Koch A, Melbye M, Sørensen P, Homøe P, Madsen HO, Mølbak K, Hansen CH, Andersen LH, Hahn GW, Garred P. Acute Respiratory Tract Infections and Mannose-Binding Lectin Insufficiency During Early Childhood. JAMA. 2001;285(10):1316-1321. doi:10.1001/jama.285.10.1316
Author Affiliations: Department of Epidemiology Research, Statens Serum Institut (Drs Koch, Melbye, Sørensen, and Mølbak and Messrs Hansen and Andersen, and Ms Hahn); Departments of Otolaryngology, Head and Neck Surgery (Dr Homøe) and Clinical Immunology, Rigshospitalet, National University Hospital (Drs Madsen and Garred), Copenhagen, Denmark.
Context Hospital-based studies have found that increased susceptibility to certain
infections is associated with low serum levels of mannose-binding lectin (MBL)
due to MBL variant alleles. However, the contribution of MBL insufficiency
to incidence of common childhood infections at a population level is unknown.
Objective To investigate the effect of MBL insufficiency on risk for acute respiratory
tract infection (ARI) in unselected children younger than 2 years.
Design and Setting Population-based, prospective, cohort study conducted in Sisimiut, Greenland.
Participants Two hundred fifty-two children younger than 2 years who were followed
up weekly between August 1996 and August 1998 for morbidity surveillance.
Main Outcome Measure Risk of ARI, based on medical history and clinical examination, compared
by MBL genotype, determined from blood samples based on presence of structural
and promoter alleles.
Results A 2.08-fold (95% confidence interval [CI], 1.41-3.06) increased relative
risk (RR) of ARI was found in MBL-insufficient children (n = 13) compared
with MBL-sufficient children (n = 239; P<.001).
The risk association was largely restricted to children aged 6 to 17 months
(RR, 2.92; 95% CI, 1.78-4.79) while less effect (RR, 1.47; 95% CI, 0.45-4.82)
and no effect (RR, 1.00; 95% CI, 0.42-2.37) was shown among children aged
0 to 5 months and 18 to 23 months, respectively.
Conclusion These data suggest that genetic factors such as MBL insufficiency play
an important role in host defense, particularly during the vulnerable period
of childhood from age 6 through 17 months, when the adaptive immune system
Acute respiratory tract infections (ARIs) are among the most prevalent
infections in childhood worldwide, with the highest incidence among children
younger than 2 years.1,2 Risk
factors for these infections include being male, living in crowded conditions,
and being exposed to child care centers and passive smoking.1,3
However, these factors explain only a fraction of the variation in incidence
between children. Mannose-binding lectin (MBL) is a serum protein believed
to play an important role in the innate immune response.4
A single gene, MBL2, located at chromosome 10, codes
for human MBL.5,6 Mannose-binding
lectin may exert its action through binding to high mannose and N-acetylglucosamine
oligosaccharides present on a variety of microorganisms, thereby activating
the complement system by MBL-associated serine proteases and interacting with
novel receptors on phagocytes.7- 9
As part of the innate immune system, the protein is considered particularly
important in the vulnerable period of infancy before adequate specific immune
protection is established by the adaptive immune system.10
Three variant alleles have been described in exon 1 of the MBL2 gene.11- 13
These variants are due to 3 single-base pair substitutions at codon 54 (allele B), codon 57 (allele C), and codon
52 (allele D), which independently cause low serum
MBL levels.11 The normal wild type allele is
commonly designated A, and the 3 mutant alleles are
designated O. All variant alleles reduce the amount
of functional MBL subunits in heterozygous individuals 5- to 10-fold.14 The variant alleles affecting the structural part
of the MBL gene are relatively common in many ethnic
groups including Eskimos (frequency of sum of all variant alleles: 0.12) and
whites (frequency of sum of all variant alleles: 0.20).14
The serum MBL concentration is further dependent on a number of nucleotide
substitutions in the promoter region of the MBL2
gene.15,16 In particular, a polymorphism
in codon −221 (X/Y type) has a significant
effect on the MBL serum concentration with the Y
promoter having high and the X having low MBL-expressing
Hospital-based studies have shown that an increased susceptibility to
bacterial and viral infections has been associated with low serum levels of
MBL and MBL variant alleles17- 21
and with a worsened prognosis in such chronic diseases as cystic fibrosis,
rheumatoid arthritis, and systemic lupus erythematosus.22- 24
However, MBL has never been shown to play a role in the incidence of infectious
diseases on a population level.
