Finding susceptibility genes for developmental
disorders of speech: the long and winding road
Susan Felsenfeld
Department of Speech-Language Pathology, Duquesne University, 600 Forbes Avenue,
Pittsburgh, PA 15282, USA
Received 27 February 2002; received in revised form 6 April 2002; accepted 6 April 2002
Abstract
Finding susceptibility genes for complex disorders is the next major challenge facing
genetics researchers. The purpose of this paper is to stimulate creative thinking about the genefinding process for developmental speech disorders (DSDs), specifically disorders of articulation/phonology and stuttering. The paper will begin with a review of existing behavioral
genetic studies of these phenotypes. This will be followed by a discussion of roadblocks that
may impede the molecular study of DSDs, research that is in very early stages of development.
As a third objective, the small number of molecular genetic studies of DSDs that have been
published or presented will be described. The paper concludes with a discussion of research
strategies that may maximize the success of molecular studies of speech phenotypes. It will be
argued that progress will most likely be enhanced if theories about biological systems and
processes can be used to narrow the search for candidate susceptibility genes. Learning
outcomes: The reader will be introduced to findings and conceptual issues that relate to
the behavioral and molecular genetic investigation of DSDs. After completing this paper,
readers should be able to (a) identify key epidemiological findings for the three speech
phenotypes that were discussed (DAS, speech delay, and stuttering); (b) summarize the
findings of the behavioral genetic studies of speech disorders that were presented; (c)
identify four specific challenges that may impede future molecular genetic studies of these
phenotypes; (d) describe the methodological sequence that led to the discovery of the
FOXP2 gene; and (e) summarize the two research strategies that were presented to
potentially reduce sample heterogeneity for future molecular genetics research. © 2002
Elsevier Science Inc. All rights reserved.
Keywords: Developmental speech disorders; Stuttering; Genetics; Familial aggregation;
Endophenotypes
E-mail address: felsenfeld@duq.edu (S. Felsenfeld).
0021-9924/02/$ - see front matter © 2002 Elsevier Science Inc. All rights reserved.
PII: S 0 0 2 1 - 9 9 2 4 ( 0 2 ) 0 0 0 8 8 - 6
330
S. Felsenfeld/Journal of Communication Disorders 35 (2002) 329-345
"The road to the City of Emeralds is paved with yellow brick." L.
Frank Baum, The Wonderful Wizard of Oz
Compared to other clinical disciplines, scientists who study speech disorders
are relative newcomers on the long and winding gene-finding road. Although it
will not be an easy road to travel, there is good reason to be optimistic that our own
version of the City of Emeralds awaits us at journey's end. The speech susceptibility genes that we seek are almost certain to exist, and finding them will
inevitably lead to a greater understanding of the nature and treatment of communication disorders of unknown origin. So, onward fellow travelers in our quest
to lay the foundational genetic bricks!
I will begin this paper by providing a brief epidemiological overview of the two
broad clinical phenotypes that are typically included under the umbrella of
developmental speech disorders (DSDs) of unknown origin (articulation/phonological disorder and stuttering). This review will be brief, since it is assumed that
most readers are familiar with the clinical features of these disorders. Following
this, I will present a selective review of existing behavioral genetic studies of both
of these phenotypes, focusing on studies that have been performed over the past
two decades. As a third objective, I will identify several common methodological
roadblocks that can complicate the molecular study of complex disorders, including disorders of speech. As a fourth objective, I will review the small number of
molecular genetic studies that have been performed to date for DSDs. As a final
objective, I will discuss two specific research strategies that may maximize the
potential for success of molecular genetic studies of these conditions.
1. Overview of the clinical phenotypes
DSDs are typically classified as either disorders affecting articulation/phonology or disorders affecting fluency (i.e., stuttering). As readers are well aware, the
clinical hallmark of articulation/phonological disorders is a reduction in the
intelligibility of speech for age. Developmental stuttering presents as a chronic
disruption in an individual's ability to produce smooth, effortless, and forwardmoving speech. Although these broad disorder categories are usually considered
discrete diagnoses, they may actually be comprised of several related but
etiologically distinct clinical subgroups.1 For example, there is growing evidence
to support the existence of at least two discrete subgroups of children affected
with a developmental articulation/phonological disorders, a speech delay subgroup and a subgroup of children with suspected developmental apraxia of
speech, or DAS. For developmental stuttering, no comparably well-established
'Larry Shriberg and his colleagues have developed a sophisticated multi-group classification
system for articulation/phonological disorders. Readers are referred to Shriberg, 1997 and Shriberg
et al., 1997.
S. Felsenfeld/Journal of Communication Disorders 35 (2002) 329-345
331
homogeneous subgroups have yet been identified, with the possible exception of
subgroups-based upon recovery status (early recovered versus persistent stuttering cases). It is likely that either this or alternative homogeneous subgroups will
emerge for developmental stuttering as genetics research in this area continues to
mature (Felsenfeld, 1996; Yairi, Ambrose, & Cox, 1996).
Identifying homogeneous subgroups for these disorders is of more than
theoretical interest. For molecular geneticists, these (narrower) subject classifications provide a more favorable analysis environment, and make it more likely that
candidate loci can be identified that underlie that particular disorder phenotype.
Of course, candidate regions that are identified for one subgroup of a disorder
(e.g., the DAS subgroup) may not generalize to other subgroups sharing similar
clinical features (e.g., the speech delay subgroup). This somewhat discouraging
reality underscores the considerable challenge that awaits us on the long genetics
road. Even if we find a promising set of candidate genes for one phenotype, this
knowledge may tell us very little about the gene or genes responsible for
provocatively similar speech disorders.
