INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Keywords: Beijing genotype; multidrug-resistant TB; RFLP; tuberculosis; W-family

The decline of tuberculosis (TB) in industrialized countries, which was accentuated by the introduction of effective treatment, was associated with a decline of interest in research on tuberculosis that left unanswered some fundamental questions concerning the transmission and spread of the disease. Renewed interest in the disease in the 1990s led to the development of new molecular techniques that can differentiate Mycobacterium tuberculosis strains, giving epidemiologists the tools to address some of these issues (1). For example, although it had been generally accepted that the majority of TB cases in low-incidence countries were the result of endogenous reactivations, a study from Denmark found that more than 50% of cases likely resulted from recent infections. Moreover, 40% of these recent transmissions were caused by only two M. tuberculosis strains (4).

The technique most commonly employed in the molecular epidemiology of tuberculosis is strain typing by restriction fragment length polymorphism (RFLP), a method based on the detection of differences in the numbers and locations of the insertion element IS6110 within the chromosomes of M. tuberculosis strains. However, this insertion element has a low but significant frequency of transposition, so that RFLP patterns show a certain amount of instability. The half-life of an IS6110 RFLP pattern has been estimated to be as short as 3 to 4 yr (5). In consequence, minor RFLP differences in the offspring of a parental strain may complicate or confuse long-term studies of molecular epidemiology.

A newer technique, spoligotyping, is based on the detection of the different DNA spacer sequences that lie between direct repeats (DRs) in the M. tuberculosis chromosome (6). The composition of the DR region appears to be a more stable characteristic of a particular strain than the IS6110-associated RFLP (7). In a previous study, spoligotyping was shown to be reliable in identifying strains that were highly similar in their IS6110 RFLPs (8).

Spoligotyping has demonstrated that one strain family of M. tuberculosis is predominant in the Beijing area of China and is common throughout Asia (8) and the western pacific region (9). This strain has been termed the "Beijing" genotype. Spoligotyping also demonstrated that the multidrug- resistant (MDR) "W" strain, which caused tuberculosis in more than 350 patients in New York City and accounted for more than 25% of all MDR TB cases in the United States in the 1990s, is a member of the strain family of the Beijing genotype (10). The Beijing genotype has been documented to be associated with the transmission of drug-resistant tuberculosis in Germany, Russia, Estonia, South Africa, and Colombia (11). In addition, two studies in Vietnam found the Beijing genotype to be significantly associated with younger age of patients, evidence suggesting that it is emerging as an important component of the tuberculosis transmission pattern in that country (11, 15).

The present study describes the introduction of a strain of the Beijing genotype into the population of Gran Canaria Island by an African refugee in 1993, and its explosive spread in this community over the subsequent few years. To our knowledge, this is the first documentation of the introduction and spread of a highly virulent M. tuberculosis strain into a community where all the cases have been systematically evaluated with molecular techniques.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The Canary Islands (Spain) are located in the Atlantic Ocean, close to the North African coast. The most populated island is Gran Canaria, which has 713,768 inhabitants, more than half of whom live in the city of Las Palmas. The island has good health care services, and all inhabitants, including immigrants and homeless persons, have access to free health care. Gran Canaria Island has a well-organized tuberculosis control program (14) and a TB notification rate per 100,000 inhabitants of 32.2 in 1988, 28.5 in 1993, 29.0 in 1996, and 29.3 in 1999 (16). Acid-fast smears and mycobacterial cultures are routinely requested from all patients with respiratory symptoms.

The study population included 566 of the 721 patients (78.5%) diagnosed with tuberculosis on Gran Canaria Island from January 1, 1993 to December 31, 1996 who had bacteriologic confirmation by culture. In addition, 85 (74 living in Las Palmas City) of 244 patients (35%) during the period 1991-1992, from whom cultures were available, and all patients with cultures from which M. tuberculosis was isolated during the last 3 mo of 1999 (40 patients) were evaluated. Medical records were reviewed to obtain demographic data, tuberculosis risk factors, and clinical characteristics of the illness. Epidemiological links were identified. The human immunodeficiency virus (HIV) test was performed on all patients with tuberculosis, accompanied by pre- and posttest counseling. All results of medical examinations were kept in strict confidence.

