close

Вход

Забыли?

вход по аккаунту

?

Complex segregation analysis of Parkinson's disease The Mayo Clinic Family Study.

код для вставкиСкачать
Complex Segregation Analysis of Parkinson’s
Disease: The Mayo Clinic Family Study
Shannon K. McDonnell, MS,1 Daniel J. Schaid, PhD,1 Alexis Elbaz, MD, PhD,2,3 Kari J. Strain, BS,1
James H. Bower, MD,4 J. Eric Ahlskog, PhD, MD,4 Demetrius M. Maraganore, MD,4
and Walter A. Rocca, MD, MPH2,4
Objective: To conduct complex segregation analyses of Parkinson’s disease (PD).
Methods: Data on the familial aggregation of PD remain conflicting. We conducted a historical cohort study of 1,234 relatives
of 162 patients with PD representative of people of Olmsted County, MN, and of 3,009 relatives of 411 patients with PD
referred to the Mayo Clinic. Relatives were interviewed and screened for parkinsonism either directly or through a proxy, those
who screened positive were examined, or a copy of their medical record was obtained to confirm the diagnosis. For subjects who
resided in Olmsted County, additional information was obtained from the archives of the Olmsted County Historical Society
and from a records-linkage system.
Results: Thirty-two relatives of population-based probands and 69 relatives of referral patients developed PD (101 in total).
Combining population-based and referral samples, the model that best explained the familial clustering of PD overall was a
major gene with additive effect on the penetrance. This model predicted an average decrease in age at onset of PD of approximately 18 years for each copy of the putative high-risk allele. The best fitting model for younger onset PD (age ⱕ 59 years) was
an autosomal recessive model. The best fitting models for older onset PD (age ⬎ 59 years) were a recessive or an additive model.
Interpretation: The familial aggregation of PD may be explained in part by a major gene with additive effect on the penetrance.
Ann Neurol 2006;59:788 –795
Existing data on the role of genetics in Parkinson’s disease (PD) remain conflicting.1 Numerous studies have
investigated the familial aggregation of PD in the past
10 years, and the majority reported a higher frequency
of PD among relatives of cases compared with relatives
of control subjects; however, the estimate of relative
risk varied considerably among studies.2–15 In all of
these studies except one,13 cases and control subjects
were interviewed to obtain information on PD among
their relatives (family history method), and there is
some evidence that this approach may overestimate the
relative risk.16 In addition, two twin studies using clinical phenotypes showed a limited role of genetic factors.17,18 By contrast, a twin study of the dopaminergic
function measured by positron emission tomography
showed a stronger genetic component.19
Few studies have examined whether the familial clustering of PD was consistent with Mendelian inheritance using segregation analyses.5,20 –24 The Mayo
Clinic Family Study of Parkinson’s Disease, using a
combination of methods to directly study the relatives
of cases and control subjects (family study method),
showed a modestly increased risk in relatives of cases
compared with relatives of control subjects.25,26 In this
article, we report the results of our complex segregation
analyses.
From the Divisions of 1Biostatistics and 2Epidemiology, Department of Health Sciences Research, Mayo Clinic College of Medicine, Rochester, MN; 3Institut National de la Sante et de la Recherche Médicale Unit 708, Hôpital de la Salpêtrière, Paris, France; and
4
Department of Neurology, Mayo Clinic College of Medicine,
Rochester, MN.
A.E. participated in this study while on leave from the Institut National de la Sante et de la Recherche Médicale Unit 360.
Received Aug 23, 2005, and in revised form Dec 21. Accepted for
publication Feb 26, 2006.
788
Materials and Methods
Study Design and Recruitment of Probands with
Parkinson’s Disease
This study included two cohorts of subjects: first-degree relatives of patients with PD derived from a well-defined population (population-based incident cases from Olmsted
County, MN) and first-degree relatives of patients with PD
referred to the Mayo Clinic (referral cases). Members of both
cohorts were followed historically through interviews, direct
examinations, review of medical records, and other special
sources of information, as described later.
