American Journal of Medical Genetics 124A:133 135 (2004)

Results of a Genome-Wide Linkage

Scan for Stuttering
Yin Yao Shugart,1 Jennifer Mundorff,2 James Kilshaw,3,4 Kimberly Doheny,1 Betty Doan,1
Jacqueline Wanyee,1 Eric D. Green,5 and Dennis Drayna3*
1

Center for Inherited Disease Research, Johns Hopkins University Bayview Research Campus, Baltimore, Maryland
Hollins Communications Research Institute, Roanoke, Virginia
3
National Institute on Deafness and Other Communication Disorders, National Institutes of Health, Rockville, Maryland
4
Stuttering Foundation of America, Memphis, Tennessee
5
National Human Genome Research Institute, National Institutes of Health, Bethesda, Maryland
2

We performed a linkage study of stuttering

using 392 markers distributed across the
genome in a series of 68 families identified
in the general outbred population of North
America and Europe. Standardized diagnosis was performed using recorded samples
of both conversation and reading, in which
stuttering dysfluencies were scored as percentage of dysfluent words and syllables.
Analysis was first performed using nonparametric methods implemented in GENEHUNTER, where we obtained maximum
statistical support for markers of chromosome 18, with a maximum NPL (Sall) of 1.51 at
D18S976. The single largest pedigree within
our sample (pedigree 0006) alone gave an
NPL of 4.72 at D18S976. For fine mapping,
we analyzed 18 markers on chromosome 18
across all families using ALLEGRO. Overall
NPL (Srobdom) scores >5 were obtained with
markers on 18p, and Zlr scores
2.5 on 18p
and proximal 18q. Furthermore, pedigree
0006 alone gave an NPL (Srobdom) of 5.35.
Overall our results suggest chromosome 18
may harbor a predisposing locus for this
disorder, and additional genes may exist.
Published 2003 Wiley-Liss, Inc.{

KEY WORDS: stuttering; linkage; speech

disorder;
non-parametric
analysis

INTRODUCTION
Stuttering is a disorder of the rhythm of speech
characterized by involuntary repetition or prolongation
of syllables, and by interruptions in the smooth flow of
speech, known as blocks. Although no clear cause of this
disorder has been identified, genetic factors in stuttering have long been suggested [Bloodstein, 1995; Yairi
et al., 1996]. A number of factors, however, have hindered genetic studies of stuttering, including a highly
distorted sex ratio, an inability to ascribe a mode of
inheritance, the high frequency of the trait in normal
young children, and greatly variable expression within
families. Our previous studies have suggested that at
least one of these complications, the distorted sex ratio,
may be less significant in familial stuttering than in the
general stuttering population [Drayna et al., 1999]. To
address the other complicating factors, we employed a
simplified approach for a linkage study, using primarily
affected family members who displayed persistent
stuttering beyond young childhood. Our strategy was
to perform a genome-wide linkage survey and use nonparametric analysis methods to identify genomic regions of interest, and then gather additional data and
perform additional analysis.
MATERIALS AND METHODS

Grant sponsor: NIDCD; Grant number: Z01 00046-03; Grant

sponsor: National Institutes of Health (The Johns Hopkins
University); Grant number: N01-HG-65403.
Yin Yao Shugarts present address is Department of Epidemiology, Johns Hopkins School of Public Health, 615 N. Wolfe St.,
Baltimore, MD 21205.
*Correspondence to: Dr. Dennis Drayna, NIDCD/NIH, 5
Research Court, Room 2B-46, Rockville, MD 20850.
E-mail: drayna@nidcd.nih.gov
Received 5 December 2002; Accepted 22 April 2003
DOI 10.1002/ajmg.a.20347

Families were recruited from the general outbred

population of North America and Great Britain via
targeted appeals to stuttering interest groups and to the
general public. Subjects were individuals over the age of
8 (mean age 31 years) who currently stuttered, and had
done so for at least 6 months. Families were ascertained
and enrolled under National Institutes of Health IRBapproved protocol no. 97-DC-0057, originally approved by NCI IRB on 1/8/97, and annually reviewed and
approved by the NINDS IRB thereafter. Informed

Published 2003 Wiley-Liss, Inc.

{
This article was prepared by a group consisting of both United
States Government employees and non-United States Government
employees, and as such is subject to 117 U.S.C. Sec. 105.

