Professional Documents
Culture Documents
www.elsevier.com/locat e/ijporl
R eceived 24 January 2006; received in revised form 8 April 2006; accepted 10 April 2006
KEYW ORDS
Otitis media;
Tonsillitis;
Comorbidity;
Upper respiratory
infection;
Genetics;
Twin study
Summary
Objective:To estimatethe relative contributionof genetic and environmental
effects to the association between recurrent otitis media and recurrent tonsillitis.
Methods:Self-reportquestionnairedata from a population-based
cohort of 9479
Norwegiantwins born from 1967to 1979.R ecurrentotitis media and recurrent
tonsillitiswere main outcomemeasures.Structuralequationmodellingwas used
to t alternativebiometric models to the twin data and to estimate
the relative
contribution of genetic and environmental effects to the association between otitis
media and tonsillitis.
Results:The lifetime prevalence was 11.7% (95% CI: 11.012.3) for recurrent tonsillitis and 11.2% (95% CI 10.511.9) for recurrent otitis media. Tetrachoric correlations were greater in monozygotic than in dizygotic twins in both males and females. A
model specifying additive genetic effects and individual environmental effects for
otitis media and tonsillitis and non-additive genetic effects for tonsillitis yielded the
best t. There was no evidence for sex differences in the genetic source or magnitude
of the genetic effects. There was a substantial overlap in genetic factors inuencing
variation in liability to otitis media and tonsillitis.
Conclusion:Common genetic factors contribute substantiallyto comorbidity
between recurrent otitis media and recurrent tonsillitis.
# 2006Elsevier Ireland Ltd. All rights reserved.
* Corresponding author at: Division of Epidemiolo gy, Norwegi an Institute of Public Health, Pb 4404 Nydalen, N - 0403 Oslo, Norway.
Tel.: +47 23408209; fax: +47 23408252.
E-mailaddresses:
ellen.kvest ad@fhi.no (E. Kvestad), k.j.kvarner@medisin.uio.no (K.J. Kvrner), espen.roysam b@psykologi.uio.no
(E. Rysam b), kristian.tambs@fhi.no (K. Tambs), jennifer.harris@fhi.no (J.R. Harris), per.magnus@ fhi.no (P. Magnus).
0165-5876
/$ see front matter
# 2006Elsevier Ireland Ltd. All rights reserved.
doi: 10.1016/j.ij porl.2006.0 4.004
1562
estima te the proport ion of genet ic and environm ental effects that are commo n and spe cic to the two
Otitis me dia is an infec tious disea se, resul ting pheno
from types.
the interpla y betw een micr obial load and imm une
respo nse. The high inciden ce and the high rate of
spont aneous
re coverysuggestthat otitis media 2. M aterials and methods
could be conside red part of the natura l mat uration
of the imm unologic al syste m in childre n [1] . 2.1. Sample
As earl yas 1957upperrespirat orydisea sewas
consid ered to play an impo rtant role in the aetiology
The Nor wegia n Institute of Publi c Health Twin Study
of middle ear disea se [2] sugge sting that the whole
is a dyna miccoho rtstudy. The cur rentdata base
respi ratory tract, includin g the middle ear, should
includesinfor mationon twins born in 1967197 9
be seen as a unit liable to the same phys iologic al
identi ed thro ugh the Medical Birth R egist ry of Norchan ges.The close relations hipbetw eenotitis
way. The R egio nal Ethics Comm ittee re viewed the
media an d tonsillitis has been demons trated in study.