Table 1 presents epidemiological observations about DSDs that were drawn
from several published sources (Felsenfeld, 1996; Shriberg, Aram, & Kwiatkowski,
1997a; Shriberg, Tomblin, & McSweeny, 1999; Yairi et al, 1996). These data are
valuable to genetics researchers because they provide insights about the distribution
of the disorder in the population. Geneticists examine epidemiological data to
determine who in the population is at greatest risk for presentation of the disorder
phenotype and when the risk period for disorder onset is most likely to have expired.
Ultimately, the distribution of affected cases is evaluated to identify departures from
random expectation; for example, researchers look to see whether one gender is
preferentially affected, if comorbidity with other disorders appears at a rate that
exceeds chance, or if familial aggregation is present. These observations are used to
help direct genetic hypothesis testing and to determine when an individual who is
participating in a genetic study has passed the typical age of onset for the disorder
and may be classified as either affected or unaffected with some confidence.
Table 1
Selected epidemiological observations for DSDs
Articulation disorder
DAS
Stuttering
Speech delay
14.0% ages 3-4;
3.8% (age 6+)
Prevalence
0.125%
Age of onset
Gender ratio
(male:female)
Average % of 1st
degree relatives
who are affected
Near speech onset
3:1-9:1
Near speech onset
1.5:1-3:1
Unknown, but may
be higher than speech
delay subgroup
20-41%
1.0% in older children
and adults; 5.0% in
preschool children
3-6 years
2:1 (preschool) 4:1
(school-aged +)
15-19%
332
S. Felsenfeld/Journal of Communication Disorders 35 (2002) 329-345
Several epidemiological observations about DSDs are worthy of comment.2
Disorder prevalence, for example, appears to vary by disorder. Of the three speech
phenotypes, suspected DAS has the lowest population prevalence rate, affecting
only about 1 or 2 children per 1000 (Shriberg, Austin, Lewis, McSweeny, &
Wilson, 1997). The speech delay subgroup, on the other hand, is considerably
more prevalent in the population, affecting perhaps as many as 14 children per 100
at age 3 and 3^1 children per 1000 at age 6 (Shriberg et al., 1999). As with speech
delay, stuttering prevalence varies as a function of age. Among preschool
children, stuttering affects approximately 5 children per 100 (Yairi et al., 1996).
By the end of the elementary school years, this prevalence rate drops to
approximately one affected individual per hundred, and stabilizes at this rate
(1%) across the remainder of the lifespan. This suggests that many preschool
children diagnosed with either stuttering or speech delay will normalize their
speech within a few years of onset, either with or without formal attempts at
intervention. This epidemiological phenomenon is of interest to geneticists
because it is possible that these outcome differences (early recovery versus nonrecovery) reflect different genetic profiles that are worthy of investigation.
All three developmental speech phenotypes manifest themselves early in the
lifespan. Theoretically, articulation/phonological disorders are detectable at or
near the time of the onset of connected speech, although formal diagnosis may not
occur until later. Stuttering also has an early though slightly more variable onset
window. At least 75% of stuttering cases report stuttering onset between 3 and 6
years, with essentially no new cases emerging after age 12 (Bloodstein, 1995).
The fact that DSDs have an almost universally early age of onset is useful for
genetics research, because it means that affected cases in families and population
samples can be identified early. That is, unlike other complex disorders whose
onset can be late in the lifespan, it is not necessary to wait until an individual
reaches adulthood to determine if he or she will develop the condition of interest.
In terms of gender, all three of the speech phenotypes affect more males than
females, although the gender ratio varies by disorder type. Suspected DAS has the
least symmetric ratio, with reports ranging from three affected males for every
affected female to as high as nine affected males for every affected female
(Shriberg et al., 1997). Interestingly, current epidemiological data for the related
articulation/phonology phenotype, speech delay, paint a quite different gender
picture. Gender ratios for the speech delay phenotype range in the literature from
only 1.5:1 at age 6 for a quasi-population sample (Shriberg et al., 1999) to 3:1 for
children of more variable ages who attended a University clinic (Shriberg &
Kwiatkowski, 1994). Stuttering also occurs more frequently among males than
females, although this ratio varies with age. Among preschool stutterers, boys are
about twice as likely to stutter as are girls. Beginning at about age 9, this ratio
2
For purposes of illustration, speech delay and DAS are considered separate articulation disorder
phenotypes in Table 1. This subgroup classification should be regarded as provisional.
S. Felsenfeld/Journal of Communication Disorders 35 (2002) 329-345
333
increases to about four affected males for every affected female. This phenomenon has been interpreted as suggesting that young girls are more likely to recover
from stuttering than are boys, for reasons that are still not fully understood (Yairi
& Ambrose, 1999).
One final epidemiological observation is of particular interest to genetics
researchers: the presence of familial aggregation. To determine if familial
aggregation is present, the status of family members of known affected cases
(probands) is compared with the status of control families for a condition of
interest. Geneticists look to see if the distribution of affected cases is significantly
higher among the biologically-related family members of proband subjects than
among the relatives of control cases or to population rates for a given disorder. If
so, then genetic factors are implicated in that condition's etiology. Published data
exist that document the occurrence of familial aggregation for both speech delay
(Felsenfeld, McGue, & Broen, 1995; Lewis, Cox, & Byard, 1993; Lewis, Ekelman, & Aram, 1989) and stuttering (Ambrose, Yairi, & Cox, 1993; Cox, Kramer,
& Kidd, 1984). Less is known about the aggregation of suspected DAS, although
there is some preliminary evidence that suggests that this phenotype may be more
highly familial than either speech delay or stuttering. As can be seen in Table 1,
the percentage of first-degree relatives that are affected varies between the speech
delay and stuttering phenotypes. In general, more relatives are affected when the
proband presents with speech delay (20^10%) than when the proband presents
with stuttering (15-19%). However, even for stuttering, these affection rates for
relatives far exceed the affection rates found in control families and in population
samples (less than or equal to 5%).
