Identifying Effective Treatments for Dystonia in Patients With Cerebral Palsy

Abstract

Recent focus on improving the recognition of dystonia in cerebral palsy (DCP) has highlighted the need for more effective treatments. Evidence supports improved functional outcomes with early interventions for patients with cerebral palsy, but it is not known which interventions are most effective for DCP. Current pharmacologic recommendations for DCP are based largely on anecdotal evidence, with medications demonstrating minimal to moderate improvements in dystonia and variable efficacy between patients. Patients, families, and clinicians have identified the need for new and improved treatments in DCP, naming this as the top research theme in a recent Neurology® publication. Precision therapeutics focuses on providing early effective interventions that are individualized to every patient and can guide research priorities to improve treatments for DCP. This commentary outlines current obstacles to improving treatment of DCP and addresses how precision therapeutics can address each of these obstacles through 4 key components: (1) identification of predictive biomarkers to select patients likely to develop DCP in the future and for whom early intervention may be appropriate to delay or prevent full manifestation of dystonia, (2) stratification of patients with DCP into subgroups according to shared features (clinical, functional, biochemical, etc) to provide a targeted intervention based on those shared features, (3) administration of an individualized dose of an effective intervention to ensure adequate concentrations of the therapeutic entity at the site of action, and (4) monitoring of objective biomarkers of response to intervention. With implementation of each of these components of precision therapeutics, new and more effective treatments for every person with DCP can be realized.

Introduction

Cerebral palsy (CP) affects an estimated 2 to 2.5 per 1,000 live births worldwide¹ and is most commonly caused by perinatal brain injury secondary to hypoxic ischemic encephalopathy (HIE) or premature birth. Children with CP exhibit 2 predominant types of hypertonia: spasticity and dystonia.² Dystonia is the predominant movement disorder in dyskinetic CP, accounting for 3%–15% of total CP diagnoses.³ However, up to 60% of children with spastic predominant CP also exhibit dystonia,⁴ and dystonia is the primary determinant of functional impairment,⁵ making it both common and severely disabling. This translates to a significant economic impact, with the Centers for Disease Control estimating that people with CP born in 2000 will accumulate a lifetime cost of $11.5 billion dollars from loss of productivity, medical costs, and premature death.⁶ Despite its high impact and prevalence, dystonia in CP (DCP) remains underdiagnosed; dystonia and spasticity are accurately identified in only 30% of children by the age of 5 years.⁷

DCP has an evolving phenotype in the first 2 years of life, reflecting continued development of brain networks responsible for its clinical manifestation.^7,8 It is logical that early treatment of CP during this period when the brain is most plastic results in better outcomes. In fact, literature shows that early motor intervention in CP can maximize plasticity.⁹ Implementation of early constraint-induced movement therapy improves short-term and probably long-term hand function,¹⁰ and Goals-Activity-Motor Enrichment improves motor and cognitive outcomes at 12 months.¹¹ The need for early treatment highlights an opportunity for early diagnosis of DCP in at-risk children and implementation of early effective interventions to improve their long-term function.

Several obstacles need to be overcome before this opportunity can be realized.

1. It is difficult to reliably predict which infants with perinatal brain injury will develop DCP and which infants will not. Differential risk of dystonia following a common insult implies that the consequences of perinatal brain injury involve multiple cellular, molecular, and biochemical mechanisms; interindividual variability in the extent to which affected pathways contribute to the presence or absence of dystonia, or the degree of functional impairment associated with the dystonia, could have important implications for decisions regarding who to treat and when to initiate treatment. Inadequate understanding of the factors contributing to acquired dystonia after perinatal brain injury and the mechanisms involved leads to a risk of delayed diagnosis, precluding early intervention and missing the window of maximal response to intervention.