The medical records-linkage system of the Rochester Epidemiology Project was used to identify all subjects residing
in Olmsted County, MN, who developed PD from 1976
through 1995. Details about the study population, the iden-
Published online April 24, 2006 in Wiley InterScience
(www.interscience.wiley.com). DOI: 10.1002/ana.20844
Address correspondence to Dr Rocca, Division of Epidemiology,
Department of Health Sciences Research, Mayo Clinic, 200 First
Street SW, Rochester, MN 55905. E-mail: rocca@mayo.edu
© 2006 American Neurological Association
Published by Wiley-Liss, Inc., through Wiley Subscription Services
tification of incident cases, and our diagnostic criteria for PD
were reported elsewhere.27–29
In addition, we recruited a series of sequential new cases
of PD referred to the Department of Neurology at the Mayo
Clinic in Rochester, MN, from July 1996 through October
2000. All cases were residents of Minnesota or one of the
surrounding four states (Wisconsin, Iowa, South Dakota,
and North Dakota) and were examined by a movement disorder specialist (D.M.M., J.H.B., or J.E.A.) following a standardized clinical assessment form.26 The diagnostic criteria
for PD were the same as those reported previously.27 For the
analyses, we distinguished a short- (120-mile radius around
Rochester, MN) and a medium-distance referral region (fivestate region). Patients referred to the Mayo Clinic from beyond the five states were excluded.26
Identification of First-Degree Relatives and
Construction of Pedigrees
Details about the identification and recruitment of firstdegree relatives were reported elsewhere.25,26 In brief, patients were administered a questionnaire including a detailed
family composition section either directly or through a
proxy. For some of the families of subjects who resided in
Olmsted County and who had no available informant, the
pedigree composition was obtained by linking existing obituary information and medical record information.25,26 We
considered ineligible for this study relatives who were
younger than 40 at the time of the study (because of low risk
for parkinsonism), half siblings, stepparents or step siblings,
and adopted relatives (because there is no blood relation).
Ascertainment of Parkinson’s Disease among Relatives
A detailed description of the methodology used to ascertain
incident cases of PD among relatives is given elsewhere.25,26
In brief, the primary methodology of the family study involved a two-phase design with individual screening of the
relatives for parkinsonism and examination of those who
screened positive. First-degree relatives were administered a
screening instrument for parkinsonism through a direct telephone interview or through a proxy interview.30 Relatives
screened positive if they had a previous diagnosis of PD, a
previous treatment with L-dopa, at least two of nine symptoms of parkinsonism, or “shaking” alone.30 Living relatives
who screened positive were invited for an examination by a
movement disorders specialist either at the Mayo Clinic or at
their home. For relatives who screened positive but were deceased (screened through a proxy) or who could not be examined, we obtained copies of their medical records. In addition, all relatives residing within 120 miles of Rochester,
MN, and aged 60 years or older were invited for examination (at Mayo or at home) regardless of the screening.26
Documentation of parkinsonism among first-degree relatives
of subjects who resided in Olmsted County was enhanced by
the use of the records-linkage system serving the population
of Olmsted County.26,31,32
For those relatives who were examined by one of the neurologists of this study or for whom a pertinent medical
record was obtained, PD and parkinsonism were defined using the criteria reported previously.27 For subjects who
screened positive but could not be examined directly and had
no pertinent medical record, parkinsonism and PD were defined as a previous diagnosis reported by the subject or by
the proxy at interview. For relatives with multiple sources of
information, the final PD status was based on the best available information.26
Statistical Analysis
First-degree relatives were considered in the analyses from
birth through one of the following events, whichever occurred first: contact for the study (or time of last medical
record information), death, or onset of parkinsonism (subjects who developed one type of parkinsonism were not considered at risk for development of another type). The age at
onset of PD was unknown for 2 relatives from the
population-based sample and for 12 relatives from referral
cases. For these 14 relatives, we substituted age at the time of
study for age at onset of PD.
Complex segregation analyses were performed using genetic regressive models,33 as implemented in the SAGE computer package.34 The SAGE REGTL module was used to
analyze age at onset, allowing for censored observations (relatives who did not have PD at their current age or at the
time of death). Two general factors, labeled A and B, were
assumed to determine three “types” of individuals, labeled
AA, AB, and BB. Under a genetic model, these three types
represent the genotypes for two alleles, A and B. For nongenetic models, these three types are arbitrary, but they allow
for population heterogeneity that is not necessarily due to
genetic differences.
Age at onset, denoted as x, was assumed to be distributed
according to the probability-density function, P(x) ⫽ [␥␣
e(␤⫹␣ x)]/[1 ⫹ e(␤⫹␣ x)]2 and the cumulative distribution
function, F(x) ⫽ ␥[e(␤⫹␣x)]/[1⫹e(␤⫹␣x)]. According to this
model, the parameter ␣ determines the rate of change in the
probability of disease by age (it determines the variability in
the age at onset). We assumed that ␣ was the same for all
three types of individuals. The parameter ␤ influences the
mean age at onset. The logistic age at onset distribution is
symmetric, similar to the normal distribution, with mean
⫺␤/␣ and variance ␲2/3␣2. The parameter ␥ is the “susceptibility” parameter (the cumulative probability that a person
will have PD if he or she lives long enough).