2. Evidence stream for genetic effects
These familial aggregation findings provide one line of evidence implicating
genetic factors in the etiology of DSDs. There are actually four descriptive
(behavioral) genetic methodologies that provide evidence to either support or
refute genetic hypotheses: case studies, with or without published pedigrees;
familial aggregation studies, with or without formal segregation analyses; twin
studies, with or without formal statistical (ACE) modeling; and full adoption
studies. All of these methodologies have been used to examine DSDs, although
the number of studies is quite small relative to other disciplines. Table 2 provides
annotated references for the primary behavioral genetic studies of articulation/
phonological disorders published over the last two decades. Table 3 provides the
same information for stuttering. For more detailed coverage of these and related
epidemiological topics, readers are referred to Felsenfeld and Drayna, 2001;
Shriberg et al., 1999; Yairi et al., 1996.
To summarize, results of behavioral genetic studies performed over the past
two decades have uniformly implicated genetic factors in the etiology of both
articulation/phonology disorders and developmental stuttering. For both of these
334
S. Fehenfeld/Journal of Communication Disorders 35 (2002) 329-345
Table 2
Annotated references for behavior genetic studies of
Reference articulation/phonological disorders
Methodology
Notes
Pedigrees
Hurst et al. (1990)
Gopnik (1990)
Vargha-Khadem et
Family studies
al. (1995) Lewis et
al. (1989)
Lewis et al. (1993)
Felsenfeld et
al. (1995)
Twin studies Lewis and Thompson
(1992)
Full adoption
study
KE family
KE family
KE family
Families of proband children had more speech, language,
and reading problems than did families of control cases
Formal segregation analysis performed; both major gene
and polygenic models fit the data Children of proband
subjects performed more poorly than children of controls
on all expressive language and articulation measures
Identical (MZ) twins were concordant for presence of
speech disorder significantly more often than were
fraternal (DZ) twins
Children with a positive biological (parental) background
of speech disorder were at an increased risk for displaying
speech disorders themselves, despite being adopted away
from their affected parent(s) at or near birth
Felsenfeld and
Plomin (1997)
phenotypes, elevated rates of speech disorder were found among the biological
relatives of proband subjects. Those family studies that modeled family pedigree
results using segregation analysis procedures were either inconclusive or tended
to favor a polygenic (multiple gene) transmission model, although a major
Table 3
Annotated references for behavior genetic studies of stuttering
Methodology
Reference
Notes
Pedigree
MacFarlane, Hanson, Walton,
and Mellon (1991) Cox et al.
(1984)
Kindred #1638 in Utah-Idaho area
Family studies
Ambrose et al. (1993)
Twin studies
Howie (1981) Andrews
et al. (1991) Felsenfeld
et al. (2000)
Segregation analysis was performed; polygenic
model provided best fit
Segregation analysis performed using relatives
of preschool probands; evidence found for
single major locus
MZ twins were concordant for presence of
stuttering significantly more often than were
DZ twins
Statistical (ACE) modeling was performed;
heritability of stuttering was found to be 0.71 in
large population sample of adult twins
Statistical (ACE) modeling was performed;
heritability of stuttering was found to be 0.70 in
large population sample of adult twins
S. Felsenfeld/Journal of Communication Disorders 35 (2002) 329-345
335
Mendelian locus could not be ruled out for all subgroups tested (e.g., Ambrose
et al., 1993). Results of the twin studies that have examined these phenotypes have
confirmed the family study findings. For both articulation/phonology disorder and
stuttering, identical (monozygotic) twins were found to be concordant for the
presence of speech disorder significantly more often than were fraternal (dizygotic) twins, a finding which is highly suggestive of genetic influence. For
stuttering, statistical (ACE) modeling results performed using two large adult
samples suggested that about 70% of the variance in liability to stuttering could be
attributed to additive genetic effects.
3. Challenges that may impede our forward progress on the road
Because a small but consistent group of behavioral genetic studies support the
hypothesis that genes contribute significantly to the liability of both articulation/
phonology disorders and stuttering, the hunt for susceptibility genes for these
disorders can rightfully commence. However, since we are new travelers on the
genetics road, it may be worthwhile to review some of the challenges that we are
likely to face as we take our first tentative steps on those little yellow bricks. The
third objective of this paper is to illustrate some of the variables that are known
to complicate the molecular study of complex disorders, including
developmental disorders of speech.
3.1. Absence of diagnostic gold standards
Although there are a number of clinical tools and guidelines available to
diagnose speech disorders for clinical purposes, there is no universally accepted
gold standard for classifying subjects as speech-affected for genetics research.
Because of this, different investigators continue to adopt different criteria for
determining affected status. Some investigators, for example, allow informant
report to substitute for direct testing or interview of subjects. In some cases,
inclusion criteria are very broad (e.g., evidence for a reading disorder alone may
classify a potential subject as speech-affected). The problem with this, of course,
is that our classifications should mirror biological reality as closely as possible.
For genetics studies, we want to identify as affected only those subjects who
truly possess the genetic liability we are investigating. If our criteria are careless
or incorrect, our progress at the molecular level will be compromised.