2. A better understanding of the pathways involved creates an opportunity to identify therapeutic interventions that modulate affected pathways contributing to acquired dystonia. Although there are numerous pharmacologic treatments for DCP, these medications are often minimally effective. A recent meta-analysis demonstrated that among current pharmacologic and neurosurgical interventions, most display marginal or no improvement in function with the quality of evidence ranging from very low to low.¹² The care pathway for the management of DCP published by the American Academy of Cerebral Palsy and Developmental Medicine (AACPDM) gives recommendations for oral medications based only on “expert opinion,” even noting that trihexyphenidyl, the second-line recommendation for treatment of generalized dystonia, may be possibly ineffective for dystonia based on current evidence.¹³

3. It is necessary to measure treatment response in an accurate, objective, and timely manner. Benefits of treatment for dystonia can take weeks to months to manifest,¹⁴ potentially leading patients and practitioners to prematurely adjust a dose or switch treatment options before the full treatment effect becomes evident. The time interval between initiation of treatment and onset of therapeutic response is accompanied by the risk that a child may receive ineffective treatment, leading to delay in effective treatment and the potential for unnecessary side effects. Finally, even when an effect is evident, many current clinical dystonia rating scales lack validity in the population with CP to reliably detect improvement in dystonia.¹⁵

In other medical specialties, the concept of precision medicine has gained traction. Medical oncology has embraced the use of predictive biomarkers to stratify patients and enrich clinical trial selection to improve efficacy and minimize medication toxicity.¹⁶ Investigations into pharmacogenomic effects on drug clearance have resulted in published genotype-based dosing guidelines for proton pump inhibitors,¹⁷ among others. Within the field of CP and spasticity management, a pharmacogenomic investigation into the variability of baclofen disposition and response after oral administration has revealed genetic factors that contribute to observed variability in baclofen clearance and clinical response, underscoring the potential for individualized therapy in CP management as well.¹⁸ However, there are no individualized approaches for patient selection and pharmacologic interventions in DCP.

A precision medicine approach can be used to facilitate early diagnosis and investigate the reasons for variable drug response among patients through 4 key components: (1) identification of predictive biomarkers to select patients likely to develop DCP in the future and for whom early intervention may be appropriate to delay or prevent full manifestation of dystonia, (2) stratification of patients with DCP into subgroups according to shared features (clinical, functional, biochemical, etc) to provide a targeted intervention based on those shared features, (3) administration of an individualized dose of an effective intervention to ensure adequate concentrations of the therapeutic entity at the site of action, and (4) monitoring of objective biomarkers of response to intervention (Figure 1). The following commentary further describes current obstacles in the management of DCP and outlines how each of the components of precision therapeutics can be applied to guide future research.

Figure 1. Application of Precision Medicine to Research in Dystonia and Cerebral Palsy.

Application of precision medicine to guide future research in dystonia and cerebral palsy requires (1) identification of predictive biomarkers of disease, (2) patient stratification to provide a targeted intervention based on individual physiology, (3) administration of an individualized dose of an effective intervention, and (4) monitoring of objective biomarkers of response to intervention.

Predictive Biomarkers

Early diagnosis and intervention in CP is crucial to maximize functional outcome.⁸ However, patients with perinatal brain injury may not demonstrate evidence of tone abnormality for months or even years after their initial injury. Even when tone and movement abnormalities are recognized, it can be difficult to identify the correct motor phenotype. Many practitioners are unaware of the nuances of dystonia diagnosis, and current evaluation tools in infants, such as the Hammersmith Infant Neurological Examination, Prechtl Qualitative Assessment of General Movements, and Developmental Assessment of Young Children, target overall motor or cognitive function and not dystonia specifically. Most dystonia-specific rating scales were developed to evaluate primary dystonia in adults, and the few that were designed to measure DCP (such as the Dyskinesia Impairment Scale and the Barry-Albright Dystonia Rating Scale) were not designed for infants and have moderate reliability at best.¹⁵ In the absence of a reliable clinical tool, there remains a need to develop an objective biomarker that can be used to determine which infants are at risk of developing dystonia.