We performed a population-based cohort study of ARI in children aged
0 to 2 years living in Sisimiut, West Greenland, to investigate whether MBL
insufficiency contributes to the general morbidity in children.
Sisimiut is the second largest town in Greenland, with approximately
5300 inhabitants, of which 88% have been born in Greenland and 12% have been
born outside of Greenland, primarily in Denmark. One health center serves
Sisimiut with up to 5 physicians and a midwife. All health services, including
medication, in Greenland are free. In this town, an open cohort of children
aged 0 to 2 years was formed by April 1, 1996, and monitored for ARI on a
regular basis from August 12, 1996, to August 6, 1998. The cohort comprised
all children living in the town as of April 1, 1996, and included all children
who were either born in Sisimiut or who had moved there before June 1, 1998.
All children irrespective of ethnicity (whether Eskimo, mixed race, or white)
were invited to participate. Specially trained Danish medical students and
local interpreters, supervised by Danish physicians, followed up this cohort.
Children were enrolled as soon as possible after identification, but
not prior to 6 weeks after birth. They were excluded from the cohort when
turning age 2 years. At enrollment written informed consent was obtained from
each child's parent or guardian and a structured background interview was
conducted. In the monitoring period, the children were visited weekly, at
which time parents were asked about their child's ARI symptoms since the last
visit. If symptoms were reported, a clinical examination was performed with
focus on the respiratory system, including otoscopy and tympanometry using
portable tympanometers (MicroTymph2; Welch Allyn, Skaneateles Falls, NY).
If any of the following physical signs were recorded, the episode was characterized
as a respiratory tract infection and defined by the symptoms reported (modified
from parameters set by Selwyn2): purulent nasal
discharge, cough, red and bulging tympanic membrane with loss of normal landmarks
and abnormal tympanometric compliance, purulent ear discharge, pharyngotonsillar
erythema or exudate, respiratory rate higher than 50/min plus cough or difficult
breathing, rales, stridor, wheezing, cyanosis, or subcostal chest indrawing.
When clear nasal secretion was the only symptom, the episode was not characterized
as an ARI.2 If the children had attended the
health center during an episode with respiratory symptoms, the diagnoses made
by the local physicians were also used to characterize an episode. These diagnoses
included rhinitis, pharyngitis, tonsillitis, acute otitis media, croup, bronchitis,
bronchiolitis, and pneumonia. A child had to be free of symptoms for 7 days
before a new episode was counted. A child was considered at risk on days he
or she was symptom free, including the first day of a new episode, but not
during the 7 days following an episode.
The study was approved by the Scientific Ethics Committee for Greenland.
Venous blood samples were drawn at the end of the study period. Genomic
DNA was isolated from blood cells drawn in EDTA containers and stored at −
80°C in Greenland, transported by − 30°C to Denmark, and stored
again at − 80°C until testing. The MBL
alleles were detected as previously described.14,15
Because all 3 variant alleles have a considerable effect on MBL concentrations,
they were grouped as allele O, while the normal allele
was designated A. Thus, the structural genotypes A/A, A/O and O/O
were obtained. On the basis of these genotypes and the effect of the promoter
variants in position − 221 on the MBL serum concentration, we were able
to define 6 MBL genotypes.22 The A/A group had 2 normal structural alleles
with high expression promoter activity (YA/YA), 1 high and 1 low-expression promoter (YA/XA), or 2 low-expression promoters (XA/XA). The A/O group had 1 variant and 1 normal structural
allele with either high-expression promoter (YA/O) or low-expression promoter (XA/O) activity. Because both the O/O group and the XA/O group have a virtually undetectable amount of MBL in the blood,22 we combined genotypes and defined an MBL-sufficient
group as A/A + YA (YA/YA, YA/XA, XA/XA, and YA/O), and an MBL-insufficient group as XA/O + O/O.
For risk factor analysis, ratios of incidence and 95% confidence intervals
(CIs) were used as measures of relative risk (RR). The number of episodes
and the number of days at risk for each child were calculated on a monthly
basis. Because each child could have respiratory events in different calendar
months, the model had to account for the possible correlation between episodes
from the same child. Therefore, a generalized estimating equation (GEE) method
with a correlation structure between measurements from the same child was
used in a model assuming a Poisson distribution for the number of episodes.