3.2. Variable expression, which may reflect genetic heterogeneity
As is the case for many complex neurodevelopmental disorders (e.g., autism,
dyslexia, schizophrenia), DSDs have a complicated clinical presentation. For
example, individuals may present with very mild or sub-clinical variants of a
336
S. Felsenfeld/Journal of Communication Disorders 35 (2002) 329-345
disorder. Should they be considered affected or unaffected with that disorder in a genetics
study? Both speech delay and stuttering are known to have high recovery rates in
childhood. If an individual clearly manifested the disorder as a child but no longer shows
signs of the disorder, is he or she still to be considered affected, from a genetic
perspective? Should only individuals who manifest a "pure" form of the disorder of
interest be included in a genetics subject pool, or is it appropriate to also include
individuals with mixed (comorbid) diagnoses? At the present time, there are no ready
answers for these questions. In general, geneticists prefer to deal first with clear and
uncomplicated cases: those with persistent, pure, and severe forms of the phenotype.
However, for practical reasons, often related to sample size, it is sometimes impractical
to include only these ideal subjects in genetics study. As a compromise, many geneticists
advise that samples from these "less optimal" cases be collected, but coded differently for
potential subgrouping analysis.
3.3. Absence of accepted analysis strategies for finding multiple
susceptibility genes
Molecular biologists have mastered the analysis techniques needed to understand
Mendelian disorders (i.e., disorders that result from one mutation that is transmitted in an
autosomal dominant, recessive, or sex-linked fashion). For these disorders, standard
linkage and association studies are appropriate and are generally successful. For
polygenic disorders, the molecular analysis picture is far less straightforward. In fact,
although a number of exploratory techniques are being tried, there are currently no
accepted molecular analysis strategies for identifying multiple susceptibility loci that
interact to produce a pathological condition. Without a doubt, such analysis techniques
will appear, perhaps in the very near future. Until that time, progress in identifying
multiple loci for polygenic conditions, including disorders of speech, will lag behind the
extraordinary progress that has been seen for Mendelian diseases.
3.4. Probable (future) replication problems
It is likely that DSDs will experience the same fate of many other neurodevelopmental disorders: initially positive genetic findings will fail to replicate. It is
unclear why the failure to replicate has been such a pervasive problem in the genetics of
complex disorders, and a full exploration of this complex topic is beyond the scope of
this paper (see Suarez, Hampe, & Van Eerdewegh, 1994 for a discussion of these issues).
However, to illustrate the dilemma, the results from another representative disorder—
bipolar affective disorder (BPAD)—will be described. BPAD (also called manicdepressive psychosis) is a complex disorder of mood and cognition. Behavioral genetic
studies of BPAD have consistently demonstrated moderately high heritability for this
disorder, suggesting that genes play a significant role in its etiology. The first genetic
linkage study of BPAD
S. Felsenfeld/Journal of Communication Disorders 35 (2002) 329-345
337
appeared in the literature in 1972. Between that time and 2001, approximately 13
molecular studies of BPAD have been published that have reported positive
linkage results (i.e., LOD scores over 3.0). What is particularly interesting is that
essentially no two studies have implicated the same genetic "hot spots" for this
disorder. Rather, positive linkage has been identified in multiple places throughout the genome, including loci on chromosomes 4, 11, 12, 13, 18, 20, 21 and X.
Thus, despite a large number of initially promising genetic "hits" for BPAD,
there is still no confirmed major locus or set of loci for the general form of the
disorder, and several initially promising loci are no longer considered likely
candidates. Perhaps most discouraging, none of these positive genetic findings
have yet proven to be directly useful for the diagnosis or treatment of BPAD
patients. Although we cannot be certain that DSDs will follow this same winding
path, we should be cautious about celebrating early linkage reports in our
discipline. Linkage findings may be specific to a group or family that was
included in a particular study and may not generalize to other subgroups or
samples. Conversely, there may be a complete failure to replicate initial findings
that will result in the appearance and then disappearance of initially promising
speech disorder loci.
4. Molecular genetic studies of DSDs
Despite these and other challenges, the molecular genetic study of DSDs has
begun. We are on the road! At the time of this writing (late 2001), there has been
only one published genetic linkage study for a developmental speech (articulation/phonological) disorder, described below. This discovery was accompanied by
some fanfare, including a write-up in the New York Times. For stuttering, there
are at least two large genetic linkage studies in progress and more in the planning
stages. However, no genetic linkage findings relating to stuttering have been
published to date.
The methodological steps that led to the identification of the first candidate
"speech gene" are outlined below. These steps are worth reviewing, because they
reflect the current standard sequence for finding candidate genes for any disorder.
The "speech gene" discovery was initially prompted by the publication of three
increasingly detailed case reports of one family in England, the KE family
(Gopnik, 1990; Hurst, Baraitser, Auger, Graham, & Norell, 1990; VarghaKhadem, Watkins, Alcock, Fletcher, & Passingham, 1995). The KE family was
of special interest because it contained a large number of members across
generations who were affected with a severe and seemingly isolated speech and
language disorder. The case reports described the phenotype of the affected KE
family members, noting the presence of a rather diffuse set of communication
problems, including dyspraxia (suspected DAS), expressive and receptive language problems, and, in some cases, mild intellectual impairment. A visual
inspection of the pedigree prompted some of the investigators to conclude that the
338
S. Felsenfeld/Journal of Communication Disorders 35 (2002) 329-345
transmission of the condition in this family might be autosomal dominant
(Gopnik, 1990; Hurst et al., 1990).
