Goldfeld AE, Delgado JC, Thim S, Bozon MV, Uglialoro AM, Turbay D, Cohen C, Yunis EJ. Association of an HLA-DQ Allele With Clinical Tuberculosis. JAMA. 1998;279(3):226-228. doi:10.1001/jama.279.3.226
From the Divisions of Adult Oncology (Dr Goldfeld and Ms Uglialoro) and Immunogenetics (Drs Delgado, Bozon, Turbay, and Yunis) and the Blood Bank (Ms Cohen), Dana-Farber Cancer Institute, Boston, Mass, and the Cambodian Health Committee, Phnom Penh (Mr Sok).
Context.— Although tuberculosis (TB) is the leading worldwide cause of death due
to an infectious disease, the extent to which progressive clinical disease
is associated with genetic host factors remains undefined.
Objective.— To determine the distribution of HLA antigens and the frequency of 2
alleles of the tumor necrosis factor α (TNF-α) gene in unrelated individuals with clinical TB (cases) compared with
individuals with no history of clinical TB (controls) in a population with
a high prevalence of TB exposure.
Design.— A 2-stage, case-control molecular typing study conducted in 1995-1996.
Setting.— Three district hospitals in Svay Rieng Province in rural Cambodia.
Patients.— A total of 78 patients with clinical TB and 49 controls were included
in the first stage and 48 patients with TB and 39 controls from the same area
and socioeconomic status were included in the second stage.
Main Outcome Measures.— Presence of HLA class I and class II alleles determined by sequence-specific
oligonucleotide probe hybridization and presence of 2 TNF-α alleles determined by restriction fragment length polymorphism analysis.
Results.— In the first stage, 7 DQB1*0503 alleles were detected among 156 alleles
derived from patients with TB, whereas no DQB1*0503 alleles were found among
the 98 alleles derived from controls (P=.04). There
was no detectable difference in the distribution of the 2 TNF-α alleles in patients with TB compared with controls. In
the second stage, we tested for the presence of a single variable, the DQB1*0503
allele, and found 9 DQB1*0503 alleles among 96 alleles derived from patients
with TB and no DQB1*0503 alleles among 78 alleles in controls (P=.005).
Conclusions.— The HLA-DQB1*0503 allele is significantly associated with susceptibility
to TB in Cambodian patients and, to our knowledge, is the first identified
gene associated with development of clinical TB.
ABOUT A THIRD of the earth's population is infected with Mycobacterium tuberculosis,1- 3
the bacteria that causes tuberculosis (TB). Infection with M tuberculosis results in a variety of conditions ranging from asymptomatic
infection to progressive pulmonary or extrapulmonary TB and death.4 Approximately 1 in 10 of those infected will progress
to active disease during their lifetime.1- 3
Tuberculosis is the leading cause of death due to an infectious disease worldwide,
accounting for approximately 3 million deaths annually.1- 3
Individuals who have impaired cell-mediated immunity due to chemotherapy,
steroid use, neoplasia, or the acquired immunodeficiency syndrome (AIDS) have
a greatly increased risk of activation of quiescent infection with M tuberculosis .4- 6
Poverty also has been implicated as a cofactor in disease progression.7,8 Certain populations are at risk for
increased susceptibility to infection and progressive disease due to M tuberculosis,9- 13
and in several populations the HLA class II DR2 serotype is associated with
clinical TB.11- 13
Mutations in the interferon-γ receptor gene have been associated with
progressive atypical mycobacterial infection14
and with Calmette-Guérin bacillus (Mycobacterium
bovis) infection.15 Tumor necrosis factor α
(TNF-α) appears to play an important role in TB pathogenesis, including
granuloma formation and containment of TB infection16,17
and impaired TNF-α secretion due to defective signaling through the
interferon-γ receptor gene may be involved in disease progression.14,15
To determine whether specific HLA class I or class II alleles are associated
with clinical TB, we performed a 2-stage study of molecular typing of HLA
class I and class II alleles and also tested for the presence of 2 TNF-α alleles in Cambodian patients with clinical TB and in control
individuals who did not have a history of TB.
The study subjects were unrelated Cambodian patients recruited from
a TB treatment program in eastern rural Cambodia. Two different groups of
patients and controls were recruited for the 2 stages of the study, the first
group in 1995 and the second group in 1996. The patients were randomly selected
from all inpatients or outpatients picking up their monthly supply of TB medicines
at Chantrea, Rumduol, or Kompong Rho District Hospitals in Svay Rieng Province.
