Discussion Brain imaging in stuttering: where next? Peter T. Fox

Transcription

1 Journal of Fluency Disorders 28 (2003) Discussion Brain imaging in stuttering: where next? Peter T. Fox Research Imaging Center, University of Texas Health Science Center at San Antonio, San Antonio, TX , USA Received 18 August 2003; accepted 21 August 2003 Keywords: Stuttering; Brain imaging; PET; TMS; fmri As the articles comprising this special issue amply demonstrate, brain functional imaging is having a significant impact on research in persistent developmental stuttering (PDS). Specifically, voxel-wise statistical parametric images (SPI) demonstrating differences in the brain activation patterns evoked by fluent versus stuttered speech are having an impact. The first such report (Fox et al., 1996), used positron emission tomography (PET) and overt paragraph reading to show that, in broad strokes, stuttering was characterized by overactivity of the right inferior premotor cortex (operculum and insula) and underactivity of auditory cortex, abnormalities which were remediated acutely by fluency induction (chorus reading). With the several papers in this issue (and numerous intervening papers), considerable consensus about these findings has emerged. The findings have been replicated with a range of immediately and temporarily effective fluency inductions as well as with sustained improved fluency produced by behavioral treatments. They have been extended from paragraph reading to spontaneous speech and to single word tasks. They have been extended from overt speech to imagined speech and speech preparation tasks. They have been extended from PET to functional magnetic resonance imaging (fmri). Despite significant methodological differences, the basic observations stand. Given this, the question becomes, Where next? How best should the PDS research community capitalize on the reliability of these findings? How do we use this new insight into stuttering? Do we try more and more task variants, to winnow our findings down to the finest, most replicable effects? Do we apply the Tel.: ; fax: address: fox@uthscsa.edu (P.T. Fox) X/$ see front matter 2003 Elsevier Inc. All rights reserved. doi: /j.jfludis

2 266 P.T. Fox / Journal of Fluency Disorders 28 (2003) latest advances in functional imaging to improve our signal to noise? Do we debate endlessly as to which of the observed effects are causal and which compensatory? Do we attempt to shoehorn the imaging data into one or another time-honored theories of stuttering? I would suggest that while each of these activities will make some slow, steady progress, none would advance our understanding of this disorder in a fundamental way. What is needed is a new view, a quantum leap. 1. Structural imaging One way for functional imaging to advance would be to discover relationships between abnormalities of functional organization and developmental abnormalities of brain structure. Clear demonstration of structural brain abnormalities particularly if confined to the left hemisphere would suggest that the rightlateralized brain activation observed with functional imaging in stuttering might best be interpreted as developmental plasticity. Functional imaging has already demonstrated inter-hemispheric transfer of functions following acquired lesions (Weiller, Chollet, Friston, Wise, & Frackowiak, 1992; Weiller, Ramsay, Wise, Friston, & Frackowiak, 1993). PDS might well be another example. Whether developmental stuttering is characterized by structural brain abnormalities remains unclear. Abnormalities of gyral anatomy in the left and right frontal operculum have been reported (Foundas, Bollich, Corey, Hurley, & Heilman, 2001) but not yet independently confirmed. If aberrant gyral patterns are present, their functional significance will need careful consideration. While neuronal migration failures and polymicrogyria can be detected by MRI and are strongly associated with abnormalities of function, these have not been reported in developmental stuttering. The physiological significance of variations in surface folding patterns, if present, is unclear. Folding patterns of secondary and tertiary sulci are highly variable (Ono, Kubik, & Abernathy, 1990) and this variability seems to have little or no predictive value regarding functional organization (Hasnain, Fox, & Woldorff, 2001). On the other hand, differences in cortical thickness and total cortical volume are heritable (Thompson et al., 2001) and, if present in stuttering, would be a stronger indication of functionally significant pathology than would gyral folding patterns. Detection of these more subtle, but likely more significant anomalies, is best performed using fully automated methods for analysis of structural images, such as deformation-field morphometry (Lancaster et al., 2003). In a recent study from our group, deformation-field morphometry was applied to magnetic resonance images to detect differences in cortical thickness between natively English-speaking Caucasians and natively Chinese-speaking Asians (Kochunov et al., 2003). The cortical differences between these two groups were limited to specific loci in the frontal, temporal and parietal lobes that were known through functional imaging studies to differentiate Chinese speakers from English speakers. We interpret these anatomical differences as evidence of neural plasticity shaped by the process of language acquisition during childhood. A deformation-field morphometric study

