Drobniewski F, Balabanova Y, Nikolayevsky V, Ruddy M, Kuznetzov S, Zakharova S, Melentyev A, Fedorin I. Drug-Resistant Tuberculosis, Clinical Virulence, and the Dominance of the Beijing Strain Family in Russia. JAMA. 2005;293(22):2726-2731. doi:10.1001/jama.293.22.2726
Author Affiliations: HPA Mycobacterium Reference
Unit, Department of Microbiology and Infection, Guy’s King’s and
St Thomas’ Medical School, (Dulwich), London, England (Drs Drobniewski,
Balabanova, Nikolayevsky, and Ruddy); Samara Regional TB Service, Samara Oblast
Dispensary, Russia Federation (Drs Balabanova and Fedorin); Samara Regional
Ministry of Health, Russia Federation (Dr Kuznetzov); Samara City TB Service,
Russia Federation (Dr Zakharova); Samara Prison TB Service, Samara, Russia
Federation (Dr Melentyev).
Context Tuberculosis and multidrug-resistant tuberculosis is a serious public
health problem in Russia.
Objective To address the extent of “Beijing strain” transmission in
the prison/civil sectors and the association of drug resistance, clinical,
and social factors with the Beijing genotype.
Design and Setting Cross-sectional population-based molecular epidemiological study of
all civilian and penitentiary tuberculosis facilities in the Samara region,
Patients Consecutively recruited patients with bacteriologically proven tuberculosis
(n = 880).
Main Outcome Measure Proportion of Beijing strains and association with drug resistance,
human immunodeficiency virus infection, imprisonment, radiological, clinical,
and other social factors.
Results Beijing-family strains (identified by spoligotyping and composed of
2 main types by mycobacterial interspersed repetitive unit analysis) were
predominant: 586/880 (66.6%; 95% confidence interval [CI], 63.4%-69.7%) with
a significantly higher prevalence in the prison population (rate ratio [RR],
1.3; 95% CI, 1.2-1.5) and those aged younger than 35 years (RR, 1.2; 95% CI,
1.0-1.3). Comparable proportions were co-infected with the human immunodeficiency
virus (≈10%), concurrent hepatitis B and C (21.6%), drank alcohol (≈90%),
smoked (≈90%), and had a similar sexual history. Drug resistance was nearly
2-fold higher in patients infected with Beijing strains compared with non-Beijing
strains: multidrug resistance (RR, 2.4; 95% CI, 1.9-3.0), for isoniazid (RR,
1.8; 95% CI, 1.5-2.1), for rifampicin (RR, 2.2; 95% CI, 1.7-2.7), for streptomycin
(RR, 1.9; 95% CI, 1.5-2.3), and for ethambutol (RR, 2.2; 95% CI, 1.6-3.2).
Univariate analysis demonstrated that male sex (odds ratio [OR], 1.5; 95%
CI, 1.1-1.9), advanced radiological abnormalities (OR, 3.3; 95% CI, 1.3-8.4),
homelessness (OR, 5.6; 95% CI, 1.1-6.3), and previous imprisonment (OR, 2.0;
95% CI, 1.5-2.7) were strongly associated with Beijing-strain family disease.
Multivariate analysis supported previous imprisonment to be a risk factor
(OR, 2.0; 95% CI, 1.4-3.3) and night sweats to be less associated (OR 0.7;
95% CI, 0.5-1.0) with Beijing-strain disease.
Conclusions Drug resistance and previous imprisonment but not human immunodeficiency
virus co-infection were significantly associated with Beijing-strain infection.
There was evidence that Beijing isolates caused radiologically more advanced
Drug-sensitive and resistant tuberculosis (TB) rates continue to increase
globally and multidrug-resistant TB (MDR-TB) has become a serious public health
problem in many regions including countries of the former Soviet Union such
as Russia, with an incidence rate reaching 83.0/100 000 in 2002.1 Molecular epidemiological techniques have contributed
to the understanding of TB transmission2- 4 and
identified related Mycobacterium tuberculosis–strain
families such as the “Beijing,” which may have distinctive properties.5 This family includes strain W responsible for MDR-TB
outbreaks in the United States and other low TB–incidence countries.2- 4
Studies have demonstrated high Beijing-strain prevalence in Asia and
former Soviet Union countries.6- 9 Explanations
for the rapid dissemination of Beijing strains have included a putative higher
virulence, enhanced transmissibility, an ability to escape from BCG vaccine–induced
protection, greater mutability, and an association with multidrug resistance
rendering cure more difficult and prolonging infectivity.1,3,5,7,9- 14 However,
by contrast, only a very weak association with resistance and the above factors
was noted by Lillebaek et al.15
Many TB cases originate among prisoners in Russia where there are almost
1 million incarcerated persons (approximately 0.7% of the total population).16 For example, the TB incidence rate was nearly 25-fold
higher among prisoners compared with the civil population in the Tomsk region.16 In the Samara region, the incidence of TB among civilians
and prisoners at the time of the study was 86.1/100 000 and 2190.0/100 000,
respectively, with incidence in pretrial detention centers of 1889.9/100 000.