more Two ques tionnaire
studieshave been conrecent epidemi ologica l studie s [35] . I n addi ducte
tion, ad,in 1992(q1) and 1998(q 2)
wit hindividu al
prosp ectivestudy of pre schoolchild ren[5] sug- respo nse rat es of 73% and 63% and pair w ise re sponse
gested that tonsillar infecti ons acco mpany comm
on of 64% and 52%, re spectiv ely. The q1 was se nt
rates
colds in somechildren, w hileothersseemto be
to all twins born during19671 974,
w ho w ere
at
pred isposedto subseq uentmiddl eear disease. least 18 yea rs old and for w hom a cur rent add ress in
Kvaerne r et al. re ported a mo derate but signica
nt
Norway
was obt ained. The q2 was sent to all twi ns
associa tion betw een recurren t ot itis med ia and
tonreceiving
q1, plus to ve new birth cohorts includin g
silliti s bot h in a cross-s ectional study of preschool
twins born during 19751 979. Th e combin ed q1 and
child ren [4] and a in a retrospect ive ques tionnaireq2 sample includes 9 479 twins who responded to at
based study of Nor wegian twins [3] . These analleast
yses,one of the ques tionnaires. 4 430 twins part iusing twin data from a sub sample of the pre sentcipate d in both studie s. Table 1 shows the num ber of
study populat ion,did not explo rethe exte ntto
pairs part icipat ing in the q1, q2 and the combi ned
which the ass ociationbet weenotitis media and
sampl e by sex and zygo sity. In q1 and q2 the re were
tonsill itiswas attrib utableto commo ngeneticor
724 and 1377 pairs from whom onl yone twin
enviro nment al facto rs. A substanti al genet ic predisrespo nded,re spectiv ely.P airswhere one twin
posit ion for otit is med ia [68] and tonsille ctomy
[9] nded to q1 and the co-tw in responded to q2
respo
has previou sly been reporte d. Herita bility estima
or tes
vice -versa are treate d a s pai r respo nders in the
of otitis me dia [10] and to nsillitis [11] based on combin
the
edsample. Th e twi n r e searchprog ram,
prese nt study popu lation has also pre viously been
includin gprocedu res
and zygositydete rminatio n,
report ed. Genetic effects , compr ising bot h addiistive
descri bed in det ail elsewhe re [12,1 3] .
and non- additive effects, expla ined 62% of the liability to recur rent tonsill itis and there were no sex
2.2. Measures
differe nces in the genet ic source or in the size of the
genet ic effects. R egardi ng ot itis med ia, the heritBoth ques tionn aires
includedthe foll owingitem s
ability compr isedonly addi tivegene ticeffects.
aboutear infec tionsand tonsill itis:Do you have
There w ere no sex differen ces in the genetic source
but a slightly g reater genet ic effect in males (0.72)
Table 1 Number of pair responders by sex and zygosity
compa red to fem ales (0.61) w as suggest ed. A mo
del
withou t these sex differen ces,w here genet ic
Group
Q1
Q2
Combined
a
effects explaine d 66% of the disea se variabi lity in
pairs(N)
P airs(N)
sample
pairs(N)
both males and femal es, gave almo st equiv alentMZM
t
416
526
677
to the data. The utilisa tion of twin studi es to esti-MZF
528
777
904
mate the genet ic and enviro nmental causes of covDZM
387
397
592
ariatio ncontribu testo bette runde rstanding
of
DZF
443
655
789
upper respirator y comorbid ity.
DZU
796
979
1285
total
2570
3334
4247
The pre sentstudy address esthe ques tionof
shared aetiolo gy in otitis media and tonsillitis , using
Q1, 1992 questionnaire; Q2, 1998 questionnaire; MZM, monozygotic
a bivar iate twin des ign. Th e aim of our study was
to males; MZF,monozygoticfemales;DZM, dizygoti c
DZF, dizygoti c females; DZU, dizygotic unlike sex.
invest igate w hether shared genet ic or environmmales;
ena
The combined sample includes twins who have responded
tal factorscontr ibuteto the associa tion
betw een
to either questionnaire 1, questionnaire 2 o r both.
otitis me diaand tonsillitis. Furthe rmore,we will
1. Introduction
R ecurr ent otitis me dia and to nsillitis: comm on disea se pre dispos ition
1563
Fig. 1 P ath model depicting the contribution of genetic and environmental factors to twin resemblance. P denotes the
phenotype. A, D, C and E are latent variables for effects of genes acting additively, genetic dominance, environmental
factors shared by cotwins and individually acting environmental factors, respectively. Lowercase letters denotes the
corresponding path coefcients for the contribution of the latent variables on the phenotype. The genetic additive
cotwin correlation is unity for monozygotic (MZ) twins and 0.5 for dizygotic (DZ) twins. The cotwin correlation between
genetic dominance is unity for monozygotic twins and 0.25 for dizygotic twins.