Following this descriptive work, a team of investigators began the genefinding process by first performing a genome-wide scan of both affected and
(some) unaffected members of the KE family (Fisher, Vargha-Khadem,
Watkins, Monaco, & Pembrey, 1998). This scan was performed to look for
genetic linkage, a statistical association between a particular genetic configuration and the presence of the disorder phenotype. One candidate region
appeared promising in this linkage analysis, a region on chromosome 7 (7q31),
which the investigators named SPCH1. This positive step prompted more
refined analysis of the relatively small SPCH1 region. Using sophisticated
bioinformatics approaches, a "fine map" of this region was constructed in a
subsequent investigation (Lai et al., 2000). In addition, an unrelated case (not a
member of the KE family) that presented with a similar phenotype was
identified and analyzed genetically (Lai et al., 2000). This unrelated case also
displayed an abnormality in SPCH1, specifically, a chromosomal translocation
that disrupted that region. This replication was considered significant because it
suggested that the SPCH1 region was associated with the expression of a severe
speech and language disorder phenotype beyond the (perhaps unique) KE
family.
As a final step in the analysis process, a point mutation (i.e., a gene) in the
mapped region was identified and analyzed (Lai, Fisher, Hurst, Vargha-Khadem,
& Monaco, 2001). This gene, called FOXP2, was found to be a member of the
FOX family of genes that are known to be "key regulators of embryogenesis"
(Lai et al., 2001). Thus, the current speculation is that a programmed disruption
occurs in this gene sometime during embryonic development in affected cases. As
a result of this disruption, areas of the developing brain that are responsible for
speech and language functions are negatively influenced, resulting in difficulty in
the processing and/or production of speech (Lai et al., 2001).
Although the FOXP2 discovery is clearly an important one, some words of
caution are in order. Given the severe and diffuse nature of the phenotype in the
KE family, it is quite possible that this mutation reflects a rare Mendelian
syndrome in which poor speech is a primary clinical feature. If so, then these
positive findings may not generalize to "garden-variety" affected cases in either
the DAS or speech delay subgroups. Some preliminary evidence to support this
interpretation was reported very recently by Schick et al. (2001). In their
investigation, a linkage analysis was performed to replicate the findings of Lai
et al. (2001) using a broader sample of speech- and/or language-affected preschool probands and their family members. Preliminary results of this study
suggested that the evidence for linkage between the speech delay phenotype and
the markers examined in the 7q31 region was "weak." However, further
molecular analyses are being performed, and the authors caution that the
generally negative findings that they reported may be subject to modification as
additional data emerge.
S. Felsenfeld/Journal of Communication Disorders 35 (2002) 329-345
339
5. Strategies for increasing the probability of success
Investigators across disciplines are striving to find strategies to assist molecular
geneticists in their task of finding susceptibility genes for complex disorders. If
the search area can be effectively narrowed or if genetic background noise can be
reduced, then the probability of a positive genetic hit increases. As a final
objective, I will present an overview of two general research strategies that may
be advantageous in our quest to find susceptibility genes for speech disorders: (a)
identifying diverse groups to study at a molecular level, and (b) reducing
etiologic heterogeneity by forming endophenotypic subgroups.
5.1. Identifying diverse groups to study at a molecular level
In general, it is useful to collect genetic data from several different population
samples when performing genetic studies for an understudied phenotype. For
example, there are well-documented advantages to examining both out-bred and
homogeneous samples in initial linkage studies (Shifman & Darvasi, 2001; Suarez
et al., 1994; Wright, Carothers, & Piratsu, 1999). In out-bred sampling, affected
cases are ascertained from the population at large through general advertisement,
at clinical sites, or through large epidemiological projects. Out-bred samples by
definition are more genetically heterogeneous, and this can provide an advantage
in terms of the potential generalizability of genetic findings. On the other hand,
homogenous samples—such as those found in geographically isolated communities—can be advantageous in the early stages of gene-hunting because these
samples reduce unwanted background "noise" that can sometimes obscure
genetic variants of interest.
Similarly, as was demonstrated by the KE family, high-density pedigrees can
also provide a useful starting point for linkage analyses. Highly dense families
may present with a rare but highly penetrant form of the disorder that can help to
focus attention on promising candidate regions in the genome. As discussed
previously, positive linkages that are found within such families may not generalize to all cases that present with a similar phenotype, but such results can provide a
starting point for gene mapping that improves upon current "shot-gunapproaches.
Finally, it may be advantageous to narrow the initial genetic search by
examining known Mendelian disorders where speech problems present as a
primary clinical symptom. It is possible that the genetic regions that are known to
be mutated in these disorders may overlap with regions that influence a similar
phenotype in more general population samples. One problem in adopting this
approach is that syndromes that present with speech problems as a prominent
feature are often those where there are numerous organic contributory conditions,
such as moderate to severe mental retardation, sensorineural hearing loss, or
severe craniofacial anomalies. These types of syndromes are less useful to
investigators looking for speech disorder loci because of their non-specific
340
S. Felsenfeld/Journal of Communication Disorders 35 (2002) 329-345
Table 4
Mendelian disorders with speech deficits as a primary clinical feature
Disorder
Gene location
Velocardiofacial syndrome
22q 11
Prader-Willi syndrome
15ql2
MRX14 syndrome
Xpll
Developmental verbal
dyspraxia syndrome
(KE family)
7q31
Speech-related features
Hypernasal resonance; poor articulation; mild
learning and language deficits Articulation
deficits; possible increased rate of stuttering;
mild-moderate cognitive deficits Moderate
mental retardation with disproportionately
poor speech intelligibility Suspected DAS;
oral apraxia; expressive language deficits;
borderline normal to mildly impaired
cognitive abilities
nature.Table 4 displays the results of an informal search of the OMIM (Online
Mendelian Inheritance in Man) database.3 This source was searched to identify
syndromes for which stuttering, or speech disorder were considered primary
clinical features. Only those syndromes with established genetic loci were
included. Clearly organically-based syndromes were excluded from the table,
using the general criteria described above. As can be seen, the number of even
marginally appropriate syndromes that remain when these criteria are applied
declines to almost zero. Moreover, and more importantly, the syndromes that
remain reflect mutations at several different loci. Thus, although theoretically
promising, this analysis strategy does not appear useful on the surface. However,
in the absence of better information, the loci identified in Table 4 may be worthy
of some increased scrutiny in genome-wide scans of DSDs.