Currently, neuroimaging remains one of the most promising predictive biomarkers in CP. Multiple studies have demonstrated the ability of structural MRI lesions to predict adverse motor or cognitive outcomes in CP, with the development of several predictive scoring systems.^19,20 Conversely, studies specifically predicting dystonia are more limited and demonstrate inconsistent associations, with injury in the subthalamic nucleus,²¹ lenticular nucleus,²² and periventricular white matter²³ all associated with dystonia. The inconsistent association between dystonia and a specific brain structure is likely because dystonia is a disorder of motor networks and is not caused by damage to any one structure.²⁴

MRI techniques investigating structural and functional connectivity may provide complementary insights into neonatal brain network aberrations that could better correlate with dystonia outcomes. Studies of premature infants have revealed regional differences in functional and structural connectivity that correlate with developmental outcomes. For example, one study found reduced fractional anisotropy in the left temporal lobe and increased fractional anisotropy in the right temporal lobe of very preterm infants (less than 30 weeks gestation), reflecting regional differences in white matter integrity that correlated with motor outcomes (as measured by the motor composite score of the Bayley Scales of Infant and Toddler Development) at 2 years of age.²⁵

Given that dystonia is a disorder of motor networks, approaches to analyze whole brain connectivity networks, such as graph theory, may be even more fitting. A whole-brain network approach can reveal the importance of specific motor networks implicated in dystonia (e.g., cortico-thalamic-basal ganglia and cerebellar networks)²⁴ and investigate the interactions between these and other networks. Using graph theory, the brain is segmented into predefined regions of interest, and pairwise connectivity metrics such as fractional anisotropy (for structural connectivity) or blood oxygenation level–dependent response (for functional connectivity) are calculated and mapped onto an adjacency matrix. These calculations are then used to quantify structural and functional connectivity metrics of network integration and segregation to compare network organization between groups²⁶ (Figure 2). A recent publication used graph theory to demonstrate significant differences in structural connectivity networks between infants with HIE and those with congenital heart disease and correlated the differences in network metrics with overall motor outcomes.²⁷

Figure 2. Illustration of Graph Theory Metrics to Characterize and Quantify Brain Networks.

The brain is segmented into regions of interest (nodes), with functional or structural connectivity expressed as a weighted measure between each pair of nodes (edges). The connectivity measures are used to create graph theory metrics of network segregation and integration. Examples of network segregation measures include modularity (the degree to which network can be subdivided into individual modules) and clustering coefficient (connection probability to nearest neighbor nodes). Network integration measures include shortest path (minimum number of edges to travel from one node to another). Connector hubs are nodes that connect 2 modules in a network to each other.

Although there are no similar pediatric studies in dystonia, a graph theory approach was used to quantify functional and structural brain network differences in adults with focal dystonia and to distinguish between dystonia phenotypes.^28,29 These studies demonstrated whole-brain network differences involving the sensorimotor and associative cortical areas in patients with focal dystonia, with specific alteration to the insular network connecter hub that may be important in coordinating the transfer of information within the larger neural network. Additional phenotypic-specific network changes were identified because patients with laryngeal dystonia exhibited abnormalities in the regional network involving the premotor cortex and superior parietal areas, while patients with writer's cramp demonstrated reorganization of occipital, primary motor, and cerebellar network hubs. With a window of opportunity between time of brain injury and onset of dystonia in infants, similar MRI whole-brain connectivity analysis could be investigated as a predictive biomarker of DCP and used to guide early interventions and enrich patient selection for therapeutic clinical trials.

Identifying a Targeted Intervention

A precision medicine approach necessitates not only identifying an affected population but also identifying effective intervention(s) that can be implemented early in the disease course to prevent or at least ameliorate any loss of function in that population. It is likely, if not certain, that subpopulations of patients with dystonia may respond best to different interventions. Neuroimaging methods described earlier can be used to subdivide affected patients into subpopulations based on network differences. To determine which intervention may be most effective at preventing the development or progression of dystonia in each subpopulation, understanding the underlying pathophysiology is necessary. Basic science investigations into the pathophysiology of DCP in hypoxic brain injury suggest a central role of striatal cholinergic interneurons. Rat models demonstrate increased number of striatal cholinergic interneurons in response to neonatal hypoxia,³⁰ and genetic mouse models of dystonia demonstrate abnormal excitation of striatal cholinergic interneurons.³¹