1564
(c) and spec ice)( enviro nment. Common envi roninstanc e, th ereare no effec tsof genet icdom imen t ref ers to environm ental factors shared bynanc
the e or commo n envi ronm ent )
bemay
statist itwins , thu s contributi ngto twin resemb lance cally compared with the observed data. A num ber of
regardle ss
of zygosity.Earl yinfec tiousexpos ures, mod els can be tte d t o the data, and the t of the
2
num ber of siblings, type of chi ld-care arr angeme
mod
nt el is assessed by a goodne ss-of-t
x -test. The
and breast -feeding pattern are norm ally examplgoal
es in mo del tting is to explain the observed data
of shared enviro nmental effec ts in childhoo d. The
as well as possib le.
non-sh ared enviro nmental effects
e) are( exposur es
The nal step in twi nanalysesis to esti mate,
that are not shared by the members of a pair, that
based on the best-t ting mo del, th e proport ion of
cause w ithin-pai r diff erences for a trait and includes
varianc e in liabil ity to otitis me dia and to nsillitis,
2
meas ureme nt error.
respec tively,
due to add itivegenet icactio n(a ),
2
Simple correlati oncompa risonscan indicate non-add itive g enetic effects
d ), (comm on envi ron2
whet her genet ic in uence is import ant for a trait.
men tal effectsc (), and ind ividual- specic envi ron2
Co-twi n simila rity for otitis me dia and to nsillitismen
can te( ). The propo rtion of varianc e in liabilit y due
2
be measur ed by estima ting tetrachor ic correla to
tions.
the sum of additi ve genet ic effects
a ) and
( non2
The tetrac horic co-twin correla tion rep resents the
addi tive effec ts
d )( is termed heri tability. Furthe r
correla tionbet weeneac h oft h e co-tw in
unde
s r- detail sof the applica tionof biome tricalgenet ic
lying liabilit y t o disease. This model fur ther assu
mod
mes
els are outli ned elsewhe re [16] .
that otitis mediaor tons illitis,respecti vely, has
a
Equali ty of prevale nces for ot itis me dia and tonmultifa ctorialaetiology involvin ga number of
silliti s, re spectiv ely,
w ithintwin pairs, betwe en
genet icand envi ronm ental
risk factors of low to
gender gro ups, between zygo sitygro upsand
mod erate effects. Und er these circums tances, the
betwe en compl ete and incomp lete pai rs, was tested
distrib utionof the liabi lityof the diseasein the
before we started the mod el-ttin g procedu re. Sevgenera l populat ion may be assu med to be app eral
roxi-mod els were t to es timate the si gnican ce of
mate ly normal [14] . Tetracho ric correla tions were
speci c para meters . The obs erved patte rn of mo noestima ted in ve gro ups, dened by sex and zygo
zygotic
sity and diz ygotic twin correla tions indicates the
(i.e. monoz ygoticmales, monozy goticfemale s, degre e t o which the diff erent factors (A, C, D and E)
dizygot icmale s,dizygoticfem alesand dizygot ic are present. A patt ern of diz ygotic twi n correla tions
opposi te sex pai rs) for: (1) twi n-cotwin re semblance
highe r than hal f the monozy gotic twi n correla tions
for otit is me dia and tonsillitis, (2) the pheno typic
indicate sthat commo nenvi ronm ent
play sa role,
(withi n per son) relations hip betw een tonsill itiswhile
and dizygotic twin corr elati ons less than half the
otitis media , and (3) the cross-t win-cro ss-trait simimonoz ygotic
twin correla tionssugge sts
that the
larity (otits media in twin 1 in a pair and to nsillitis
genet
in iceffe ct may be partly non-addi tive[14] .
twin 2 in the same pai r).