A banded Toeplitz correlation structure with 6 bands was applied. With this
correlation structure, episodes from the same child with a maximal interval
of 7 months were considered correlated, while those with larger intervals
were regarded as independent. Relative risk estimates were obtained from the
GEE model, and CIs were calculated using a robust covariance estimator for
the estimated effects. We used the Wald test to assess the effect of any risk
factor, assuming that estimates based on the GEE method were asymptotically
normally distributed.25 We have previously
found that age was a strong risk factor for ARI (data not shown), and to allow
dependence on age, sex, ethnicity, and calendar period, we adjusted for these
factors in all risk-factor analyses. The GENMOD procedure in SAS version 6.12
was used for the GEE model.26
The population-attributable risk percentage, which is an estimate of
the fraction of the total number of ARIs that would not have happened if the
effect of a specific risk factor had been eliminated, was estimated as described
by Bruzzi et al27 on the basis of adjusted
RRs and the distribution of exposure in the episodes.
Of the 338 children eligible for participation, 294 agreed and 44 refused,
for a participation rate of 87%. Acute respiratory tract infection was not
stated by any of the families as a reason for not participating in the study.
None of the participating children were recognized as having severe chronic
diseases. Of the 294 enrolled children, 288 were at risk for ARI (ie, having
no symptoms for 7 days before having an episode of ARI). Of these, blood could
be sampled and MBL genotypes determined for 252 children. These included 204
Greenlandic Eskimo children, 27 children of mixed race, and 8 white children
according to parents' place of birth (13 children were of unknown descent).
The ethnic distribution of the 252 children with known MBL genotypes did not
differ from that of children with unknown MBL genotypes (Fisher exact test, P = .29). One hundred ninety-one children (75.8%) were
homozygous for the normal allele A (genotype A/A), 6 children (2.4%) were homozygous
for variant alleles (genotype O/O), and 55 (21.8%) were heterozygous (genotype A/O) (Table
1). Prevalence of the normal genotype A/A was higher in Eskimo children than in children of mixed
descent, both of which were higher than in white children, which did not reach
statistical significance (Fisher exact test, P =
.32). All variant alleles were observed in Eskimo children with B as the predominant allele (frequency: 0.08), followed by D and C at lower frequencies (0.04 and 0.01,
respectively). When combining the structural alleles with promoter alleles,
239 children (94.8%) were MBL-sufficient (A/A + YA), while 13 children (5.2%)
were MBL-insufficient (XA + O/O). There was a significant association between ethnicity
and MBL sufficiency or insufficiency (Fisher exact test, P = .03) with Eskimo children having the lowest frequency of MBL insufficiency.
The risk factor analysis showed an increased risk for ARI in children
who were heterozygous or homozygous for variant alleles, with those homozygous
for variant alleles having the highest risk (Table 2). When combining those heterozygous or homozygous for variant
alleles, the risk was still significant compared with children homozygous
for normal structural alleles (A/A). An even more pronounced effect was seen among MBL-insufficient
children compared with MBL-sufficient children. There was no difference in
the risk for ARI among MBL-sufficient children, while both groups of MBL-insufficient
children had significantly higher RRs of ARI than the former. Compared with
all MBL-sufficient children, we found MBL-insufficient children to have a
2.08-fold increased risk for ARI (P<.001). Table 3 presents the full multivariate
model for the analysis of MBL-sufficient children vs MBL-insufficient children.
Table 4 presents an analysis
of interaction between age and MBL genotypes. When only looking at the structural
alleles, the most pronounced effects of MBL were observed for the age groups
6 through 17 months. A clear age-dependent effect of MBL was seen, however,
when comparing the groups defined as being MBL-sufficient and MBL-insufficient
(XA/O + O/O). Thus, the highest risk was seen for MBL-insufficient children aged
6 through 17 months (RR, 2.92; 95% CI, 1.78-4.79). Younger children also had
an increased, although not statistically significant, RR of 1.47, while there
was no effect of MBL insufficiency for older children (RR, 1.00).
There was a population-attributable risk for ARI of 7.2% associated
with the presence of variant alleles (A/O and O/O vs A/A), and of 2.5% associated with
To our knowledge, our study is the first prospective population-based
study of the effect of MBL on the risk for common childhood infections. Overall,
MBL insufficiency (genotypes XA/O + O/O) was associated with a significant 2-fold increased RR of ARI.