134

Shugart et al.

consent was obtained from all subjects or their legal

guardians, and an additional assent was obtained from
minor subjects. A complete diagram of the pedigrees of
all families included is available from the corresponding
author. We excluded families with documented or reported stuttering in both the maternal and paternal
lineages, as such bilineal families can obscure nonparametric statistical analysis. The sample consisted of
68 families; 23 of these families consisted of one generation, 33 contained two generations, and 12 contained
three generations. Overall, the families contained an
average of 2.72 affected individuals per family. The
sample contained 48 sibships with 2 affected, 9 sibships
with 3 affected, 2 shipships 4 affected, and one pair of
affected half-sibs. Of the 48 sibships with 2 affected, 22
had zero parents, 13 had 1 parent, and 13 had 2 parents
genotyped and included in the analysis.
For diagnostic consistency, stuttering was evaluated
by a single clinician (J.M.), using a standardized reading
text of 500 words containing balanced numbers of each
class of speech sounds [Webster, 1978] plus 5 min of free
conversation speech. Text of the standard reading passage can be obtained from the corresponding author at:
drayna@nidcd.nih.gov. Speech samples were recorded
and stuttering dysfluencies (repetitions, prolongations,
and blocks) were counted and evaluated as both percent
of words spoken and percent of syllables spoken. Individuals were classified as affected if they demonstrated

4% word dysfluencies in either reading or free speech,
a cut-off typically regarded as mild-to-moderate stuttering in clinical settings. All subjects were over the age
of 8 and had documentation of stuttering for 6 months
or more. DNA samples were obtained as previously
described [Muellenbeldt et al., 1995].
A total of 226 individuals in 68 families were included,
188 of whom were affected.
Genotyping was performed by standard fluorescent
methods on an ABI 3700. The markers used were a
modification of the CHLC version 9 marker set (392
markers, average spacing 9 cM, average heterozygosity
0.76). See www.cidr.jhmi.edu for details on genotyping
methods. The error rate for the genome scan, based on
2,993 paired genotypes from blind duplicate samples,
was 0.15%. The overall missing data rate was 14.4%,
largely due to failed genotyping reactions with buccal
swab DNA in the high throughput system employed.
After accounting for failed genotyping reactions, a total
of 81,928 genotypes were generated and analyzed in the
initial genome-wide scan. Marker allele frequencies
were computed using data from one individual randomly chosen from each pedigree. Initial genome-wide
linkage analysis was performed coding unaffected
individuals as such. Because of diagnostic uncertainties
in subsequently sampled individuals in pedigree 0006,
high resolution analysis on chromosome 18 was performed in this family coding individuals as either
affected or unknown.
RESULTS
Initial analysis across the entire genome using
GENEHUNTER [Kruglyak et al., 1996] indicated

evidence for linkage to a number of markers distributed

broadly across chromosome 18p and proximal 18q. The
summed NPL Sall scores for all families was 1.51 at
D18S976 (P 0.03). Analysis of our largest family alone
(pedigree 0006, containing six affected individuals)
gave an NPL of 4.72 at D18S976. This genome scan also
gave NPL scores greater than 1 but less than 1.5 at loci
on chromosome 1, 2, 10, and 13. Interestingly, pedigree
0006 only gave highly positive scores on chromosome
18q. Therefore, we then focused attention on chromosome 18 chromosomal region for additional analysis.
For fine mapping, we typed 12 additional markers
spanning approximately 60 cM on chromosome 18 for all
families, resulting in a final combined marker spacing of
3.3 cM in this region. We then performed both parametric and non-parametric statistical analyses using
ALLEGRO [Gudbjartsson et al., 2000]. For the parametric analysis, an incomplete dominant mode of inheritance model was assumed, based on the overall
pattern of stuttering in our family sample. Specifically,
the penetrance was set to be 0.9 and the disease
frequency was set to be 0.001.
While the overall LOD score across all families were
negative (LOD <  10.23 through the whole chromosome), the heterogeneous LOD (HLOD) was 1.42 at the
marker D18S78, suggesting genetic heterogeneity in
this data set. The estimated proportion of families linked
to the locus of interest was 78%. Penetrance values
varied from 5085% were also used in additional parametric analysis, and the results for HLOD did not
change severely, indicating our penetrance of 0.9 was a
reasonable estimate.
Because of the dominant inheritance pattern that
stuttering presents in many of our more informative
families, for the non-parametric analysis we employed
the new statistic Srobdom implemented in ALLEGRO,
which has been shown to be more powerful than other
non-parametric statistics under a dominant model
[McPeek, 1999]. Statistically, Srobdom was defined as
Sri2Acli  1, where A is the set of all alleles observed
for the locus among the affected individuals in the
pedigree, cl(i) is the number of affected individuals in
the pedigree with at least one copy of allele I, and r is
the relative risk. McPeek [1999] further stated that
under the assumption of a dominant model with predisposing allele frequency approaching zero, the power
to detecting linkage is not very sensitive to the choice
of r.
In this analysis, the best evidence for linkage was
again found at D18S78, which gave a NPL score of 5.143,
and a Zlr score of 2.63, with P 0.0043. In addition, a
second region of positive linkage scores was observed on
proximal 18q, surrounding marker D18S847 (Zlr score
of 2.47), which gave P 0.0068. In these 68 outbred
families, our data gives a Zlr score of 2.63 (equivalent to a
LOD of 1.50) and a heterogeneous parametric LOD of
1.42, which suggest a locus on chromosome 18 may be
important in the genesis of this disorder.
The region of chromosome 18 implicated in this
analysis is quite large, and contains several hundred
known and putative genes. Possible interesting candidate genes include a cluster of genes in the desmoglein/