3 P.T. Fox / Journal of Fluency Disorders 28 (2003) comparing persons who do stutter with persons who do not is presently underway in our laboratory and should give a definitive answer to the question of whether cortical anatomical abnormalities exist and, if present, how their locations relate to functional abnormalities. Abnormalities of myelin structure have been found in the left rolandic operculum of persons with PDS (Sommer, Koch, Paulus, Weiller, & Büchel, 2002). The functional interpretation of this finding appears more straightforward: a disconnection of superior temporal and inferior frontal language regions of the left hemisphere underlies stuttering. Disordered inter-regional connectivity in stuttering has also been suggested by magnetoencephalographic studies (Salmelin et al., 1998) showing that the temporal sequence of speech-associated activation is abnormal in stuttering speakers. In regards to the functional imaging evidence of excess right inferior premotor activation, this might well indicate that a compensatory shift of language processes into the non-dominant hemisphere has occurred. A fundamental difficulty for accepting a causal link between structural and functional abnormalities, however, is the normalization of functional activation patterns seen during fluency induction and following treatment. If function has migrated away from a structural lesion, why would function shift back into a structurally abnormal region? If the shift to right-hemispheric processing evolves over years during development, how can it be so readily reversed? Taken more generally, congruence between functional and structural observations may be more difficult to achieve than one might naively presume. 2. Epidemiology and genetics It is now well established that stuttering is a heritable disorder (Ambrose, Cox, & Yairi, 1997). Another potential avenue for brain imaging to advance our understanding of stuttering would be to use functional (and structural) imaging characteristics as additional phenotypes to facilitate pedigree analyses and linkage mapping. Functional imaging patterns during memory tasks have been demonstrated to be reliably associated with the APOE-4 gene, which transmits risk for Alzheimer s disease (Bookheimer et al., 2000). Structural abnormalities in speech-motor cortex and subcortical nuclei have been demonstrated to be reliably associated with the FOXP2 gene, which transmits risk for speech dyspraxia (Watkins et al., 2002). If the abnormal patterns of brain activation which are now reliably detected in group-average images of persons with PDS can be quantified on a per-subject basis, and if they are present in lesser degrees in non-stuttering relatives of stuttering probands, they could prove to be a substantial aid in disambiguating the inheritance pattern of stuttering and, through DNA linkage analyses, in identifying candidate loci for in-depth study. However, even if (when) the genetics underlying developmental stuttering are elucidated, there is an additional level of explanation that must be addressed. How do the genetic traits result in the behavior we recognize as stuttering?