The increased number and excitability of cholinergic neurons in hypoxic ischemia models of dystonia suggest that anticholinergic medications may be particularly promising for treatment of dystonia and perhaps even as an early intervention directly targeting the underlying pathophysiology of dystonia. In fact, a prospective open-label clinical trial of trihexyphenidyl in children with DCP demonstrated significant improvement in upper extremity motor function at 15 weeks, but a large subgroup had worsening of hyperkinetic movements while on the medication. The authors ultimately concluded that the evidence remains equivocal for its use in CP.¹⁴

If trihexyphenidyl holds promise as a medication that could target the underlying pathophysiology of dystonia and affect disease progression, why have studies been plagued by variabilities in response and excessive adverse effects? Reasons could be numerous, including variable receptor density, heterogenous patterns of brain injury affecting underlying pathophysiology, medical comorbidities, or epigenetic factors. To individually assess all the contributing factors affecting variability in the response to a medication, it is necessary to first ensure that poor response is not simply a consequence of inadequate drug concentration of the site of drug action.

Individualized Dosing

Systemic exposure refers to how much of a drug is present in systemic circulation and is affected by drug absorption, distribution, metabolism, and excretion (ADME). Systemic exposure is the product of drug concentration over time and is easily visualized in a plasma concentration-time profile of a drug, with area under the curve reflecting the systemic exposure. For a given drug, systemic exposure varies significantly between patients based on individual factors affecting ADME (Figure 3A).

Figure 3. Illustration of Concepts of Drug Bioavailability and the Variability Between Patients.

Data used to construct figure are derived from Leeder 2019.³² (A) The solid line represents the average plasma drug concentration-time profile for all patients. The gray shaded area represents the area under the curve, which reflects the systemic exposure to a drug at a given dose. The green box represents drug concentrations required to achieve a desired therapeutic response. (B) Individual patient data plotted demonstrate variability in drug concentration dependent on CYP2D6 genotype: poor metabolizers (no functional alleles; red), intermediate metabolizers (1 null allele and 1 partial function allele; yellow), extensive metabolizers (1 functional allele; green), and ultrarapid metabolizers (2 functional alleles; blue). Colored trend lines demonstrate the variability in average concentration between groups. Points above the green box are within toxic concentrations and may produce side effects; points below the green box are subtherapeutic and may not lead to a physiologic response.

For instance, many drugs are metabolized in the liver through the cytochrome P450 (CYP) system including enzymes such as CYP2C9, CYP2D6, and CYP3A4. CYP2D6 is of particular interest because it is a highly polymorphic gene, with more than 170 identified allelic variants that can affect the metabolic activity of the expressed enzyme.³³ Based on an individual's CYP2D6 diplotype, an individual may be classified as a poor, intermediate, extensive, or ultrarapid metabolizer. As an example of the functional consequence of CYP2D6 allelic variation on system exposure to a CYP2D6 substrate medication, a previous study demonstrated a 50-fold range in systemic exposure to atomoxetine (Strattera) after administration of a weight-based dose of the medication (0.5 mg/kg).³⁴ This broad range of exposures following a standard weight-based dose of the medication has potential consequences for both efficacy and toxicity; individuals who are CYP2D6 poor metabolizers (2 nonfunctional copies of the gene) have higher systemic exposures and may be at an increased risk of side effects, while individuals with normal or ultrarapid metabolism (2 or more functional copies of the gene) may not reach a systemic exposure sufficient to achieve the desired response to the drug, even at the highest recommended dose of the drug (Figure 3B).

Even within a genotype, there may be large variability in dose-normalized systemic exposure. In a genotype-stratified pharmacokinetic study of pravastatin in pediatric patients,³⁵ the authors replicated the approximately 2-fold increase in systemic exposure per copy of variant SLCO1B1 allele, as reported in adults (between genotype groups), but more importantly, demonstrated an 11-fold difference in systemic exposure within patients with the same SLCO1B1 genotype. This observation highlights the importance of additional nongenetic factors affecting ADME processes in individual patients that can influence systemic exposure.