Genetic contributions to the model are incorporated into
the transmission parameters ␶AA, ␶AB, and ␶BB (the probability that individuals of a given type transmit the A factor to
their offspring). Mendelian models restrict the transmission
parameters to ␶AA ⫽ 1.0, ␶AB ⫽ 0.5, and ␶BB ⫽ 0; however,
more general models provide the ability to test whether
transmission fits these Mendelian expectations. The population frequencies of the three types of founding parents in the
pedigrees, the sum of which must be 1, are represented by
PAA, PAB, and PBB. Assuming Hardy–Weinberg equilibrium,
these three frequencies depend on q, the population frequency of allele A (PAA ⫽ q2, PAB ⫽ 2q[1 ⫺ q], and PBB ⫽
[1 ⫺ q]2). The effect of “type” is considered in the context
of two mathematical models. Model I assumes that the susceptibility is the same in all subjects, but that the age-atonset parameter (␤) depends on the person’s “type” (genotype AA, AB, or BB for the genetic models). Model II allows
“type” to influence a person’s susceptibility (␥), but not the
McDonnell et al: Familial Aggregation of PD
789
age-at-onset parameters (␣ and ␤). Our primary analyses focus on Model I. To correct for the method of ascertainment
of the probands, we conditioned the likelihood for each pedigree on the age at onset of PD in the proband.
Complex segregation analyses were performed by fitting
four Mendelian models (dominant, recessive, additive, and
codominant) and three non-Mendelian models (a general unrestricted model; an environmental model without generation effects [␶ ⫽ q], and an environmental model that allows
for generation differences [␶ ⫽ q]). In both environmental
models, the three “types” are allowed to differ in disease
probability (population heterogeneity). Hypotheses of disease
transmission were tested by likelihood ratio tests comparing a
restricted model with the general model allowing estimation
of all parameters (the best fitting model corresponds to the
least significant p value). The Akaike Information Criterion
(AIC) was used to compare nonhierarchical models (the
model with the smallest AIC provides the best fit to the observed data). Analyses were performed for all pedigrees combined and in two strata defined by age at onset of PD in the
proband (first tertile vs second and third tertiles combined).
Results
Characteristics of Probands
From a total of 202 incident cases of PD from Olmsted County, 162 were included (80%; 97 men and 65
women) and 40 (20%) refused to participate. The cases
excluded were more often alive (64%) than the cases
included (47%; p ⫽ 0.05). However, the distribution
by age and sex was similar in cases included and excluded. For 18 PD probands, the ethnic group was unknown. All remaining patients except one were white,
and the majority declared both parents of European
descent (one was Native American). The initial interview was direct for 47 cases and through a proxy for 90
cases. The remaining 25 probands were not interviewed, and the pedigree was constructed by passive
linkage of information from the Olmsted County Historical Society and the records-linkage system serving
Olmsted County. The 162 cases yielded 1,234 firstdegree relatives aged 40 years or older, of which 1,118
(91%) had data adequate for complex segregation analyses (disease status; age at onset, age at death, or age at
study).
We also recruited 411 patients with PD referred to
the Mayo Clinic (255 men and 156 women). A total
of 200 patients resided within the 120-mile radius
around Rochester, MN, whereas 211 patients resided
in a broader 5-state region (50 remainder of Minnesota, 67 remainder of Iowa, 44 remainder of Wisconsin, 20 North Dakota, and 30 South Dakota). Of the
referral patients with PD eligible for the study and invited to participate, only 9% refused to participate. Almost all the referral cases were white (407 cases; 99%),
and the majority declared both parents of European
origin (342 cases; 83%). Almost all of the initial interviews to construct the pedigree were direct (98%; only
790
Annals of Neurology
Vol 59
No 5
May 2006
10 interviews were with a proxy). The 411 referral
cases yielded 3,009 first-degree relatives aged 40 years
or older, of which 2,948 (98%) had data adequate for
complex segregation analyses.
Table 1 summarizes the demographic characteristics
of the probands, both overall and by geographic region
of referral (geographic stratum). The three strata were
similar for sex of the proband and number of firstdegree relatives. However, age at onset of PD decreased
with increasing distance of referral (test for linear
trend, p ⬍ 0.0001).
Overall Segregation Analyses
Eighty-two of the 573 families studied had at least 1
first-degree relative affected by PD (14%). The majority of these families had a single secondary case (66
families with 1; 14 families with 2; 1 family with 3; 1
family with 4). The results from fitting segregation
models in the combined population-based and referral
samples are presented in Table 2. The best fitting
model is that with the largest p value (the least statistical significance) and the smallest AIC. The estimated
transmission parameters from the general unrestricted
model are consistent with a Mendelian inheritance (see
Table 2). Both environmental models are clearly rejected because of inadequate fit compared with the
general model. Among the genetic models, the best fit
was provided by the additive model assuming that the
average age at disease onset for heterozygotes was exactly midway between the two types of homozygotes.