4 268 P.T. Fox / Journal of Fluency Disorders 28 (2003) Neural system s modeling Ultimately, what is needed are explanations at the neural system level as to how speech production is organized and executed, how the speech system is dysregulated so as to produce the execution errors collectively termed stuttering, and how fluency inductions and treatments achieve behavioral normalization. This level of explanation is well beyond descriptions of brain activation patterns or structural abnormalities. It is well beyond box-and-arrow models of the connections among brain regions implicated in language àlájürgens (2002), even when accompanied by verbal descriptions of the functions of each box àlápetersen, Fox, Posner, Mintun, and Raichle (1989). What is needed is a computational model of the neural systems of speech a model that describes the system as a whole and quantifies both the functioning of individual regions and the functional interactions among regions. How do we begin to build such a model? Can it be built from functional imaging data? Control-system models, neural network models, and realistic models of neuronal conduction properties have been applied to great advantage to invasively acquired electrophysiological recordings. Heuristic, non-quantitative models (e.g., Jürgens, 2002; Mayberg et al., 2000) have been developed from functional imaging data. Heuristic models can be used as general guides for experimental design but not for data modeling and model-based analysis. Mathematical modeling techniques that add more rigor to functional imaging have been developed in recent years, but remain quite limited in scope. Functional connectivity modeling inter-regional covariances in functional activation levels during task performance was introduced to functional imaging as an explicit technology transfer from multi-electrode electrophysiology (Friston, 1994; Grafton, Sutton, Couldwell, Lew, & Waters, 1994), in an attempt to impart modeling rigor to functional imaging. Using functional connectivity to demonstrate discrete paths of information exchange in the motor system, Liu, Gao, Liotti, Pu, and Fox (1999) showed high information exchange between supplementary motor area and the dentate nucleus and between the primary motor cortex and the external segment of the globus pallidus, externa, but little information exchange between these two subsystems. In the speech motor system, functional connectivity was used to model differences in inter-regional information flow between speech and tongue movement (He et al., 2003). Functional connectivity modeling, however, has two serious limitations. First, it models information exchange but not information flow. That is, the covariance between two regions is expressed as a single, bi-directional value, rather than separate values for each direction. Second, data are obtained during task performance, making them task-specific. Thus, connectivity values obtained during any given task likely won t apply to any other task. Further, changes over time in connectivity values (i.e., treatment effects) are almost certainly contaminated by alterations in task strategy (i.e., which brain areas are recruited and to what degree) and is not an unambiguous measure of changes in synaptic efficiency (receptor density), the most likely mechanism of motor learning.

5 P.T. Fox / Journal of Fluency Disorders 28 (2003) Structural equation modeling (SEM), also termed Path Analysis, was introduced to functional imaging by McIntosh and Gonzalez-Lima (1991), and Grafton (Grafton & DeLong, 1997; Grafton et al., 1994). SEM addresses (some of) the shortcomings of functional connectivity analysis. As applied to functional imaging data, SEM (a technology transfer from social psychology) adds directionality to the model by assuming/postulating specific anatomical connections and omitting others. The raw inter-regional covariances are then modeled onto this assumed connectivity, to derive unidirectional path weights. SEM, however, still has severe limitations. Perhaps most glaring is the need to postulate the existence of specific connections and the non-existence of others, with such assumptions being based on animal work in all SEM studies reported to date. There are many systems (e.g., the speech system) in which such assumptions are highly problematic. Second, like functional connectivity analysis, SEM remains task-specific, in that the data from which covariances are computed are obtained during a task. Task-independent methods of measuring inter-regional connectivity have been established as an important technical objective in several laboratories specifically to address this shortcoming of functional connectivity modeling and SEM modeling. Fox et al. (1997) and Paus et al. (1997) introduced the conjoined application of transcranial magnetic stimulation (TMS) and PET, to measure inter-regional connectivity in a task-free manner. TMS directly depolarizes neurons in cortex immediately beneath coil face. This local depolarization physiologically propagates to remote regions connected to the target region. The TMS-induced changes in neural activity, both local and remote, can be readily imaged using PET. Further, TMS/PET provides path-weighting values (r-values) for the strengths of each connection. As it is unreasonable to expect that treatments will cause new physical connections between regions (new fiber pathways), changes in TMS/PET-derived path weights are best interpreted as changes in synaptic weighting of existing pathways, i.e., as synaptic plasticity. TMS/PET has been used to demonstrate TMS-induced synaptic plasticity in persons with PDS (Tandon et al., 2000). To date, however, no laboratory has used TMS/PET data to constrain a SEM of functional imaging data, nor to test the mechanisms of action of a non-tms therapy. Resting-state fmri is being explored as an alternative to TMS/PET for taskindependent measurement of inter-regional connectivity (Xiong, Gao, & Fox, 1999). In the resting state, regional neuronal firing rates are periodically variable. Regions that are anatomically connected show temporal covariation at rest. The time scale of this periodicity is sufficiently slow (seconds) that it can be detected by functional MRI performed with short repetition time (TR = 1 s). Resting-state fmri connectivity measures have the advantage of providing connectivity information for any and all brain regions imaged, rather than being limited to a stimulated area, as with PET/TMS. Resting-state fmri is not yet a well-validated or widely accepted method. Still, it has the potential to supplement (possibly even supplant) TMS/PET for measuring inter-regional connectivity, testing for synaptic plasticity, and for providing anatomical constraints for SEM.