Overcrowding, inadequate ventilation, and long periods of incarceration facilitate
transmission. Poor patient adherence to treatment during and after incarceration
and high loss to follow-up after release from prison encourage the development
of drug resistance, which facilitates transmission into the general population.13,17,18
We performed one of the largest (and the largest in Russia) population-based
molecular epidemiological studies to systematically address: (1) Beijing strain
transmission in the prison and civil sectors in the Samara region; and (2)
correlation of drug resistance, clinical, and social factors with Beijing
genotype to support or contradict immunopathological studies suggesting that
Beijing strains had either enhanced pathogenicity or association with drug
Consecutive pulmonary TB patients aged 18 years and older, from all
civilian TB dispensaries (18) and the prison TB hospital (admitting all TB
cases in the prison sector) were included over 1 year (September 2001-September
2002). All patients provided written consent. Patients were interviewed by
a team of trained Russian health staff who recorded clinical (including details
of radiological examination) and social data. A structured questionnaire was
used to collate data that were supplemented and verified with information
from the medical notes.
The questionnaire and study were developed and approved by Federal Tuberculosis
Institutions, Samara TB Service and Ethics Committee, and the Samara Health
Ministry under whose auspices the study was conducted. The following data
were collected: age, sex, accommodation, previous imprisonment, presence of
TB symptoms and signs (cough, phlegm, hemoptysis, weight loss, night sweats,
fever, shortness of breath, chest pain, fatigue, and body mass index), investigations
including chest radiograph (cavitation, lung areas affected, and nature of
pathology), blood pressure, erythrocyte sedimentation rate, and human immunodeficiency
virus (HIV) status (HIV testing was performed routinely on all TB patients),
history of treatment, history of TB contacts, social and sexual health history
(including alcohol consumption, smoking history, history of sexually transmitted
infections, and sexual activity).
Strains of Mycobacterium tuberculosis were
isolated at the Regional TB laboratory, and drug susceptibility testing was
performed with the resistance ratio method using Lowenstein–Jensen media
in Samara, Russia and London, England. Spoligotyping was used to identify
Beijing family isolates and performed as previously described.19 Mycobacterial
interspersed repetitive unit (MIRU) analysis at 12 loci20 was
also performed on 113 Beijing isolates selected proportionally from patients
in the prison and civilian TB facilities.
Data were analyzed using EpiInfo version 6 (US Centers for Disease Control
and Prevention, Atlanta, Ga), and SPSS version 10 (SPSS Inc, Chicago, Ill),
statistical software packages. Pearson χ2 or Fisher 2-tailed
test and the Mann-Whitney U test were used to compare
categorical and continuous variables, respectively.
All variables were included in a univariate analysis. From the univariate
analysis, all variables that had a P value of ≤.01
(as multiple comparisons were used) were included in a subsequent multivariate
logistic regression analysis. Multivariate analysis was performed by binary
logistic regression using forward stepwise and Wald statistical criteria for
logistic regression collinearity and confounding, and interactions were assessed.
The final regression model used 93.3% of all available information due to
some missing clinical and social data.
Nearly all eligible individuals participated in the study (<2% did
not participate). Beijing family strains were dominant overall among the 880
patients recruited (253 civilians and 333 prisoners; 586/880, 66.6%; 95% confidence
interval [CI], 63.4%-69.7%) of all isolates (Table 1).21 A significantly higher
prevalence was seen in the prison population (rate ratio [RR], 1.3; 95% CI,
1.2-1.5), and was more common in those aged younger than 35 years compared
with older patients (RR, 1.2; 95% CI, 1.0-1.3). Two MIRU-12 loci profiles
were predominant (Table 2) as recently
described in a smaller study of isolates in St Petersburg.22 In
this study, however, MIRU type 2 was more common among prisoners while MIRU
type 1 was more common among civilian patients.
The number and proportion of Beijing and non–Beijing-infected
patients with TB symptoms and signs, chest radiological appearances, social
problems, and sexually transmitted diseases are given in Table 3. Comparable proportions in the 2 groups were co-infected
with HIV (≈10%) and hepatitis B/C (≈20%), drank alcohol, smoked, and
had a similar history of sexually transmitted diseases (Table 3). The number of episodes of imprisonment for Beijing and
non–Beijing-infected patients is given in the Figure. Nearly three quarters of patients infected with Beijing
strains had been in prison at some time and many had been imprisoned more
than once, typically 2 to 4 times, while more than half of patients infected
with non-Beijing strains had a history of imprisonment. Recreational drug
use with opiates was common, usually intravenously in both groups. Prison
appears to be the major site for contacts with TB cases.