This model predicted an average decrease in age at onset of PD of approximately 18 years for each copy of
the putative high-risk allele. The codominant model
(which does not restrict the age at onset for heterozygotes) gave similar results, providing further support
for the additive model.
All seven models were also fit without restriction to
Hardy–Weinberg equilibrium. There was no evidence
for a lack of equilibrium in any of the seven models
(data not shown). Another potential source of heterogeneity was the sex of the relative. We fit all seven
models allowing for sex-specific estimates and found no
evidence of a sex effect for the gene influencing age at
onset (see Table 2). Parameter estimates were similar
for men and women.
As secondary analyses, each of the seven models of
transmission was fit using Model II that allows type
(eg, genotype) to influence susceptibility rather than
age at onset. All four Mendelian genetic models fit the
data equally, and the exclusion of environmental models was less definitive ( p ⫽ 0.06 without generation
effect; p ⫽ 0.04 with generation effect). Comparing
the results of Models I and II using AIC, Model I provided a better fit in all analyses (data not shown).
Table 1. Characteristics of the 573 Probands with Parkinson’s Disease
Geographic Stratum
Characteristics of Probands
Median age at onset, yr (range)
Female sex, n (%)
Median number of first-degree
relatives (range)
Median sibship size (range)
Median duration of PD, yr (range)b
Fathers
With adequate data
Affected, n (%)
Mothers
With adequate data
Affected, n (%)
Siblings
With adequate data
Affected, n (%)
Offspring
With adequate data
Affected, n (%)
Spouses
With adequate data
Affected, n (%)
Olmsted
County
(N ⫽ 162)
120-Mile
Radius
(N ⫽ 200)
5-State
Region
(N ⫽ 211)
All
Probands
(N ⫽ 573)
70 (40–97)a
65 (40.1)
10 (2–25)
65 (31–85)a
68 (34.0)
9 (2–25)
62 (28–82)a
88 (41.7)
10 (3–25)
65 (28–97)
221 (38.6)
9 (2–25)
3 (0–13)
8 (0.3–24)
162
101
1 (1.0)
162
116
5 (4.3)
458
457
20 (4.4)
326
326
3 (0.9)
126
118
3 (2.5)
3 (0–17)
3 (0.2–26)
198
188
4 (2.1)
198
187
7 (3.7)
626
626
14 (2.2)
318
318
0 (0.0)
118
117
2 (1.7)
3 (0–16)
3 (0.3–27)
210
192
13 (6.8)
210
192
13 (6.8)
712
712
14 (2.0)
299
299
0 (0.0)
120
117
2 (1.7)
3 (0–17)
5 (0.2–27)
570
481
18 (3.7)
570
495
25 (5.1)
1,796
1,795
48 (2.7)
943
943
3 (0.3)
364
352
7 (2.0)
a
We observed a linear trend of decreasing age at onset of Parkinson’s disease (PD) with increasing distance of geographic referral ( p ⬍
0.0001).25
b
The longer median duration of PD among Olmsted County cases is an artifact due to the selection of referral cases with shorter duration.26
Stratified Segregation Analyses
A formal test of heterogeneity across the three geographic strata was conducted by fitting segregation
models to each group separately and then comparing
the sum of the three subset likelihoods with the likelihood from the combined groups. No evidence of
heterogeneity was observed for any of the seven models considered (see Table 2). To evaluate possible genetic heterogeneity, we split the pedigrees into two
strata based on the age at onset of PD in the proband
(youngest tertile ⱕ 59 years vs second and third tertiles combined ⬎59 years) and determined the best fit
for all seven models. This age cutoff was chosen to yield
a similar number of affected relatives in both strata (44
secondary cases in the younger stratum and 57 in the
older stratum). All of the models, including the purely
environmental models, provided an adequate fit to the
data for both strata (Tables 3 and 4). In the younger
onset stratum, an autosomal recessive model provided
the best fit, whereas in the older onset stratum, both the
additive and the recessive model had similar fits. Our
difficulty in distinguishing between the models may be
due to the small number of secondary cases in each stratum.