All patients received a standard 6-mo treatment with isoniazid (H), rifampin (R), and pyrazinamide for the initial 2 mo, and then with H and R for the remaining 4 mo (17).

The M. tuberculosis strains were stored in a 40° C freezer in liquid Dubos medium and subcultured on Löwenstein-Jensen medium. One isolate from each patient was typed by the internationally standardized method of restriction fragment length polymorphism (RFLP), using insertion sequence IS6110 as a probe (1). A cluster of M. tuberculosis isolates was defined as two or more isolates with identical RFLP patterns consisting of five or more bands. To ensure that clustered isolates were not the result of laboratory cross-contamination, patients with a single positive culture but negative acid-fast smears were considered to be false positives if their sputum samples were processed in a laboratory on the same day as a positive smear from another patient with the same RFLP pattern.

After RFLP pattern analysis, all patient included in clusters were further studied by conventional contact tracingclinical records and personal interview of all the patientsfor evidence of epidemiological linkage. Epidemiological linkage was considered to be present when two patients belonged to the same population in a small area.

Spoligotyping was used to confirm that strains belonged to the Beijing genotype. Spoligotyping characterizes the polymorphic direct repeat (DR) region of the M. tuberculosis chromosome (6). The Beijing genotype is defined as strains containing only the last 9 of the 43 possible spacers (8).

RFLP and spoligotyping patterns were subjected to computer- assisted analysis using GelCompar 4.1 (Applied Maths, Kortrijk, Belgium). The results were compared with an extensive database containing the RFLP and spoligotyping patterns of thousands of strains isolated in diverse geographical locations maintained at the National Institute of Public Health and Environmental Protection (RIVM) in The Netherlands and with the database of Spanish strains maintained at the University of Zaragoza (Zaragoza, Spain), including MDR TB strains (18). The value of 75% was specified for minimum correlation and Dice coefficient as similar coefficient for identification of the Beijing genotype because all strains known to belong to the Beijing family of strains are included in this percentage of similarity in our databases.

For statistical analysis, ² and Fisher exact tests were used to analyze for linear trend. A p value of < 0.05 was considered significant.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Five hundred and sixty-six of the 721 M. tuberculosis isolates (78.5%) from the period 1993-1996 were typed by RFLP typing and 409 (72.3%) were found to belong to 1 of 78 clusters containing at least 2 identical strains. The largest cluster (Cluster 1) consisted of 75 isolates with an identical IS6110 RFLP pattern of 15 bands. From 1991 to 1996, an increasing percentage of isolates belonged to Cluster 1 (Table 1, p < 0.001 for trend in proportions). This RFLP pattern was not found in any of the 85 M. tuberculosis strains isolated in the period from 1991 to 1992, suggesting that the strain entered the island in 1993. 

Among the 75 cases in Cluster 1, the median age was 35.0 ± 14.8 yr (range, 3-81 yr). Epidemiological linkage was confirmed for 51 (68%) of the cases in this cluster: 29 (39%) were intravenous drug users, 26 (35%) were HIV positive, 14 (19%) were immigrants (4 from sub-Saharan West African countries, all of them living in the same block), 12 (16%) had spent time in prison, and 9 (12%) were homeless. The median age of all TB cases was 39.9 ± 17.6 yr (range, 1-87 yr), and 13% were intravenous drug users, 15% were HIV positive, 7% were immigrants, 7% had spent time in prison, and 6% were homeless. Figure 1 illustrates the geographic proximity of the clustered cases.