5.2. Reducing etiologic heterogeneity by forming endophenotypic subgroups
An alternative approach to reducing subject heterogeneity for genetic analyses
is to subdivide large groups of affected cases into smaller subgroups that are
presumed to be more genetically homogeneous on the basis of some classification
system. These classifying or grouping variables are sometimes referred to as
"micro" or endophenotypes. More precisely, endophenotypes are typically
defined as neurobiological or neurobehavioral correlates of a disorder that are
close to or directly linked to gene expression (Callicott et al., 1998; Goldsmith &
Lemery, 2000). The presumption, sometimes difficult to test empirically, is that
subjects within a well-conceived endophenotypic subgroup are more likely to
manifest "genetic profiles" that are similar to one another than are randomly
collected individuals who share a broad phenotypic label, such as stuttering. The
first step in this analysis process is to form these subgroups properly, not an easy
task itself. Once this is done, geneticists can analyze the subgroups separately to
determine if each subgroup does in fact display unique mutations.
The website for OMIM is http://www.ncbi.nlm.nih.gov/Omim/searchomim.html
S. Felsenfeld/Journal of Communication Disorders 35 (2002) 329-345
341
For disorders of speech, at least three endophenotypic subgrouping systems are
possible. First, subgroups can be formed based on the presence or absence of a
specific clinical symptom or outcome {phenotype-based subgrouping). Shriberg
and his colleagues present a good example of phenotype-based subgrouping when
they propose using the clinical symptom of inappropriate stress as a marker for
suspected DAS (Shriberg, Aram, & Kwiatkowski, 1997b). For stuttering, overt
behaviors (such as the presence or absence of speech blocks) could theoretically
be used as a phenotype-based grouping variable. More promising, however, is a
subgrouping system that is based upon recovery status; that is, whether an
individual recovers from stuttering prior to middle childhood or persists in
stuttering beyond this point. The hypothesis in this case would be that individuals
who recover from stuttering and those who persist will differ in their underlying
genetic make-up (see Yairi & Ambrose, 1999).
A second strategy for forming endophenotypic subgroups is to use outcomes
on psychophysical tasks as the grouping variable (performance-based subgrouping). In this system, individuals who share a diagnostic label (e.g., as articulation
disordered) are subdivided on the basis of their performance on some task or set of
tasks that are presumed to tap into a processing system that underlies the disorder.
For example, it may be possible to subgroup stuttering individuals into a group of
good versus poor performers on tasks involving sequential finger tapping, manual
crank turning, dichotic listening, or vocal reaction time. For articulation/phonological disorders, a similar procedure could be used to divide affected cases on the
basis of performance on phonological processing tasks such as nonsense-word
repetition, word segmentation, or phoneme decoding efficiency (see Shriberg et
al., 1999).
The third strategy for forming subgroups is to use the presence or absence of
neuroanatomic or other physiological markers as the grouping variable (biologically-based subgrouping). To do this, candidate physiological measures that are
associated with the disorder phenotype must be identified, and subjects must be
assessed to determine their status on that variable. Candidate physiological
measures might include evoked potentials, hormone or neurotransmitter levels,
or neuroanatomic and neurophysiological regions of interest in brain. Because of
their presumed proximity to the core biological deficit(s), biologically-based
grouping variables may prove to be the most promising of the three systems.
Although well-established biological markers are presently lacking for developmental disorders of speech and language, it is likely that these markers will
emerge. For example, brain imaging (PET and MRI) studies of speech and
language disordered probands have already been successful in identifying candidate regions of interest that distinguish affected from unaffected cases (cf.,
DeNil, Kroll, Kapur, & Houle, 2000; Fox et al., 1996; Plante, Swisher, Vance, &
Rapcsak, 1991). In the near future, this information might be used to form
endophenotypic subgroups-based, for example, upon the presence or absence of
an anomaly in a specific brain region—that will help us find those elusive
emeralds at the end of the road.
342
S. Felsenfeld/Journal of Communication Disorders 35 (2002) 329-345
6. Summary
Compared to other clinical disciplines, the molecular genetic study of DSDs
has been slow to develop, despite evidence from pedigree, family, twin, and
adoption studies that have uniformly implicated genetic factors in the etiology of
both articulation/phonology disorders and developmental stuttering. A number of
conceptual and methodological challenges are likely to face geneticists as they
look for susceptibility genes for speech disorders. These include the absence of
diagnostic gold standards for these conditions, variable expression of the phenotypes, the absence of accepted analysis strategies for finding multiple susceptibility genes, and possible (future) replication problems. Two general research
strategies were proposed to reduce heterogeneity for molecular genetic analyses:
(a) identifying diverse groups to study, and (b) reducing etiologic heterogeneity by
forming endophenotypic subgroups. It is argued that progress in the molecular
study of speech disorders will be enhanced if theories about biological systems
and processes can be used to narrow the search for candidate susceptibility genes.
