. Six months later he was readmitted to hospital with a new episode of tuberculosis and an INH-resistant M. tuberculosis strain as demonstrated by DRE-PCR and RFLP analysis ( Figures 3 and 4). He received standard therapy for 8 months; then the treatment was discontinued due to drug-induced hepatitis. After 15 months he was readmitted to the hospital and died one month later. The strains isolated in 1991 had a 16-band-RFLP pattern and those isolated in 1992 and 1994 had a 4-band pattern.

Patient 32 underwent standard therapy for 9 months. After 2 years he was readmitted to the hospital with sputum-positive culture and clinical and radiological signs of disseminated tuberculosis and died 3 days later. The DRE-PCR and RFLP analysis revealed that the patient was reinfected with a new strain of M. tuberculosis ( Figures 3 and 4). The first isolate showed a 18-band-RFLp pattern and the second a 16-band pattern.


The possibility that persons previously infected with M. tuberculosis could be exogenously reinfected has been debated for decades ( Stead 1967; Romeyn 1970). However, this was thought to occur rarely because of the immunity conferred by the initial infection. The pandemic of HIV has modified the epidemiology, pathogenesis and clinical manifestations of TB ( Hopewell 1992; Dolin et al. 1994 ). Recommendations for treatment are based on studies with short follow-ups, and there is great concern about the long-term validity of the treatment regimens when applied to HIV-infected patients ( Hawken et al. 1993 ; Pulido et al. 1997 ).

The observation that HIV-infected tuberculous patients might have subsequent episodes of the disease as a result of exogenous reinfection with M. tuberculosis, complicates the evaluation of clinical trials of chemotherapy of the tuberculosis. Currently, relapses after therapy are assumed to result from inadequacies in the regimen's efficacy or from patient noncompliance. However, if exogenous reinfection is common, relapse may reflect reinfection after inadequate initial therapy. Previous reports have documented by DNA fingerprint methods the exogenous reinfection in HIV-positive and negative patients ( Nardell et al. 1986 ; Small et al. 1993 ; Horn et al. 1994 ; Sahadevan et al. 1995 ; Schafer et al. 1995 ; Turet et al. 1997 ). Recently, van Rie et al. (1999) studied 698 patients from an area with a high rate of TB over a period of 6 years and found 16 HIV-seronegative patients with a second episode of tuberculosis. Using RFLP analysis, they proved that in 75% of those cases, the second episode was attributable to exogenous reinfection of a new M. tuberculosis strain.

Although IS6110 RFLP analysis is the international standard method for DNA fingerprint studies on M. tuberculosis strains ( van Embden et al. 1993 ; Small et al. 1994 ; Yang et al. 1994, 1995 ; Torrea et al. 1996 ; Bishai et al. 1998 ), it has several limitations in developing countries. Recently, the discriminative power of the 12 typing methods of M. tuberculosis was compared in an interlaboratory study ( Kremer et al. 1999 ). DRE-PCR yielded 63 banding patterns in a set of 90 M. tuberculosis strains from 38 countries, while RFLP yielded 84 patterns. When we compared the M. tuberculosis patterns obtained by IS6110 RFLP and DRE-PCR methods, the discriminatory power of DRE-PCR was far inferior to that of RFLP typing. Many isolates with the same DRE-PCR patterns have RFLP patterns with a low degree of similarity. By contrast none of the RFLP clusters was subdivided on the basis of DRE-PCR analysis (F. C. O. Fandinho, personal communication). However, laboratory methods need to be simplified to increase their accessibility for clinical laboratories in developing countries. As established by Friedman et al. (1995) , DRE-PCR is a simple and rapid screening technique for typing purposes. It can be performed in any laboratory with basic equipment and facilities for molecular techniques. Although DRE-PCR is not recommended for population-based epidemiological studies with a large number of isolates, it seems to be useful in differentiating the new episode of tuberculosis as a reactivation of latent M. tuberculosis infection from an exogenous reinfection by comparing multiple isolates from the same patient.

RFLP analysis confirmed the results obtained from DRE-PCR on the strains from patients 7, 11 and 32, who were considered to have an exogenous reinfection with new M. tuberculosis strains. In cases with different DRE-PCR patterns, tested strains will have different RFLP patterns, making additional typing unnecessary.

For nine of 12 patients with subsequent positive culture for M. tuberculosis, DRE-PCR analysis confirmed a true relapse. Identical isolates by DRE-PCR were confirmed by RFLP method. More important, however, was the demonstration of exogenous reinfection in three patients. Considering different patterns in the isolates from one patient, a consequence of cross-contamination in the laboratory does not seem likely, since the institution is a 30-bed hospital, with few clinical specimens to be processed every day. Secondly, all patients with exogenous reinfection showed clinical and radiological evidence of tuberculosis compatible with the bacteriological results. Another hypothesis against exogenous reinfection could be the evolution of the strain; however, the degree of change in the RFLP patterns shown by the isolates from the same patient demonstrated that they were indeed reinfected, making this hypothesis implausible.

The lack of association between fingerprint patterns and drug susceptibility was exemplified in patient 5, whose original strain became isoniazid-resistant during treatment while the DNA pattern remained unchanged. This phenomenon has already been observed by others ( Small et al. 1993 ; Sahadevan et al. 1995 ).

Studies with multiple isolates of M. avium have demonstrated that HIV-infected patients can present mixed infections with more than one strain of M. avium simultaneously ( Tsang et al. 1992 ; Saad et al. 1997 ). Contrary to what has been described with M. avium disease, our study showed that the patients with M. tuberculosis disease did not present mixed infections. In all strains simultaneously isolated from the same patient we detected only one DRE-PCR and RFLP pattern. Those patients had disseminated tuberculosis, which is a frequent event in AIDS patients in our setting ( Grinsztejn et al. 1997 ).

It has been shown in HIV-seropositive patients that reinfection can occur even during treatment ( Small et al. 1993 ; Horn et al. 1994 ). But in our study exogenous reinfection occurred only after a cure of the initial episode of TB. Because of the small number of patients studied, we are not able to establish the prevalence of reinfection tuberculosis in HIV-infected individuals. But it is theoretically possible that patients with advanced HIV infection who have completed effective antituberculosis therapy remain susceptible to reinfection and develop tuberculosis when re-exposed to M. tuberculosis. Most patients were followed up until death, and the survival time after tuberculosis (about 16 months) may not be long enough for patients with advanced HIV infection to develop a new episode of tuberculosis.

Although we were not looking for an epidemiological connection among cases, comparison of the DNA fingerprints from all patients did not show any correlation. These results indicate that the patients were probably infected by M. tuberculosis strains from the community. Our study demonstrated that DRE-PCR typing can be used as a rapid method to exclude epidemiological relationship, as the evaluation of a new episode of tuberculosis after culture-negative M. tuberculosis in HIV-infected patients.


This research was supported by CNPq, FAPERJ and FINEP.