A risk factor analysis was conducted for Beijing-strain family–associated
infection. Univariate analysis demonstrated that male sex (OR, 1.5; 95% CI,
1.1-1.9), older age (OR, 1.3; 95% CI, 1.1-1.7), homelessness (OR, 5.6; 95%
CI, 1.1-6.3), and previous imprisonment (OR, 2.0; 95% CI, 1.5-2.7) were strongly
associated with Beijing-strain family–associated disease (Table 4). Advanced radiological damage (multiple zones affected
with fibrotic changes and widespread cavitation) was also significantly associated
based on univariate analysis (OR, 3.3; 95% CI, 1.3-8.4). There was less association
based on univariate analysis with night sweats (OR, 0.7; 95% CI, 0.5-0.9).
Multivariate analysis (Table 4) supported
previous imprisonment to be a risk factor (OR, 2.0; 95% CI, 1.4-3.3).
The number of prison episodes did not differ significantly between Beijing
and non-Beijing–infected patients, suggesting that a single episode
was sufficient to acquire Beijing-strain TB. Previous treatment for longer
than 4 weeks and male sex failed to reach statistical significance (the latter
presumably related to the confounding effect of imprisonment on sex, because
the majority of prisoners were men). Although homelessness failed to reach
significance on multivariate analysis, living in one’s own apartment
(ie, the reverse of homelessness in terms of independent stable accommodation
and a surrogate marker of wealth) was less associated with Beijing infection
(Table 4). There were no significant
differences in BCG vaccination rates in patients infected with Beijing strains
(88.6%; 452/510) and in non-Beijing strains (90.6%, 213/235; 95% CI for the
difference, −2.6% to 6.6%); vaccination status was unknown for 74 and
55 patients, respectively.
Drug susceptibility testing on 561 viable isolates detected high levels
of resistance with significantly higher rates of resistance among strains
belonging to the Beijing family (Table 5).
Resistance to isoniazid, rifampicin, multidrug resistance, streptomycin, and
ethambutol were more than 2-fold higher in Beijing- vs non-Beijing–infected
patients (Table 5).
The high prevalence of drug-resistant Beijing–associated TB and
increasing rates of both legal and illegal migration from Russia to the United
States and Western Europe makes this study relevant to all physicians treating
patients with TB. The expansion of the European Union includes states of the
former Soviet Union with a substantial ethnically Russian population that
until recently could freely migrate within the Russian Federation and cases
of MDR-TB in former Russian nationals have been identified.
Beijing family strains comprised two thirds of the isolates in this
population spanning the entire prison and civilian TB systems. Drug resistance
rates were significantly higher, particularly for MDR-TB among those infected
with the Beijing genotype. Co-infection with HIV did not appear to have a
significant impact on Beijing-associated characteristics and high rates of
co-infection (approximately 10%) were seen in both groups. High rates of concurrent
hepatitis B and C infection as well as HIV were seen overall, reflecting on
the importance of intravenous drug abuse in this population.
There was some evidence that clinical presentation differed between
the Beijing- and non–Beijing-infected groups. Although a smaller study
in the Netherlands23 showed no difference in
radiological appearance, this study did show that Beijing-infected patients
were more likely to have radiological features of advanced disease supporting
in vivo studies demonstrating a different immunopathological response in Beijing-infected
mice.24 There was a lower likelihood of night
sweats in Beijing-associated disease. The underlying cause of night sweats
remains unclear but is presumably associated with a heightened TH1
cytokine response, which might be reduced in Beijing infection. The study
by Lopez et al,24 which demonstrated that BALB/c
mice infected with Beijing isolates exhibited a more severe pathology also
noted a high but transient tumor necrosis factor α (TNF-α) and
low interferon γ output. Consistent with these observations were the findings
of a study in which a biologically active lipid species (a polyketide synthase–derived
phenolic glycolipid [PGL]) produced by Beijing strains inhibited the release
of proinflammatory cytokines including tumor necrosis factor α; PGL
appeared to be responsible for the “hyper-lethality” of Beijing
strains in murine models.25
Beijing-infected patients were more likely to have been treated previously.
Whether this implies greater treatment failure or relapse due to intrinsic
properties of Beijing strains per se or is due to the association with drug
resistance requires further study.
The younger age of those infected with Beijing strains would support
recent and active transmission, particularly among prisoners who are younger
than civilian TB patients and who are more likely to be homeless or to share
communal accommodation before imprisonment and on release.