View larger version (27K): [in this window] [in a new window]   Figure 1.   Epidemiological link of patients included in Cluster 1. Map of Las Palmas City, showing the residences of the various patients.

The first case with the strain of Cluster 1 recognized on the island (the apparent index case) was a refugee from Liberia diagnosed with smear-positive pulmonary tuberculosis, with a smear grading of 4+ and laryngeal tuberculosis that had not been diagnosed in July 1993, 6 mo after his arrival on the island. He was an intravenous drug user (IVDU), and HIV negative. He was enrolled in a program with directly observed therapy (17), but because of frequent changes in his domicile he failed to take the prescribed medications on a regular basis and his treatment had to be extended over a prolonged period. The patient's tuberculosis had not been cured when the public health system of the island lost contact with him at the end of 1996. The last M. tuberculosis isolate taken from this patient, in August 1995, was still susceptible to all drugs tested. The patient initially lived with other homeless persons on an old ship in the Las Palmas harbor. The index case had contact with 14 members of 5 communities. Many of the other patients belonging to Cluster 1 (20%) also lived in this part of the city (Port-La Isleta), which is a nexus for illegal drug sales, illegal immigrants, and the homeless. Four patients from this community transmitted TB to another three communities. The entry and prolific spread of this genotype M. tuberculosis strain in the island is shown in Figure 2.

View larger version (32K): [in this window] [in a new window]   Figure 2.   Sequence of entry and spread of the "Beijing" genotype Mycobacterium tuberculosis strain in Gran Canaria Island. Arrows indicate the trend in relations among the cases. *Patients living in the same block.

The RFLP pattern of Cluster 1 was found to be at least 75% identical to the RFLPs of 943 previously identified Beijing genotype strains from 24 different geographic areas, including Beijing, Mongolia, and Vietnam, documented in the RIVM (Figure 3). The database contained four strains, identical to Cluster 1, that were isolated in Africa, one each, in Somalia, Ethiopia, Sierra Leone and Liberia (Figure 4). The 75 strains isolated in Cluster 1 were also analyzed by spoligotyping, and all hybridized to oligonucleotides representing spacers 35 to 43, which is characteristic of the family of strains of the Beijing genotype. Spoligotyping was performed on all 40 M. tuberculosis strains isolated on Gran Canaria Island during the last 3 mo of 1999. Nine of these 40 isolates (22.5%) were found to have the Beijing genotype, suggesting an association with Cluster 1. 

View larger version (51K): [in this window] [in a new window]   Figure 3.   Comparison of RFLPs of Mycobacterium tuberculosis strains in central IS6110 fingerprint database at RIVM. The respective patterns matched within a 75% range with 943 previously found Beijing genotype strains from 24 different geographic areas. Strains with patterns identical to those of Cluster 1 from Gran Canaria Island are presented in Figure 4.

View larger version (62K): [in this window] [in a new window]   Figure 4.   Four patterns of Mycobacterium tuberculosis strains identical to Cluster 1. Lane 1, isolated from Somalia; lane 2, isolated from Ethiopia; lane 3, isolated from Sierra Leone; lane 4, isolated from Liberia; lane 5, isolated from Cluster 1 (GC749).