6 270 P.T. Fox / Journal of Fluency Disorders 28 (2003) Functional imaging research sorely needs the rigor provided by mathematical modeling. A modeling framework structural equation modeling has been developed which appears appropriate for modeling human imaging data in a manner which captures some (but certainly not all) of the important characteristics of the neural systems of speech. SEM is most rationally applied when task-independent (i.e., resting state ) anatomical connectivity data are used, in addition to task-induced regional covariations. To date, this has not been reported by any laboratory. A reasonable next step would be to apply these advanced imaging and modeling methods to a treatment trial in developmental stuttering. If successful, such a study has the potential of defining a new strategy by which functional imaging is applied to disease pathophysiology and treatment mechanisms of action and should provide fundamental insights into the nature of stuttering. 4. Developmental stuttering as a proving ground for imaging methods From the perspective of a scientist heavily involved in the development of brain imaging strategies, persons with persistent developmental stuttering offer a truly unique resource. This is a disorder whose symptoms can be rapidly and briefly eliminated (by fluency induction), allowing subjects to be imaged with and without stuttering in a single imaging session. Further, it is a disorder for which reasonably effective treatments exist, allowing the mechanisms of action of therapy to be studied with imaging before and after treatment. Subjects with PDS are medically, psychiatrically and neurologically unimpaired, making it much easier for them to participate in research than for persons with virtually any other neural system disorder. Evidence, both from behavioral studies and from imaging studies, indicates that PDS is quite restricted in the neural systems compromised and non-progressive. This makes PDS far easier to study than degenerative disorders (e.g., Parkinson s disease, Alzheimer s disease). All of these factors combine to make PDS an extremely attractive model system within which to develop and validate imaging methods for modeling neural systems and neural system disorders and within which to learn to use imaging as a phenotype for genetic analyses. Using PDS as a vehicle for prototyping new imaging strategies clearly benefits persons with PDS, but should also bring benefits to persons with a wide range of brain disorders in whom these new techniques subsequently are applied. References Ambrose, N. G., Cox, N. J., & Yairi, E. (1997). The genetic basis of persistence and recovery in stuttering. Journal of Speech, Language Hearing Research, 40, Bookheimer, S. Y., Strojwas, M. H., Cohen, M. S., Saunders, A. M., Pericak-Vance, M. A., Mazziotta, J. C., & Small, G. W. (2000). Patterns of brain activation in people at risk for Alzheimer s disease. New England Journal of Medicine, 343,