Detailed analyses showed, however, that the effect of MBL insufficiency was
restricted to the period 6 through 17 months of age. This age group had nearly
a nearly 3-fold increased risk for ARI. These findings are particularly interesting
because 1 of the 2 theories of the importance of MBL in the immune response
envisages the protein to play a major role in the immune defense in the vulnerable
period between 6 and 18 months of age when children are depleted of passively
acquired maternal antibodies and the adaptive immune system is still immature.28 However, this has not previously been demonstrated.
The other hypothesis proposes a role for MBL at the time of primary contact
with any pathogen at any age and before an IgM antibody response can be mounted—the
so-called anteantibody hypothesis.29
Although not mutually exclusive, overall, our findings lend support to the
first hypothesis, but we cannot exclude that the latter hypothesis is valid
in certain situations.
It has been shown that hospitalized children and immunocompromised patients
with infectious diseases including ARI and meningococcal disease have a higher
frequency of MBL mutations than controls hospitalized for other reasons.17- 19,21 In
contrast to this and to our findings, studies of children with recurrent otitis
media have failed to show an association with MBL mutations.30,31
There may be a number of reasons for this discrepancy. Apart from methodological
differences in prospective vs retrospective studies and in the composition
of study populations, a major difference is the age group involved. We found
the maximal effect of MBL insufficiency in those aged 6 through 17 months,
while no effect could be demonstrated among children aged 18 through 23 months.
It is difficult to extrapolate from our results to children older than 2 years,
but our findings suggest that the effect of MBL insufficiency as an independent
risk factor for ARI in these children is limited. By contrast, in older children
and in adults, the main effect of MBL insufficiency may rather be as a disease-modifying
locus in an already established disease or in situations with a concurrent
immunodeficiency. This is indicated by studies showing that the clinical course
and severity of diseases such as cystic fibrosis, systemic lupus erythematosus,
and rheumatoid arthritis may be worse in patients carrying variant MBL alleles.22- 24
Another difference may be that we determined MBL insufficiency on the basis
of both MBL structural alleles and promoter alleles
while others considered only the structural alleles. We have previously shown
that the former measure may in fact reflect the serum level of MBL better
than the structural alleles by themselves, because heterozygous children (A/O) may be both MBL-sufficient
and insufficient, depending on the promoter alleles.22
Correspondingly, we found a significantly increased risk for ARI in children
heterozygous and homozygous for variant alleles (A/O, O/O;
RR, 1.46), but it was not as high as when promoter alleles were considered
Finally, unlike previous investigations, we addressed ARI as a whole.
The spectrum of infections was wide from rhinitis to pneumonia and bronchiolitis.
The effect of MBL mutations may differ when observing different clinical manifestations
or etiological agents. Because our study population was of limited size and
had a low frequency of MBL variant alleles compared
with populations in other studies, we were not able to further explore specific
clinical manifestations or the role of microbiological agents. Different ARIs
are, however, often associated. Thus, rhinitis may precede other infections
such as otitis media and pneumonia. A large part of ARI is viral in origin,
while acute otitis media is more often a bacterial infection. Thus, even if
MBL insufficiency is found associated with bacterial infections, it may primarily
be associated with decreased protection against viral infections, which in
turn predispose to bacterial infections. Indeed, some of the initial observations
on the role of MBL in the innate immune system were made on such viruses as
the influenza virus.32
The influence of MBL insufficiency on a population basis should theoretically
be more pronounced in other populations with higher incidences since we found
relatively few children to be MBL-insufficient. We have previously reported
that frequencies of heterozygous and homozygous variant alleles are found
in 38% of white and in 48% of black populations compared with 22% and 20%
as found earlier among Greenlandic Eskimos.13,31
The latter figures are comparable to the frequency of 24% found in the present
population, where 81% were ethnic Eskimos. Accordingly, the population-attributable
risks for MBL-structural alleles and promoter alleles in our population are
probably lower than in other populations.
The high frequency of MBL variant alleles in different populations indicates
that MBL polymorphisms represent a balanced genetic system favoring variant
alleles arising from general selection.14 We
have previously suggested that this might be due to the putative disadvantage
MBL confers toward intracellular bacteria and parasites, which use the complement
system to gain access to the intracellular compartment.33
Whether this genetic mechanism plays a role in current populations is unknown.
In conclusion, on a population level, we found that MBL insufficiency
is significantly associated with increased risk for ARI among those aged 6
through 17 months, illustrating the importance of innate immune defense factors
prior to the maturation of the adaptive immune system.