Discussion
Comparison with Previous Studies
Our study suggests that the familial aggregation of PD
may be explained, in part, by a major gene following a
Mendelian model.2–15,25 In the combined populationbased and referral samples, we were able to clearly reject both environmental models. The best fit to the
data was provided by a model with an additive effect
on the penetrance. This result is consistent with Maher
and colleagues’23 finding of an additive major gene influencing age at onset. However, our allele frequency
was larger (q ⫽ 0.18 vs q ⫽ 0.02), and the decrease in
the average age at onset of PD for each copy of the
putative high-risk allele was smaller (18 vs 34.5
years).23
Although a single-locus model provided an adequate
fit to our data, it is important to recognize that segregation analyses cannot distinguish between a single locus for all families and multiple loci across different
families. For example, if there were several different
major genes segregating in different groups of families,
and all acted in an additive fashion, then our observed
estimates of penetrance would be the average of the
penetrances of the multiple loci, and the estimated al-
McDonnell et al: Familial Aggregation of PD
791
Table 2. Complex Segregation Analyses for Age at Onset of Parkinson’s Disease (Model I): Combined Sample of Families of
Probands from Olmsted County, the 120-Mile Radius, and the 5-State Region
Environmental Models
Mendelian Models
Parameter
Dominant Recessive
Additive
With
Without
Generation Generation
Effects
Effects
Codominant
qA
0.035
0.295
0.176
␶AA
1.0b
1.0b
1.0b
␶AB
0.5b
0.5b
0.5b
b
b
␶BB
0.0
0.0
0.0b
␤AA
⫺11.93a ⫺12.89
⫺13.27
␤AB
⫺11.93a ⫺16.62a ⫺17.01
␤BB
⫺15.14 ⫺16.62a ⫺20.76
␣
0.176
0.196
0.208
␥
0.116
0.110
0.131
⫺2 lnL
1,396.43 1,395.15 1,388.15
Model fit ␹2
9.18
7.90
0.90
df
4
4
4
p
0.057
0.095
0.925
AICc
1,406.43 1,405.15 1,398.15c
Test of heterogeneity across the three geographic regionsd
Model fit ␹2
10.38
11.14
9.12
df
10
10
10
p
0.408
0.347
0.521
a
0.176
1.0b
0.5b
0.0b
⫺13.25
⫺17.01
⫺20.57
0.208
0.130
1,388.13
0.88
3
0.830
1,400.13
11.24
12
0.509
General
Model
Sex-Dependent
General Model
Female
Male
0.387a
0.314
0.167
0.156
0.387a
0.502a
0.91
1.00
0.387a
0.502a
0.62
0.57
a
0.387
0.502a
0.0
0.0
⫺11.93
⫺12.46
⫺12.88 ⫺13.41 ⫺12.84
⫺15.38
⫺15.82
⫺16.40 ⫺17.39 ⫺15.98
⫺15.38
⫺15.76
⫺19.51 ⫺22.75 ⫺19.04
0.186
0.191
0.201
0.208
0.202
0.098
0.097
0.126
0.132
0.132
1,406.01
1,402.76 1,387.25
1,382.60
18.75
15.51
—
4.65
3
2
—
5
0.0003
0.0004
—
0.46
1,418.00
1,416.76 1,405.25
1,410.60
9.18
12
0.688
15.50
14
0.345
18.48
18
0.425
—
—
—
Parameters are constrained to be equal within each column.
b
c
Parameter is fixed at value shown.
The lowest Akaike Information Criterion (AIC) was observed for the additive model.
d
A formal test of heterogeneity across the three geographic strata of probands indicated no evidence of heterogeneity for any model tested.
df ⫽ degrees of freedom.
lele frequency would be the sum of the allele frequencies at the multiple loci.
When the pedigrees were stratified by age at onset of
PD in probands, it was difficult to distinguish among
models, probably due to the small number of secondary cases in each stratum. Despite this difficulty, significant heterogeneity was observed for all seven models across the two subsets, possibly indicating different
genetic and environmental mechanisms for younger
and older onset PD (see Table 3).
Eleven genetic loci related to PD have been identified through linkage analysis, and causal mutations
have been identified in six genes to date (SNCA, parkin, UCHL1, DJ1, PINK1, and LRRK2).1,35 Some of
these genes are involved in autosomal dominant PD,
whereas others are involved in autosomal recessive
PD.1,36 This genetic heterogeneity may explain some
of the difficulties with segregation analyses. Although
these genes have been shown to explain PD in some
families, their combined frequency explains only a
small fraction (not yet known) of PD in the population. Therefore, additional genetic contributions to PD
remain to be elucidated. Model parameters provided by
our best fitting additive model can be used to specify
the gene frequency and age-specific penetrance for future parametric linkage analyses.
792
Annals of Neurology
Vol 59
No 5
May 2006
Strengths and Limitations of This Study
First, an important strength of this study was the quality of phenotypic assessment among the first-degree relatives. The presence of PD was confirmed by direct
examination in 27 relatives of cases, by medical record
documentation in 29, and by direct or proxy interview
in 45 (1 direct and 44 proxy interviews). In addition,
387 relatives of cases were examined by an author and
found not to be affected by PD despite a positive
screening. In these analyses, we included only relatives
affected by PD, rather than parkinsonism, because our
familial aggregation findings suggested a genetic effect
restricted to PD.25 However, some misclassification of
parkinsonism and PD might have occurred (despite
our best efforts against it).