7 P.T. Fox / Journal of Fluency Disorders 28 (2003) Foundas, A. L., Bollich, A. M., Corey, D. M., Hurley, M., & Heilman, K. M. (2001). Anomalous anatomy of speech-language areas in adults with persistent developmental stuttering. Neurology, 57, Fox, P. T., Ingham, R. J., George, M. S., Mayberg, H. S., Ingham, J. C., Roby, J., Martin, C., & Jerabek, P. (1997). Imaging human intra-cerebral connectivity by PET during TMS. Neuroreport, 8, 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, Friston, K. J. (1994). Functional and effective connectivity in neuroimaging: A synthesis. Human Brain Mapping, 2, Grafton, S. T., & DeLong, M. (1997). Tracing the brain s circuitry with functional imaging. Nature Medicine, 6, Grafton, S. T., Sutton, J., Couldwell, W., Lew, M., & Waters, C. (1994). Network analysis of motor system connectivity in Parkinson s disease: Modulation of thalamocortical interactions after pallidoctomy. Human Brain Mapping, 2, Hasnain, M., Fox, P. T., & Woldorff, M. G. (2001). Structure-function spatial covariance in the human visual cortex. Journal of Cerebral Cortex, 11, He, A. G., Tan, L. H., Tang, Y., James, G. A., Wright, P., Eckert, M., Fox, P. T., & Liu, Y. (2003). Modulation of neural connectivity during tongue movement and reading. Human Brain Mapping, 18, Jürgens, U. (2002). Neural pathways underlying vocal control. Neuroscience and Biobehavioral Review, 26, Kochunov, P., Lancaster, J., Tan, L. H., Amunts, K., Zilles, K., Mazziotta, J., & Gao, J. H. (2003). Localized morphological brain differences between English-speaking Caucasians and Chinese-speaking Asians: New evidence of anatomical plasticity. Neuroreport, 14, Kochunov, P., Thompson, P., & Fox, P. (2003). Asymmetry of the brain surface from deformation field analysis. Human Brain Mapping, 19, Liu, Y., Gao, J. H., Liotti, M., Pu, Y., & Fox, P. T. (1999). Temporal dissociation of parallel processing in the human subcortical outputs. Nature, 400, Mayberg, H. S., Brannan, S. K., Tekell, J. L., Silva, J. A., Mahurin, R. K., McGinnis, S., & Jerabek, P. A. (2000). Regional metabolic effects of fluoxetine in major depression: Serial changes and relationship to clinical response. Biological Psychiatry, 48, McIntosh, A. R., & Gonzalez-Lima, F. (1991). Structural modeling of functional neural pathways mapped with 2-deoxyglucose: Effects of acoustic startle habituation on the auditory system. Brain Research, 547, Ono, M., Kubik, S., & Abernathy, C. D. (1990). Atlas of cerebral sulci. New York: Thieme Medical Publishers, Inc. Paus, T., Jech, R., Thompson, C. J., Comeau, R., Peters, T., & Evans, A. C. (1997). Transcranial magnetic stimulation during positron emission tomography: A new method for studying connectivity of the human cerebral cortex. Journal of Neuroscience, 17, Petersen, S. E., Fox, P. T., Posner, M. I., Mintun, M., & Raichle, M. E. (1989). Positron emission tomographic studies of the processing of single words. Journal of Cognitive Neuroscience, 1, Salmelin, R., Schnitzler, A., Schmitz, F., Jäncke, L., Witte, O. W., & Freund, H. J. (1998). Functional organization of the auditory cortex is different in stutterers and fluent speakers. Neuroreport, 9, Sommer, M., Koch, M. A., Paulus, W., Weiller, C., & Büchel, C. (2002). Disconnection of speechrelevant brain areas in persistent developmental stuttering. Lancet, 360, Tandon, N., Fox, P., Ingham, R. J., Ingham, J. C., Collins, J., Pridgen, S., & Lancaster, J. (2000). TMS-induced modulation of cerebral blood flow in stutterers. NeuroImage, 11, S279. Thompson, P. M., Cannon, T. D., Narr, K. L., van Erp, T., Poutanen, V.-P., Huttunen, M., Lonnqvist, J., Standertskjöld-Nordenstam, C.-G., Kaprio, J., Khaledy, M., Dail, R., Zoumalan, C. I., & Toga, A. W. (2001). Genetic influence on brain structure. Nature Neuroscience, 4,

8 272 P.T. Fox / Journal of Fluency Disorders 28 (2003) Watkins, K. E., Vargha-Khadem, F., Ashburner, J., Passingham, R. E., Connelly, A., Friston, K. J., Frackowiak, R. S. J., Mishkin, M., & Gadian, D. G. (2002). MRI analysis of an inherited speech and language disorder: Structural brain abnormalities. Brain, 125, Weiller, C., Chollet, F., Friston, K. J., Wise, R. J., & Frackowiak, R. S. J. (1992). Functional reorganization of the brain in recovery from striatocapsular infarction in man. Annals of Neurology, 31, Weiller, C., Ramsay, S. C., Wise, R. J., Friston, K. J., & Frackowiak, R. S. J. (1993). Individual patterns of functional reorganization in the human cerebral cortex after capsular infarction. Annals of Neurology, 33, Xiong, J., Gao, J. H., & Fox, P. T. (1999). Interregional connectivity to primary motor cortex revealed using MRI resting state images. Human Brain Mapping, 8,