Initial isolates from all Cluster 1 patients were susceptible to all antibiotics tested. Of these 75 patients, 72 were cured by a standard 6-mo treatment protocol. In two cases, multidrug-resistant strains were subsequently isolated. One of these was HIV positive, alcoholic, and used intravenous drugs, whose tuberculosis was diagnosed in December 1994. After the patient had defaulted from treatment several times, an H and R-resistant strain was isolated from her sputum. She was subsequently hospitalized for 5 mo of treatment with second-line drugs, and in 1999 her sputum was negative by direct examination and culture. The other patient, who was diagnosed in September 1995, also adhered poorly to the treatment regimen, and has remained culture positive with a strain resistant to H and R since December 1997. These patients were two of the four cases, from the whole of Spain in 1998, with a strain of the Beijing genotype registered in the database at the University of Zaragoza.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Molecular epidemiology is a valuable tool in studying the transmission of tuberculosis. It can verify transmission patterns suspected from traditional epidemiologic investigation, reveal unexpected transmission routes, and, as demonstrated here, identify the contribution of highly transmissible M. tuberculosis strains. This report has documented that a strain of the Beijing genotype, not previously identified on Gran Canaria Island, spread over a period of only 4 yr to become the most common isolate on the island. Since its introduction in 1993, this strain has had a considerable impact on the TB incidence on Gran Canaria Island, causing a small increase in the notification rates in the 1993-1996 period in a well-organized tuberculosis control program (16). More than one in eight of all tuberculosis cases diagnosed between 1993 and 1996, and more than one in five cases in late 1999, were caused by this strain. This cluster of 75 patients is one of the largest recorded clusters, comparable to one described in Greenland (19). The emergence of another strain of the Beijing genotype has been reported in Vietnam (11, 15) and Iran (20), where it was isolated from a high percentage of young tuberculosis patients (11). Less than 4% of MDR strains systematically typed in the database at the University of Zaragoza were found to belong to the Beijing genotype when spoligotyping was performed.

The international Gel Compare database, including thousands of RFLP patterns of M. tuberculosis strains, has demonstrated that the Beijing genotype has a worldwide distribution. The Cluster 1 strain of the Beijing genotype apparently entered Gran Canaria Island with an African refugee. Only four strains identical to those of Cluster 1 have been identified in the international Gel Compare database, one each in Somalia, Ethiopia, Sierra Leone, and Liberia. The latter was the country of origin of the African refugee in this study. The Cluster 1 strain rapidly spread to close contacts in his community, which contained individuals with the high-risk characteristics described in several other TB outbreaks (21). Its extensive spread might be explained by a high degree of susceptibility and close contact among the population into which it was introduced and may be due to the exceptional virulence of this strain (24). The key role of prolonged exposure due to poor adherence to treatment, loss to follow-up without documented cure, and the development of multidrug resistance in some individuals in the cluster must also be taken into account. Nonetheless, other strains isolated from members of the same community have not demonstrated the pattern of rapid spread seen with the Cluster 1 strain (17).

Reports documenting the Beijing genotype in predominant, highly transmissible strains found in distinct geographic locations support the idea that it is more virulent than other strains of M. tuberculosis. Chance would predict that a few strains would eventually emerge as predominant in any given population, but this cannot explain how a single genotype has become predominant in many different, geographically separate populations, and implies that the Beijing genotype has some advantage over other strains in its ability to be transmitted and cause disease (25). A previous report (26) from the same community has demonstrated the higher frequency of the same Beijing genotype of the Cluster 1 strain among recurrent cases caused by reinfection than among those resulting from reactivation of the initial strain. These results might suggest a higher level of virulence of the strain demonstrated by its ability to reinfect patients who had been previously cured of disease caused by other strains.

Previous studies have reported a significant correlation between the Beijing genotype and drug resistance (9, 27). Perhaps the virulence of this strain is related to an exceptional capacity to replicate within the patient, and the greater bacterial load increasing the odds for selecting a drug-resistant mutant. The present study was unable to confirm the relation between this strain of the Beijing genotype and the development of drug resistance, as poor adherence to therapy was so obviously a key factor in those cases that developed multidrug resistance.

The outbreak reported here, on an island in Western Europe with a moderate notification rate of tuberculosis, together with other reports from Asia and the former Soviet Union, suggest that this strain could be responsible for an important component of the worldwide resurgence of tuberculosis, and may represent an epidemic within the larger tuberculosis epidemic. A key priority in tuberculosis research is the development of an effective vaccine (28). The integration of work on the molecular epidemiology, immunology, and molecular genetics of tuberculosis, with an emphasis on study strains of differing virulence, might help in the development of a more effective vaccine.