In pharmacologic research, the concepts of ADME and systemic exposure are applied to investigations of the dose → exposure → response relationship, meaning that an individual patient is given a particular dose of drug to achieve a certain systemic exposure, which produces the desired response. For a clinician, it is logical to turn this statement on its head to response → exposure → dose; to produce the desired clinical response or outcome for a given patient (i.e., improvement of dystonia), what systemic exposure is needed in that patient, and what dose of the drug is required to provide the desired systemic exposure for that individual?³⁶

Turning back to trihexyphenidyl, little is known about its metabolism and thus the relationship between administered dose and systemic exposure, especially in children with dystonia. Although PubChem reports limited information on trihexyphenidyl metabolism, a study in 1978 (under trihexyphenidyl's previous name of benzhexol) identified hydroxylated metabolites in the urine of patients taking trihexyphenidyl, suggesting at least some degree of metabolism.³⁷ Tools such as SMARTCyp³⁸ are now available to provide insight into the potential sites of drug metabolism by CYPs based on molecular structure. SMARTCyp predicts trihexyphenidyl metabolism by CYP2D6 and CYP3A4, which would produce metabolites consistent with the reported structures of benzhexol (trihexyphenidyl) metabolites. In fact, a 1985 study by Burke and Fahn³⁹ investigating pharmacokinetics of trihexyphenidyl in adults noted substantial variability in plasma concentration-time profiles between patients, further supporting the contention that individual patient factors affecting ADME of trihexyphenidyl can affect systemic exposure.

Variability in systemic exposure due to genetic variation (and additional as yet uncharacterized nongenetic factors) could explain why trihexyphenidyl may be effective for some patients, while others experience excessive side effects or minimal improvement.¹⁴ Further investigation of how trihexyphenidyl is cleared by the body, the pathways involved, and how an individual's pharmacogenotype affects trihexyphenidyl disposition will inform individualized dosing strategies and allow desired drug exposures to be achieved in each patient.

While achieving the desired systemic exposure for a medication may be the first step, additional questions remain regarding the relationship between systemic exposure and clinical response. How should studies be designed to collect this information? One option, especially for drugs such as atomoxetine with considerable variability in the dose-exposure relationship due to genetic variation, is to conduct a study in which doses are individualized to provide each patient/study participant with an equivalent drug exposure and then escalate the exposure (as opposed to the dose) to identify factors contributing to variability in response. But a necessary step for applying this paradigm to drugs prescribed to pediatric patients with dystonia, such as trihexyphenidyl, is defining an objective measure of response.

Response Biomarkers

The final component required for effective implementation of precision therapeutics in DCP is an objective biomarker of response. Interventions in dystonia may take months to see a clinical effect, leading to significant delays in determining appropriate medication dosing or even premature discontinuation of drugs that could be effective. For example, pediatric patients in the clinical trial of trihexyphenidyl did not demonstrate significant improvement until 16 weeks of treatment.¹⁴ Deep brain stimulation for dystonia also measures improvements over the course of months, rather than weeks.⁴⁰ Even when a clinical response becomes evident, current rating scales used to evaluate dystonia are often not validated in children and may not be sensitive enough to detect changes in dystonia, especially in the setting of mixed movement phenomenologies.¹⁵ Using quality-of-life scales, patients and families can comment on visible improvements in function, but this remains an indirect measure of response.

A quantifiable response biomarker would allow for objective assessment of dystonia to evaluate efficacy of an intervention. This biomarker would be invaluable in clinical trials to establish the necessary systemic exposure of a drug required to achieve the desired therapeutic response. Ideally, a response biomarker would reflect underlying physiologic changes occurring during an intervention, potentially even before a clinical change is apparent. Just as neuroimaging graph theory metrics hold promise as predictive biomarkers of dystonia, these metrics could also be monitored to evaluate treatment response. A recent study in pediatric patients with bipolar disorder demonstrated initial differences in graph theory metrics (clustering coefficient and path length), which resolved after 6 weeks of treatment with lithium or quetiapine.⁴¹ Yet, limitations in the use of imaging response biomarkers exist. It is not known how sensitive these metrics would be to quantify treatment response in dystonia, and these types of metrics are not currently used as biomarkers in the clinical setting. However, they are derived using data available from most MRI scanners and could be implemented across care settings with appropriate infrastructure.