Second, we were able to include in the study both a
population-based series of incident cases of PD and a
referral series of cases. The population-based series
avoided possible biases related to survival (incidenceprevalence bias) and to selection for inclusion in the
study (referral bias).37 However, the series was somewhat limited by the size of the Olmsted County population. To increase our statistical power, we added a
larger referral series of cases, and we showed absence of
heterogeneity of findings across the three geographic
regions considered.
Table 3. Complex Segregation Analyses (Model I) for Families of Probands with Younger Onset of Parkinson’s Disease: Age at
Onset ⱕ 59 Years (First Tertile); total of 44 cases among first-degree relatives.
Environmental Models
Mendelian Models
Dominant
Parameter
qA
0.009
␶AA
1.0b
␶AB
0.5b
␶BB
0.0b
␤AA
⫺12.30a
␤AB
⫺12.30a
␤BB
⫺15.34
␣
0.189
␥
0.197
⫺2 lnL
556.68
Model fit ␹2
7.73
df
4
p
0.102
AICc
566.68
Test of heterogeneity between the two
Model fit ␹2
20.95
df
5
p
0.0008
a
Additive
0.176
0.152
1.0b
1.0b
0.5b
0.5b
0.0b
0.0b
⫺13.60
⫺13.65
⫺17.40a
⫺17.04
⫺17.40a
⫺20.42
0.214
0.209
0.193
0.219
554.28
556.44
5.33
7.49
4
4
0.255
0.112
564.28c
566.44
strata by age at onsetd
22.46
13.14
5
5
0.0004
0.022
Codominant
0.176
1.0b
0.5b
0.0b
⫺13.56
⫺17.34
⫺17.34
0.213
0.193
554.28
5.33
3
0.149
566.28
15.53
6
0.017
With
Generation
Effects
0.164a
0.164a
0.164a
0.164a
⫺18.24
⫺24.12
⫺29.22
0.361
0.167
553.01
4.06
3
0.255
565.01
0.128
0.236a
0.236a
0.236a
⫺19.07
⫺24.97
⫺29.99
0.369
0.167
550.63
1.68
2
0.432
564.63
31.19
6
0.00002
General
Model
0.208
0.097
0.100
0.281
⫺19.13
⫺24.98
⫺29.96
0.369
0.168
548.95
—
—
—
566.95
33.41
7
0.00002
21.84
9
0.009
Parameters are constrained to be equal within each column.
b
c
Recessive
Without
Generation
Effects
Parameter is fixed at value shown.
The lowest Akaike Information Criterion (AIC) was observed for the recessive model.
d
A formal test of heterogeneity between strata by age at onset of Parkinson’s disease (PD) among probands suggested heterogeneity for all seven
models.
df ⫽ degrees of freedom.
Table 4. Complex Segregation Analyses (Model I) for Families of Probands with Older Onset of Parkinson’s Disease: Age at
Onset ⬎ 59 Years (Second and Third Tertiles Combined); total of 57 cases among first-degree relatives.
Environmental Models
Parameter
Dominant
Recessive
Additive
Codominant
Without
Generation
Effects
qA
␶AA
␶AB
␶BB
␤AA
␤AB
␤BB
␣
␥
⫺2 lnL
Model fit ␹2
df
p
AICc
0.048
1.0b
0.5b
0.0b
⫺11.84a
⫺11.84a
⫺15.07
0.175
0.084
818.80
2.34
4
0.674
828.80
0.333
1.0b
0.5b
0.0b
⫺12.25
⫺15.77a
⫺15.77a
0.183
0.085
818.41
1.95
4
0.745
828.41c
0.259
1.0b
0.5b
0.0b
⫺12.33
⫺15.39
⫺18.45
0.185
0.096
818.57
2.11
4
0.716
828.57
0.277
1.0b
0.5b
0.0b
⫺12.25
⫺15.41
⫺17.25
0.184
0.091
818.32
1.86
3
0.602
830.32
0.348a
0.348a
0.348a
0.348a
⫺11.27
⫺13.79
⫺15.57
0.171
0.085
821.80
5.34
3
0.149
833.80
Mendelian Models
a
General
Model
0.294
0.522a
0.522a
0.522a
⫺11.77
⫺14.28
⫺15.46
0.171
0.084
818.72
2.26
2
0.323
832.72
0.236
1.00
0.73
0.089
⫺12.32
⫺15.32
⫺16.57
0.183
0.085
816.46
—
—
—
834.46
Parameters are constrained to be equal within each column.
b
c
With
Generation
Effects
Parameter is fixed at value shown.
The lowest Akaike Information Criterion (AIC) was observed for the recessive and additive models.
df ⫽ degrees of freedom.