The diagnosis of clinical TB was made on site on the basis of light
microscopy demonstrating the presence of acid-fast bacilli in sputum, pleural
fluid, or lymph node drainage. One of us (S.T.) conducted family interviews
of the household members of the patients and verified that no patients in
the study were related. Control individuals were recruited from patients visiting
the same hospitals for minor complaints. Based on detailed clinical history,
controls did not have a history of TB or current symptoms consistent with
TB. All patients and controls were followed up for 6 months to confirm their
diagnosis or control status.
After consent was obtained, blood was drawn from each subject and stored
at 4°C for 4 to 10 days. Plasma and peripheral blood mononuclear cells
(PBMCs) were prepared according to standard techniques and stored at −70°C.
Plasma samples were screened for evidence of antibodies to human immunodeficiency
virus (HIV) types 1 and 2 and human T-lymphotropic virus (HTLV) types 1 and
2 by enzyme-linked immunosorbent assay.
The DNA was prepared from PBMCs by a quick isolation method,18 and allele-specific polymerase chain reaction (PCR)
was performed. For HLA class I typing, PCR amplification of exons 2 and 3
of the HLA-A and HLA-B loci was performed. For class II typing, PCR amplification
of exon 2 of the HLA-DRB, DQB1, and DQA1 loci was performed.19- 21
The PCR products were separated by agarose gel electrophoresis, stained with
ethidium bromide, and photographed. Dot-blot, prehybridization, and hybridization
procedures were carried out according to the manufacturer's instructions (Lifecodes
Corp, Stamford, Conn). Class I and class II alleles were identified in the
PCR-amplified products by sequence-specific oligonucleotide probe hybridization.20,22- 25
The TNF-α alleles were determined by
restriction fragment length polymorphism (RFLP) analysis using primers designed
to incorporate a polymorphic site (the nucleotide A vs G) at position −308
nucleotides (nt) relative to the TNF-α transcription
start site. The TNF2 (A at −308 nt) polymorphism
creates an Nco I restriction site and can be differentiated
by size (107 nt for the TNF1 and 87 nt for the TNF2 allele) from the TNF1 allele
by agarose gel electrophoresis.26
The frequencies of independent HLA alleles and TNF-α alleles in patients and controls were determined by direct counting.
The statistical significance of the difference in frequency of individual
HLA and TNF-α alleles between the 2 groups
was calculated by the Fisher exact test with the aid of INSTAT software (GraphPad,
San Diego, Calif), and levels of significance were reported as P values along with 95% confidence intervals according to the program
used. In the first stage of the study, since each subject was tested for several
HLA alleles and 2 TNF-α alleles, and the same
data were used to compare the frequency of all detected alleles, significant
associations may have arisen by chance due to multiple comparisons. In the
second stage, we tested for the presence of a single allele identified in
the first stage, and thus, correction of the data for multiple comparisons
was not necessary.
The first stage of our study included 78 patients with TB (mean age,
47 years; range, 11-77 years; 24% male; 76 patients with pulmonary TB, 1 with
pleural TB, and 1 with scrofula) and 49 control individuals (mean age, 40
years; range, 15-68 years; 53% male). Of these 127 individuals, no patient
with TB had antibodies to HIV-1 or HIV-2, and 1 patient with TB was positive
for antibodies to HTLV-1.
Among the 156 alleles derived from 78 patients with TB, there were 7
DQB1*0503 alleles, whereas this allele was not found among the 98 alleles
from the 49 controls (P=.04) (Table 1). When HLA-DR15 and HLA-DR16 alleles were combined, there
was a slight increase in HLA-DR2 alleles (52 of 156) among patients with TB
compared with 32 of 98 alleles in controls (Table 1). HLA-B38 was present in 16 of 138 alleles among patients
with TB compared with 2 of 96 alleles in controls (P=.005)
(data not shown).
An RFLP analysis scoring for the 2 TNF-α
promoter variants revealed 11 TNF2 alleles of 156
alleles derived from patients with TB and 8 TNF2
alleles of 96 alleles derived from controls with no difference in TNF2 variant expression in either group. Seventeen of the 19 TNF2 alleles were found in HLA-DR3–positive subjects
(data not shown). Thus, as in other populations,26
the TNF2 allele was in linkage disequilibrium with
Based on the first stage of our analysis, we chose to further investigate
the association of the single class II allele, DQB1*0503, and TB. This second
stage included 48 patients with pulmonary TB (mean age, 46 years; range, 25-76
years; 40% male) and 39 controls (mean age, 40 years; range, 19-67 years;
30% male). One control subject tested positive for antibodies to HIV-1 and
was excluded from the HLA analysis. Among the 96 alleles derived from 48 patients
with TB, there were 9 DQB1*0503 alleles. No DQB1*0503 alleles were detected
among the 76 alleles derived from 38 controls (P=.005).