Becau se
c andd cann ot be esti mated si multaneo usly
in the struct ural mo dels, either AC E or AD E is chosen
2.3.2. Model tting using Mx
as the full model based on the obs erved correla tion
patte rn.In the mo deltt ingproce duresthe full
The path mo del represe nts the hypothesi zed causal
conne ctionsw ithina systemand is a series of
mod el is compare d with neste d su b-models. A mo del
regress ion equa tions that describ e the relationswith
hipsonly non- additive g enetic effects and no addibetwe enthe tetra choriccorrela tionsand the
tive genet iceffect is not realistic becau seeven
hypothe sized
latent variable s( genotype s
and
under full dom inance most of the genet ic variance
enviro nments ) [16] . Fig. 1 represen ts the genet
will usuall
ic
y be addi tive [17] . Rat her than acce pting
path mo delfor twi nstudies,where the obs erved a model w ithoutaddi tivevariance ,d w asconcorrela tions can be descri bed as the sums of paths
strain ed not to exceed
a in the models test ed.
betwe en the phen otypes of twin1 and twin2. Iden-By meansof bivariateanalysis,the cova riance
tical twins are perfect ly corr elated for effects due
betwe en the phen otypes can to be partition ed into
to additi veand non -additive
genet iceffects ,and
genet ic and envi ronm ental compo nents. Moreover,
frater naltwi nsare correla ted0.5 for addi tive sex-spe cic effects may be invest igated w hen the
genet iceffects and 0.25 for non- additive
genet ic data incl udes infor mation from same- sex and oppoeffects in our models . Th ese effects (A, C, D andsite-sex
E)
twins [14] . Models of g eneral sex-lim itation
are used when theoret ical models are built.
specify that bot h the sourc e and the magnitu de of
To estima tethe twi ncorrelati onsand perform genet ic and envi ronm ental
effe cts may vary
the mo del-ttin procedurea
g
compute r prog ram, betwe enmalesand female s.The mo rerestr icted
such as Mx [16] , is nee ded. Mo del-tti ng approach
mod es
elof commo nsex-l imitatio only
n
allows the
involve con structing
a mo delthat descri besthe
magnitu de of the se effec ts to vary bet ween males
obser veddata. The model s pre dictions(for
and female s.
R ecurr ent otitis me dia and to nsillitis: comm on disea se pre dispos ition
1565
3. Results
T able 2 P henotypic correlation, twin co-twin and crosstwin crosstrait correlations in a study of 4247 twin pairs,
according to zygosity and sex
Correlations
MZM
MZF
DZM
DZF
DZU
0.71
(0.62,0.79)
0.71
(0.61,0.79)
0.64
(0.57,0.71)
0.61
(0.52,0.67)
0.35
(0.20,0.48)
0.10
( 0.07,0.27)
0.25
(0.13,0.36)
0.14
(0.03,0.25)
0.33
(0.22,0.40)
0.24
(0.14,0.32)
Crosstwin cross
trait correlations
R ecurrent tonsillitis in twin1 0.20
versus otitis media in twin 2 ( 0.04,0.34)
0.25
(0.15,0.35)
0.12
( 0.05,0.28)
0.18
(0.06,0.29)
0.14
(0.04,0.24)
P henotypic correlations
R ecurrent otitis media
versus tonsillitis
0.43
(0.34,0.52)
0.33
(0.19,0.46)
0.37
(0.27,0.46)
0.44
(0.36,0.52)
Twinco-twin correlations
R ecurrent otitis media
R ecurrent tonsillitis
0.42
(0.29,0.53)
MZM, monozygotic males; MZF, monozygoti c females; DZM, dizygoti c males; DZF, dizygotic females; DZU, dizygotic unlike sex.
1566
Table 3 Bivariate model tting statistics. All models are compared to the common sex-limitation ADE-model
Sex-limitation
Model
Common
ADE
ADtons E
AE
ADE
ADtons E
AE
No
df
Dx
37903
37906
37908
37909
37911
37912
2.109
228.014
6.949
9.028
454.435
Ddf
AIC
3
5
6
8
9
.550
<.001
.326
.340
<.001
3.891
218.014
5.051
6.972
436.435
Fig. 2 P ath model depicting the contribution of genetic and environmental factors to the association between otitis
media and tonsillitis within one twin in the best tting model. A, D and E are latent variables for effects of genetic
additivity and dominance and non-shared environmental factors, respectively. Lowercase letters denotes the corresponding path coefcients for the contribution of the latent variables on the phenotype.
ra and re are correlations
reecting the degree to which genetic dominance, additive genetic effects and unique environmental effects inuencing
otitis media correlate with the corresponding factors inuencing tonsillitis.
R ecurr ent otitis me dia and to nsillitis: comm on disea se pre dispos ition
1567
1568