Conclusion

Dystonia can have a devastating effect on the function and quality of life of an individual with CP. This deleterious effect highlights an opportunity to apply an early effective intervention in children that can affect the remainder of their lives. Application of precision therapeutics can improve treatment of DCP by addressing research priorities including (1) identification of predictive biomarkers to select appropriate patients for early intervention, (2) stratification of patients to provide a targeted intervention based on their individual physiology, (3) administration of an individualized dose of an effective intervention, and (4) monitoring of objective biomarkers of response to intervention.

However, implementation of precision therapeutics to improve treatment of DCP will not be possible without increasing engagement in the community to promote early recognition and treatment of DCP. The CP community includes not only clinicians, but patients and families as well. Because only 30% of children at high risk of CP have an identified motor phenotype by age 5 years,⁷ large numbers of children with dystonia are not being treated simply because they are not being diagnosed. A recent survey of clinicians at the Child Neurology Society Cerebral Palsy interest group meeting indicated that poor identification of motor phenotype could be partly due to lack of knowledge, with 30% of respondents reporting no or limited knowledge of motor phenotyping.⁴²

But even beyond improving motor phenotyping by neurologists, many children with CP first encounter the medical system through physiatry, orthopedic surgery, or general pediatric providers who may have a very different knowledge base and priorities of care. Organizations such as the Cerebral Palsy Research Network, Cerebral Palsy Foundation, AACPDM, Child Neurology Society, and Dystonia Medical Research Foundation can enhance collaborations between specialties and aid in the recruitment of underrepresented populations for research studies. Of most importance, engaging with these organizations and with patients and their families ensures that future research will be both impactful and accessible to this often underserved and under-resourced population.

Precision therapeutics brings personalized medicine from the bench to the bedside, helping clinicians advocate for the individual needs of their patients, identify their specific motor phenotype, and consider treatments that account for each patient's functional goals. Future developments in model-informed precision dosing⁴³ could allow clinicians to make individualized dosing decisions on mobile apps or web-based tools without requiring expertise in pharmacokinetic modeling, translating into immediate treatment benefits for patients with DCP. A recent survey of the CP community, including patients and families, indicated that developing new treatments is the top research priority for DCP.⁴⁴ This desire to improve treatments further underscores that now is the time to partner with the CP community to tackle these research priorities and bring precision therapeutics to DCP.

Glossary

AACPDM

American Academy of Cerebral Palsy and Developmental Medicine

ADME

absorption, distribution, metabolism, and excretion

CP

cerebral palsy

CYP

cytochrome P450

DCP

dystonia in CP

HIE

hypoxic ischemic encephalopathy

Appendix. Authors

Name	Location	Contribution
Rose Gelineau-Morel, MD	Division of Neurology, Children's Mercy Kansas City; School of Medicine, University of Missouri-Kansas City; Department of Pediatrics, University of Kansas Medical Center, Kansas City	Drafting/revision of the article for content, including medical writing for content; major role in the acquisition of data; study concept or design; and analysis or interpretation of data
Christopher Smyser, MD, MS	Department of Pediatrics, Department of Neurology, and Mallinckrodt Institute of Radiology, Washington University in St. Louis, MO	Drafting/revision of the article for content, including medical writing for content
J. Steven Leeder, PharmD, PhD	School of Medicine, University of Missouri-Kansas City; Department of Pediatrics, University of Kansas Medical Center; Division of Clinical Pharmacology, Toxicology & Therapeutic Innovation, Children's Mercy Kansas City, MO	Drafting/revision of the article for content, including medical writing for content; study concept or design

Study Funding

This work was supported by the Eunice Kennedy Shriver National Institute of Child Health and Human Development (T32HD069038).

Disclosure

The authors report no relevant disclosures. Go to Neurology.org/N for full disclosures.