McDonnell et al: Familial Aggregation of PD
793
Third, a unique feature of this study was the access
to the records-linkage system serving the community of
Olmsted County.26,31,32 The system provided medical
record documentation for 486 relatives of probands
who resided in the same county. In addition, combining information from the system with data from the
Olmsted County Historical Society, we were able to
construct pedigrees and to study the familial aggregation of parkinsonism in 25 families of Olmsted County
probands who could not be interviewed.26
Our study has a number of limitations. The first
limitation is the relatively small sample size of the
population-based sample leading to a small number of
events among relatives of cases and, therefore, to limited power in this group. Unfortunately, our series of
incident cases with PD was determined by the size of
the Olmsted County population. This limitation was
addressed by the addition of 411 families of referral
cases. In total, we included in the study 573 probands
and their 4,243 relatives, yielding 306,228 person-years
of follow-up, and 101 secondary cases.
Second, despite our best efforts, we failed to study
any of the relatives of 40 cases (20%). To investigate
any possible bias introduced by this nonparticipation,
we compared the demographic and clinical characteristics of the cases excluded and included, and we found
no major differences except that cases excluded were
more often alive than cases included.26,37 Third, despite our best efforts, the presence of PD could be documented only through the interview of a proxy informant for 44 of the 101 secondary cases (44%). For
these secondary cases, we cannot exclude the presence
of a residual family information bias.16,37 However,
our results for the familial aggregation of PD were similar after excluding these secondary cases with less certain diseases status.25
This study was supported by the NIH grants NINDS R01
NS33978 (S.K.M., D.J.S., K.J.S., J.H.B., J.E.A., D.M.M., W.A.R.)
and NIEHS R01 ES10751 (D.M.M., W.A.R.). In addition, the
NIH grant NIAMSD R01 AR30582 provided infrastructural support for the Rochester Epidemiology Project. Finally, A.E. was
funded by a postdoctoral fellowship from INSERM Unit 360, and
from the Mayo Foundation. Some of the results of this article were
obtained by using the program package SAGE, which is supported
by the National Center for Research Resources (US Public Health
Service Resource Grant RR03655).
We thank R.R. Black, K.J. Brown, J.A. Cogswell, C.J. Ellefson,
C.D. Garvey, K.M. Kuntz, D.J. Pilgrim, V.S. Schwartz, and T.L.
Shevlin for their assistance with data collection, and K.F. Tennison
for typing the manuscript.
References
1. Vila M, Przedborski S. Genetic clues to the pathogenesis of
Parkinson’s disease. Nat Med 2004;10:S58 –S62.
2. Semchuk KM, Love EJ, Lee RG. Parkinson’s disease: a test of
the multifactorial etiologic hypothesis. Neurology 1993;43:
1173–1180.
794
Annals of Neurology
Vol 59
No 5
May 2006
3. Morano A, Jiménez-Jiménez FJ, Molina JA, Antolı́n MA. Riskfactors for Parkinson’s disease: case-control study in the province of Cáceres, Spain. Acta Neurol Scand 1994;89:164 –170.
4. Payami H, Larsen K, Bernard S, Nutt J. Increased risk of Parkinson’s disease in parents and siblings of patients. Ann Neurol
1994;36:659 – 661.
5. Bonifati V, Fabrizio E, Vanacore N, et al. Familial Parkinson’s
disease: a clinical genetic analysis. Can J Neurol Sci 1995;22:
272–279.
6. De Michele G, Filla A, Volpe G, et al. Environmental and genetic risk factors in Parkinson’s disease: a case-control study in
southern Italy. Mov Disord 1996;11:17–23.
7. Marder K, Tang M-X, Mejia H, et al. Risk of Parkinson’s disease among first-degree relatives: a community-based study.
Neurology 1996;47:155–160.
8. McCann SJ, LeCouteur DG, Green AC, et al. The epidemiology of Parkinson’s disease in an Australian population. Neuroepidemiology 1998;17:310 –317.
9. Elbaz A, Grigoletto F, Baldereschi M, et al. Familial aggregation of Parkinson’s disease. A population-based case-control
study in Europe. Neurology 1999;52:1876 –1882.
10. Rybicki BA, Johnson CC, Peterson EL, et al. A family history
of Parkinson’s disease and its effect on other PD risk factors.
Neuroepidemiology 1999;18:270 –278.
11. Taylor CA, Saint-Hilaire MH, Cupples LA, et al. Environmental, medical, and family history risk factors for Parkinson’s
disease: a New England-based case control study. Am J Med
Genet 1999;88:742–749.
12. Autere JM, Moilanen JS, Myllylä VV, Majamaa K. Familial aggregation of Parkinson’s disease in a Finnish population. J Neurol Neurosurg Psychiatry 2000;69:107–109.