Stuttering Linkage Scan

desmocolin family on 18q12.1, and the neuronal cadherin 2 gene on 18q11.2, both of which are known to be
involved in cell adhesion and intercellular communication. Such communications may be important in the
neurons involved in speech production in the brain.
There has been one other report of the results from a
genome-wide linkage survey, which was performed in
the Hutterites, a highly genetically isolated population
[Cox and Yairi, 2000]. This survey also gave support for
linkage on chromosomes 1, 13, and 16, where we obtained NPL scores of 1.1, 1.38, and 0.518, respectively.
Although it is not clear whether the markers that gave
positive NPL scores from our genome scan overlap with
the markers investigated by Cox and Yairi, it appears
that chromosomes 1 and 13 may warrant further study.
The lack of linkage on chromosome 18 in the Hutterites
suggests stuttering may display locus heterogeneity in
different study populations. Because stuttering displays
complex inheritance, it will be important to confirm our
results in additional independent samples. We propose
to collect additional families, and to make our genotypic
data available to facilitate these necessary confirmatory
studies.

135

in our family ascertainment and enrollment. Genotyping services at C.I.D.R. were funded through the
National Institutes of Health via The Johns Hopkins
University, Contract Number N01-HG-65403.
REFERENCES
Bloodstein O. 1995. A handbook of stuttering. 5th edn. San Diego: Singular
Publishing Group.
Cox N, Yairi E. 2000. Genetics of stuttering: Insights and recent advances.
ASHA Leader 5(16). Bethesda: American Speech Language Hearing
Association. Abstract no. 674.
Drayna D, Kilshaw J, Kelly J. 1999. The sex ratio in familial persistent
stuttering. Am J Hum Genet 65:14731475.
Gudbjartsson DF, Jonason K, Frigge ML, Kong A. 2000. Allegro, a new
computer program for multipoint linkage analysis. Nat Genet 25(1):12
13.
Kruglyak L, Daly MJ, Reeve-Daly MP, Lander ES. 1996. Parametric and
nonparametric linkage analysis: A unified multipoint approach. Am
J Hum Genet 58(6):13471363.
McPeek RM. 1999. Optimal allele-sharing statistics for genetic mapping
using affected relatives. Genet Epidemiol 16(3):225249.
Meulenbelt I, Droog S, Trommelin GJ, Boomsa D, Slagboom E. 1995. Highyield non-invasive human genomic DNA isolation method for genetic
studies in geographically dispersed families and populations. Am J Hum
Genet 57:12521254.

ACKNOWLEDGMENTS

Webster R. 1978. Empirical considerations regarding stuttering, Chapter 6.

In: Gregory H, editor. Controversies about stuttering therapy. Baltimore: University Park Press.

We thank the assistance from the Stuttering Foundation of America and the British Stammering Association

Yairi E, Ambrose N, Cox N. 1996. Genetics of stuttering: A critical review.

J Speech Lang Hear Res 39:771784.