Definitive TB contacts were identified in 250/582 (43.0%) of cases for
Beijing and 98/289 (33.9%) for non-Beijing–individuals with the majority
occurring in the prison (145/243, 59.7% and 50/92, 54.3%), respectively. Nevertheless,
there was no other significant difference in the nature of TB contact between
Beijing and non–Beijing-infected groups. Although 2 MIRU types predominated
in both the prison and civilian sector, the proportion of each type differed
(ie, MIRU type 223325153533 was more common among civilians and type 223325173533
was more common in prisoners with TB). The exact significance of this is unclear
but the allelic difference may confer variation in pathogenicity or transmissibility
within the populations (ie, sub-Beijing families may have different and advantageous
properties, which confounds comparison of all patients infected with Beijing
and non-Beijing strains). A larger comparative analysis of patients with the
2 types is needed to resolve this.
The higher rates of drug resistance among Beijing isolates in the prison
(nearly twice as high as in the civilian population) supports imprisonment
as the principal source of drug resistance. Smaller molecular epidemiological
studies in the former Soviet Union6- 9,19 demonstrated
that a high proportion of TB strains overall belonged to the Beijing family
but were limited to either prison or civilian patients or did not include
knowledge of patients’ HIV or clinical status.
When introduced de novo, Beijing-strain family isolates have been shown
to spread successfully. For example, on Gran Canaria, 1 of the 7 Canary Islands,26 rapid dissemination of the Beijing genotype was observed
over a 3-year period from its introduction onto the island in 1993. It has
been postulated that Beijing isolates are more virulent and also that they
may be escape mutants circumventing BCG-induced immune protection. In a country
such as Russia with a historical policy of multiple BCG vaccinations throughout
childhood, it is possible that the widespread prevalence of Beijing isolates
in this setting is because of an evolutionary advantage over other strains.
Nevertheless, we have not discovered significant differences between the BCG
status of patients infected with Beijing and non-Beijing strains. The widespread
prevalence of Beijing isolates in this setting is, therefore, likely to be
a combination of high levels of drug resistance that reduce cure rates and
maintain a pool of infectious cases, and the opportunity for institutional
spread particularly in prison and intrinsic properties of the Beijing family
itself in the interaction with the host.
Given the existing epidemiology, rapid and inexpensive methods for diagnosing
rifampin-, isoniazid-, and multidrug-resistant TB are urgently required, coupled
with improved institutional cross-infection procedures for both TB and HIV.
Correction of current weaknesses in the system for TB health care delivery,
accompanied by the implementation of a directly observed therapy short course
(DOTS-style program), supplemented by rapid diagnosis of resistance and appropriate
regimens for the treatment of highly drug-resistant TB are required. Delivery
of antiretroviral therapy for those with HIV infection is essential.
This was a large prospective population-based study but larger clinical
and immunopathological studies are needed with more discriminating molecular
epidemiological tools to verify enhanced virulence of the Beijing family (or
other strain-family groups) in humans.
Corresponding Author: Francis Drobniewski,
MD, PhD, HPA Mycobacterium Reference Unit, Clinical Sciences Center, Institute
of Cell and Molecular Sciences, Queen Mary’s School of Medicine, Newark
Street, London E1, England (firstname.lastname@example.org).
Author Contributions: Dr Drobniewski had full
access to all of the data in the study and takes responsibility for the integrity
of the data and the accuracy of the data analysis.
Study concept and design: Drobniewski, Balabanova,
Acquisition of data: Drobniewski, Balabanova,
Ruddy, Zakharova, Fedorin.
Analysis and interpretation of data: Drobniewski,
Drafting of the manuscript: Drobniewski, Balabanova,
Critical revision of the manuscript for important
intellectual content: Drobniewski, Balabanova, Ruddy, Kuznetzov, Zakharova,
Statistical analysis: Balabanova, Drobniewski.
Obtained funding: Drobniewski.
Administrative, technical, or material support:
Drobniewski, Balabanova, Ruddy, Kuznetzov, Zakharova, Melentyev, Fedorin.
Study supervision: Drobniewski, Balabanova,
Ruddy, Zakharova, Fedorin.
Financial Disclosures: None reported.
Funding/Support: This work was funded by the
UK Department for International Development (CNTR 00 0134).
Role of the Sponsor: The funder had no role
in the design or conduct of the study and no role in the collection, management,
analysis, or interpretation of the data or preparation, review, or approval
of the manuscript.
Acknowledgment: We thank the staff of the Moscow
Federal Central Tuberculosis Research Institute and Research Institute for
Phthsiopulmonology and Samara Regional TB service dispensaries and hospitals
for their valuable advice and support. We are particularly grateful to the
bacteriologists, physicians, and nurses as well as the patients who took part
in the study.