13. Sveinbjörnsdóttir S, Hicks AA, Jónsson T, et al. Familial aggregation of Parkinson’s disease in Iceland. N Engl J Med 2000;
343:1765–1770.
14. Payami H, Zareparsi S, James D, Nutt J. Familial aggregation
of Parkinson disease. A comparative study of early-onset and
late-onset disease. Arch Neurol 2002;59:848 – 850.
15. Marder K, Levy G, Louis ED, et al. Familial aggregation of
early- and late-onset Parkinson’s disease. Ann Neurol 2003;54:
507–513.
16. Elbaz A, McDonnell SK, Maraganore DM, et al. Validity of
family history data on PD. Evidence for a family information
bias. Neurology 2003;61:11–17.
17. Tanner CM, Ottman R, Goldman SM, et al. Parkinson disease
in twins. An etiologic study. JAMA 1999;281:341–346.
18. Wirdefeldt K, Gatz M, Schalling M, Pedersen NL. No evidence
for heritability of Parkinson disease in Swedish twins. Neurology 2004;63:305–311.
19. Piccini P, Burn DJ, Ceravolo R, et al. The role of inheritance
in sporadic Parkinson’s disease: evidence from a longitudinal
study of dopaminergic function in twins. Ann Neurol 1999;45:
577–582.
20. Maraganore DM, Harding AE, Marsden CD. A clinical and
genetic study of familial Parkinson disease. Mov Disord 1991;
6:205–211.
21. Lazzarini AM, Myers RH, Zimmerman TR Jr, et al. A clinical
genetic study of Parkinson’s disease: evidence for dominant
transmission. Neurology 1994;44:499 –506.
22. Zareparsi S, Taylor TD, Harris EL, Payami H. Segregation
analysis of Parkinson disease. Am J Hum Genet 1998;80:
410 – 417.
23. Maher NE, Currie LJ, Lazzarini AM, et al. Segregation analysis
of Parkinson disease revealing evidence for a major causative
gene. Am J Med Genet 2002;109:191–197.
24. Korchounov A, Schipper HI, Preobrazhenskaya IS, et al. Differences in age at onset and familial aggregation between clinical types of idiopathic Parkinson’s disease. Mov Disord 2004;
19:1059 –1064.
25. Rocca WA, McDonnell SK, Strain KJ, et al. Familial aggregation of Parkinson’s disease: the Mayo Clinic Family Study. Ann
Neurol 2004;56:495–502.
26. Rocca WA, Peterson BJ, McDonnell SK, et al. The Mayo
Clinic Family Study of Parkinson’s Disease: study design, instruments, and sample characteristics. Neuroepidemiology
2005;24:151–167.
27. Bower JH, Maraganore DM, McDonnell SK, Rocca WA. Incidence and distribution of parkinsonism in Olmsted County,
Minnesota, 1976-1990. Neurology 1999;52:1214 –1220.
28. Bower JH, Maraganore DM, McDonnell SK, Rocca WA. Influence of strict, intermediate, and broad diagnostic criteria on
the age- and sex-specific incidence of Parkinson’s disease. Mov
Disord 2000;15:819 – 825.
29. Rocca, WA, Bower JH, McDonnell SK, et al. Time trends in
the incidence of parkinsonism in Olmsted County, Minnesota.
Neurology 2001;57:462– 467.
30. Rocca WA, Maraganore DM, McDonnell SK, Schaid DJ. Validation of a telephone questionnaire for Parkinson’s disease.
J Clin Epidemiol 1998;51:517–523.
31. Kurland LT, Molgaard CA. The patient record in epidemiology. Sci Am 1981;245:54 – 63.
32. Melton LJ III. History of the Rochester Epidemiology Project.
Mayo Clin Proc 1996;71:266 –274.
33. Bonney GE. Regressive logistic models for familial disease and
other binary traits. Biometrics 1986;42:611– 625.
34. SAGE. Statistical Analysis for Genetic Epidemiology, Release
3.1. Computer program package from the Department of Epidemiology and Biostatistics, Rammelkamp Center for Education and Research, MetroHealth Campus, Case Western Reserve University, Cleveland, OH, 1998.
35. Brice A. How much does dardarin contribute to Parkinson’s
disease? Lancet 2005;365:363–364.
36. Hardy J, Cookson MR, Singleton A. Genes and parkinsonism.
Lancet Neurol 2003;2:221–228.
37. Sackett DL. Bias in analytic research. J Chron Dis 1979;32:
51– 63.
McDonnell et al: Familial Aggregation of PD
795
Документ
Категория
Без категории
Просмотров
0
Размер файла
87 Кб
Теги
complex, segregation, clinics, family, stud, mayo, analysis, disease, parkinson
1/--страниц
Пожаловаться на содержимое документа