Appendix A. Continuing education
Finding susceptibility genes for developmental disorders of speech: the long
and winding road
1. Which of the following best summarizes what is known about the gender
distribution of the three DSDs that were discussed in the paper?
a. Males and females are equally likely to be affected with a DSD during
the preschool years, but after this, more males are likely to be affected
than females (with ratios ranging from 2:1 to 4:1)
b. More males than females are likely to be affected with a DSD at every
age, although there is a trend for the male:female ratio to increase
(become more pronounced) with subject age
c. More females than males are affected with stuttering in the preschool
years; for all other disorders, males are more likely to be affected with a
DSD at every age
d. No consistent gender pattern can be identified for DSDs; the gender
ratios for each of the three disorders are distinctly different
2. Which of the following interpretations is not supported by the findings of
the behavioral genetic studies of DSDs reviewed in this paper?
a. Families containing a proband (a speech-affected member) are more
likely to have other biological relatives who are speech-affected than
are families of control cases
b. If one member of an identical (MZ) twin pair has a speech disorder, his
or her cotwin is also more likely to have a speech disorder than is the
cotwin of a fraternal (DZ) twin who is speech-affected
S. Felsenfeld/Journal of Communication Disorders 35 (2002) 329-345
343
c. Having a biological parent with a positive history of speech disorder
may increase the risk for speech problems in children who were
adopted away at or near birth
d. Familial aggregation has been well-established for articulation/
phonological disorders, but not for stuttering, which shows no such
aggregation
3. All of the following have been identified as potential challenges that may
complicate the molecular genetic study of DSDs except
a. Difficulty finding a sufficient number of affected cases in the
population who will agree to participate in genetics research
b. Difficulty that may occur in replicating initially promising genetic
findings
c. Difficulty finding universally recognized diagnostic criteria that reflect
biological reality
d. Difficulty in determining how to manage the complicated clinical
presentation of the disorder phenotypes
4. Which of the following best illustrates the methodological sequence
that was followed in the discovery of the FOXP2 "speech disorder"
gene?
a. A genome-wide scan was performed to identify linkage; the SPCH1
region was fine mapped; the KE pedigree was described in detail; the
mutated gene (FOXP2) was identified and analyzed
b. The SPCH1 region was fine mapped; a genome-wide scan was
performed to identify linkage; the mutated gene (FOXP2) was
identified and analyzed; the KE pedigree was described in detail
c. The KE pedigree was described in detail; a genome-wide scan was
performed to identify linkage; the SPCH1 region was fine mapped; the
mutated gene (FOXP2) was identified and analyzed
d. The KE pedigree was described in detail; the SPCH1 region was fine
mapped; a genome-wide scan was performed to identify linkage; the
mutated gene (FOXP2) was identified and analyzed
5. Which of the following will probably not be a valid system for forming
homogeneous endophenotypic subgroups for future molecular genetic
studies of speech disorders?
a. Subgrouping based upon some clinical symptom or outcome of the
disorder
b. Subgrouping based upon a score on a widely used norm-referenced
diagnostic test
c. Subgrouping based upon performance on a carefully selected task that
is presumed to tap into a processing system that underlies the disorder
d. Subgrouping based upon a biological marker, such as a specific
neuroanatomic region
344
S. Felsenfeld/Journal of Communication Disorders 35 (2002) 329-345
References
Ambrose, N. G., Yairi, E., & Cox, N. J. (1993). Genetic aspects of early child stuttering. Journal of
Speech and Hearing Research, 36, 701-706. Andrews, G., Morris-Yates, A., Howie, P., &
Martin, N. (1991). Genetic factors in stuttering
confirmed. Archives of General Psychiatry, 48, 1034-1035. Bloodstein, O. (1995). A handbook
on stuttering (5th ed.). Chicago, IL: National Easter Seal Society. Callicott, J. H., Egan, M. E,
Bertolino, A., Mattay, V. S., Langheim, E J., Frank, J. A., & Weinberger,
D. R. (1998). Hippocampal JV-acetyl aspartate in unaffected siblings of patients with
schizophrenia: A possible intermediate neurobiological phenotype. Biological Psychiatry, 44,
941-950. Cox, N. J., Kramer, P. L., & Kidd, K. K. (1984). Segregation analysis of stuttering.
Genetic
Epidemiology, 1, 245-253. DeNil, L. E, Kroll, R. M., Kapur, S., & Houle, S. (2000). A positron
emission tomography study of
silent and oral single word reading in stuttering and nonstuttering adults. Journal of Speech,
Language & Hearing Research, 43, 1038-1053. Felsenfeld, S. (1996). Progress and needs in the
genetics of stuttering. Journal of Fluency Disorders,
21, 77-103. Felsenfeld, S., & Drayna, D. (2001). Stuttering and genetics: Our past and our future.
In S. E. Gerber
(Ed.), The handbook of genetic communicative disorders (pp. 151-174). New York: Academic
Press. Felsenfeld, S., Kirk, K. M., Zhu, G., Statham, D. J., Neale, M. C, & Martin, N. G. (2000). A
study of
the genetic and environmental etiology of stuttering in a selected twin sample. Behavior Genetics,
30, 359-366. Felsenfeld, S., McGue, M., & Broen, P. A. (1995). Familial aggregation of
phonological disorders:
Results from a 28-year follow-up. Journal of Speech and Hearing Research, 38, 1091-1107.
Felsenfeld, S., & Plomin, R. (1997). Epidemiological and offspring analyses of DSDs using data
from the Colorado Adoption Project. Journal of Speech, Language, & Hearing Research, 40,
778-791. Fisher, S. E., Vargha-Khadem, E, Watkins, K. E., Monaco, A. P., & Pembrey, M. E.
(1998).