In Cambodia, which has a population of approximately 10 million, it
is estimated that up to 40000 new cases of TB and 13000 deaths due to TB occur
every year.27,28 Tuberculosis
is the major cause of morbidity and mortality in Cambodian men aged 18 to
40 years.27- 29
Although there are no reliable data, it is assumed that the majority of Cambodians
harbor M tuberculosis and thus are chronically infected
with TB,30 but that only a subset of these
individuals progress to active pulmonary or extrapulmonary TB.4
In the Cambodian population we studied, poverty (ie, an annual family income
of about $180 per year) and poor nutritional status, known cofactors of TB
progression, are pervasive and were equivalent between the patient and control
groups. In an ongoing pilot study in Svay Rieng, of 450 people randomly screened,
75% were purified protein derivative positive with a skin reaction of greater
than 5 mm of induration (S.T. and A.E.G., unpublished data). Although HIV-1
infection and AIDS are rapidly increasing in Cambodia,31
we ruled out HIV-1 as contributing to TB progression in this study group.
We did not detect a difference in the presence of the TNF2 allele in patients with TB vs controls consistent with the lack
of detectable effect of this polymorphism on TNF-α gene expression.32 Rather, the TNF2 allele appears to serve as a marker in the HLA region
for genetic associations with susceptibility to certain inflammatory and infectious
Previous studies using serologic testing methods reported an association
between progressive TB and the HLA-DR2 serotype in populations from India,
Indonesia, and Russia.11- 13
However, serologic methods can result in false assignment of the HLA class
II type in up to 25% of samples when compared with more sensitive molecular
DNA–based methods.33 Another study using
molecular typing failed to identify a specific allele that was associated
with disease progression although it supported the general association between
the HLA-DR2 serotype and TB progression in an Indian population.34
We found that when HLA-DR2 alleles were combined (ie, the HLA-DR15 and HLA-DR16
alleles), they were slightly increased in patients with TB vs in controls.
However, no specific HLA-DR2 alleles were increased significantly in the patient
group even before correction for multiple comparisons (Table 1). We found an association between clinical TB and the HLA-DQB1*0503
allele, and established the significance of this association by its confirmation
in a second study sample.
Based on the crystal structure of the class II molecules,35
it has been proposed that peptides bound by the HLA molecules form hydrogen
bonds with amino acid residues conserved in most class II alleles. The side
chains of the residues of antigenic peptides are accommodated in smaller cavities,
called pockets, in the binding site of the HLA molecules.36
These pockets appear to determine the peptide-binding specificity of the different
class II molecules.36 Thus, differentiation
of HLA alleles is crucial for the interpretation of HLA and disease associations
because point mutations in the class II genes are critical for peptide binding
and presentation and commonly occur in or near the peptide-binding pockets.35
Our evidence for an association between a specific allele, HLA-DQB1*0503,
and progressive clinical TB is particularly intriguing because the DQB1*0503
allele encodes for a change at amino acid position 57 of the β chain
(β57), which influences the charge in the putative peptide binding pocket,
P9, of the DQ molecule.36 The DQB1*0503 allele,
which is part of the DQ1 serologic specificity, encodes the negatively charged
aspartic acid at β57 in place of the more common, uncharged, and hydrophobic
amino acid valine. The MHC-restricted presentation of peptides by TB-infected
macrophages may be affected in patients who express this particular P9 pocket.
The negatively charged P9-binding pocket may bind TB antigens less effectively
or elicit a diminished immunogenic response.
In our combined samples of 126 patients with TB, 15 (12%) carried the
HLA-DQB1*0503 allele. Analysis of the protein sequences of all DQ molecules
reveals that only 2 other DQB1 alleles (DQB1*0603 and 0607), both also part
of the DQ1 serologic specificity, encode an identical P9 pocket. In the first
stage of our study, we found 2 additional patients with TB who carried the
HLA-DQB1*0603 allele, but we did not find either the DQB1*0603 or the 0607
allele among any of the controls. Thus, 13.5% of the patients with TB carried
the HLA-DQB1*0503 or DQB1*0603 alleles (1 patient was homozygous for HLA-DQB1*0503).
Our findings identify the HLA-DQB1*0503 allele as, to our knowledge,
the first gene associated with TB progression. These results provide a clue
to the complex process of mycobacterial antigen presentation and containment
by the host immune system and support the hypothesis that variability in the
human major histocompatibility complex confers relative susceptibility or
resistance to infectious disease.