Localization of a gene implicated in a severe speech and language disorder. Nature Genetics, 18,
168-170. Fox, P. T., Ingham, R. J., Ingham, J. C, Hirsch, T. B., Downs, J. H., Martin, C, Jerabek,
P., Glass, T.,
& Lancaster, J. L. (1996). A PET study of the neural systems of stuttering. Nature, 382, 158-162.
Goldsmith, H. H., & Lemery, K. S. (2000). Linking temperamental fearfulness and anxiety
symptoms: A behavior-genetic perspective. Biological Psychiatry, 48, 1199-1209. Gopnik, M.
(1990). Feature blind grammar and dysphasia. Nature, 344, 715. Howie, P. M. (1981). Concordance
for stuttering in monozygotic and dizygotic twin pairs. Journal of
Speech and Hearing Research, 24, 317-321. Hurst, J. A., Baraitser, M., Auger, E., Graham, E, &
Norell, S. (1990). An extended family with a
dominantly inherited speech disorder. Developmental Medicine and Child Neurology, 32,
347-355. Lai, C. S. L., Fisher, S. E., Hurst, J. A., Levy, E. R., Hodgson, S., Fox, M., Jeremiah,
S., Povey, S.,
Jamison, D. C, Green, E. D., Vargha-Khadem, E, & Monaco, A. P. (2000). The SPCH1 region on
human 7q31: Genomic characterization of the critical interval and localization of translocations
associated with speech and language disorder. American Journal of Human Genetics, 67, 357-368.
Lai, C. S. L., Fisher, S. E., Hurstz, J. A., Vargha-Khadem, E, & Monaco, A. P. (2001). A forkheaddomain gene is mutated in a severe speech and language disorder. Nature, 413, 519-523. Lewis,
B. A., Cox, N. J., & Byard, P. J. (1993). Segregation analysis of speech and language
disorders. Behavior Genetics, 23, 291-297. Lewis, B. A., Ekelman, B. L., & Aram, D. M. (1989).
A familial study of severe phonological
disorders. Journal of Speech and Hearing Research, 32, 713-724.
S. Felsenfeld/Journal of Communication Disorders 35 (2002) 329-345
345
Lewis, B. A., & Thompson, L. A. (1992). A study of developmental speech and language disorders in
twins. Journal of Speech and Hearing Research, 35, 1086-1094. MacFarlane, W. B., Hanson,
M., Walton, W., & Mellon, C. D. (1991). Stuttering in five generations
of a single family: A preliminary report including evidence supporting a sex-modified mode of
transmission. Journal of Fluency Disorders, 16, 117-123. Plante, E., Swisher, L., Vance, R., &
Rapcsak, S. (1991). MRI findings in boys with specific language
impairment. Brain and Language, 41, 52-66. Schick, S. L., Kundtz, A. M., Taylor, H. G.,
Shriberg, L. D., Freebairn, L. A., Lewis, B. A., &
Iyengar, S. K. (2001, October). Evidence of linkage for speech impairment with chromosome 7
markers. Poster presented at the annual meeting of the American Society of Human Genetics, San
Diego, CA. Shifman, S., & Darvasi, A. (2001). The value of isolated populations. Nature
Genetics, 28, 309-310. Shriberg, L. (1997). Developmental phonological disorder(s): One or many?
In B. W. Hodson & M.
L. Eswards (Eds.), Applied phonology: Perspectives and clinical applications. Gaithersburg, MD:
Aspen. Shriberg, L. D., Aram, D. M., & Kwiatkowski, J. (1997a). Developmental apraxia of
speech: I.
Descriptive and theoretical perspectives. Journal of Speech, Language, & Hearing Research, 40,
273-285. Shriberg, L. D., Aram, D. M., & Kwiatkowski, J. (1997b). Developmental apraxia of
speech: III. A
subtype marked by inappropriate stress. Journal of Speech, Language, & Hearing Research, 40,
313-337. Shriberg, L. D., Austin, D., Lewis, B., McSweeny, J. L., & Wilson, D. L. (1997). The
Speech
Disorders Classification System (SDCS): Extensions and lifespan reference data. Journal of
Speech, Language, & Hearing Research, 40, 123-140. Shriberg, L. D., & Kwiatkowski, J.
(1994). Developmental phonological disorders: I. A clinical
profile. Journal of Speech and Hearing Research, 37, 1100-1126. Shriberg, L. D., Tomblin, J. B.,
& McSweeny, J. L. (1999). Prevalence of speech delay in 6-year-old
children and comorbidity with language impairment. Journal of Speech, Language, & Hearing
Research, 42, 1461-1481. Suarez, B. K., Hampe, C. L., & Van Eerdewegh, P. (1994). Problems of
replicating linkage claims in
psychiatry. In E. S. Gershon & C. R. Cloninger (Eds.), Genetic approaches to mental disorders
(pp. 23-46). Washington, DC: American Psychiatric Press. Vargha-Khadem, E, Watkins, K.,
Alcock, K., Fletcher, P., & Passingham, R. (1995). Praxic and
nonverbal cognitive deficits in a large family with a genetically transmitted speech and language
disorder. Proceedings of the National Academy of Science USA, 92, 930-933. Wright, A. F,
Carothers, A. D., & Pirastu, M. (1999). Population choice in mapping genes for
complex diseases. Nature Genetics, 23, 397-404. Yairi, E., & Ambrose, N. (1999). Early
childhood stuttering: I. Persistence and recovery rates.
Journal of Speech, Language, & Hearing Research, 35, 755-760. Yairi, E., Ambrose, N., & Cox,
N. (1996). Genetics of stuttering: A critical review. Journal of Speech
and Hearing Research, 39, 111-184.