Clinical Actionability of Genetic Findings in Cerebral Palsy

Key Points

Question

Does a genetic diagnosis provide clinical utility for preventive, focused symptom-management or disease-modifying primary treatments for cerebral palsy?

Findings

In this systematic review and meta-analysis, a moderate number of patients with cerebral palsy undergoing exome sequencing had an identifiable genetic etiology in a total of 243 different genes. A number of actionable findings for outcome severity, intervention safety, and treatment efficacy were found.

Meaning

Precision medicine opportunities exist for patients with cerebral palsy, demonstrating a benefit of genetic sequencing in this population.

This systematic review and meta-analysis evaluates the use of genetic testing results in changes in care for individuals with cerebral palsy.

Abstract

Importance

Single gene variants can cause cerebral palsy (CP) phenotypes, yet the impact of genetic diagnosis on CP clinical management has not been systematically evaluated.

Objective

To evaluate how frequently genetic testing results would prompt changes in care for individuals with CP and the clinical utility of precision medicine therapies.

Data Sources

Published pathogenic or likely pathogenic variants in OMIM genes identified with exome sequencing in clinical (n = 1345) or research (n = 496) cohorts of CP were analyzed. A systematic literature review for evidence of effective therapies for specific genetic etiologies was performed.

Study Selection

Nonstandard interventions that led to a detectable improvement in a defined outcome in individuals with variants in the gene of interest were included.

Data Extraction and Synthesis

Literature was evaluated using PRISMA guidelines. A diverse, expert working group was established, scoring rubrics adapted, and scoring consensus built with a modified Delphi approach.

Main Outcomes and Measures

Overall clinical utility was calculated from metrics assessing outcome severity if left untreated, safety and practicality of the intervention, and anticipated intervention efficacy on a scale from 0 to 3.

Results

Of 1841 patients with CP who underwent exome sequencing, 502 (27%) had pathogenic or likely pathogenic variants related to their phenotype. A total of 243 different genes were identified. In 1841 patients with identified genetic etiologies of CP, 140 (8%) had a genetic etiology classified as actionable, defined as prompting a change in clinical management. Also identified were 58 of 243 genes with pathogenic or likely pathogenic variants with actionable treatment options: 16 targeting the primary disease mechanism, 16 with specific prevention strategies, and 26 with specific symptom management. The level of evidence was also graded according to ClinGen criteria; 45 of 101 interventions (44.6%) had evidence class D or below. The potential interventions have clinical utility with 98 of 101 outcomes (97%) being moderate-high severity if left untreated and 63 of 101 interventions (62%) predicted to be of moderate-high efficacy. Most interventions (72 of 101 [71%]) were considered moderate-high safety and practicality.

Conclusions and Relevance

The findings indicate that actionable genetic findings occurred in 8% of individuals referred for genetic testing with CP. Evaluation of potential efficacy, outcome severity, and intervention safety and practicality indicates moderate-high clinical utility of these genetic findings. Genetic sequencing can identify precision medicine interventions that provide clinical benefit to individuals with CP. The relatively limited evidence base underscores the need for additional research.

Introduction

A genetic etiology for 15% to 34% of cerebral palsy (CP) cases can be identified using exome sequencing,¹ and commercial CP gene sequencing panels are available for clinical use. Although clinicians are increasingly using both panels and exome sequencing as part of the diagnostic workup of patients with CP, the impact of genetic findings on clinical care is still being determined. The current genomic landscape of CP is broad, with hundreds of individually rare single-gene causes.¹ Given the increased use of clinical genetic testing² and the heterogeneity of CP-associated genes being identified, there is a need to evaluate how genetic findings might impact the care of individuals with CP.

The actionability of genetic findings in epilepsy has been evaluated using an iterative process of expert review and curation. A study³ of 2008 adults with epilepsy reported that 11% had identifiable causative variants. In 56% of these cases, the genetic finding was expected to be actionable, prompting a change in clinical management.

ClinGen⁴ is a resource supported by the US National Institutes of Health. Its goal is to define the clinical relevance of genes and their variants for use in precision medicine and research. A ClinGen working group previously published a landmark rubric to assess the clinical actionability of genetic results,⁵ and a Gene Curation Expert Panel specifically dedicated to CP was formed in March 2022.⁶ However, the existing actionability rubric focuses on assessing incidental or secondary findings identified during clinical genetic sequencing rather than etiologic findings identified during a diagnostic workup.

By analyzing published data, we sought to build on prior efforts to determine to what extent genetic findings could impact the clinical care of individuals with CP. A working group comprised of content experts was organized and began with the established framework for determining actionability used by studies of epilepsy genetics. A systematic review of the literature and synthesis of evidence were used to compile a list of genes with actionable genetic findings. We then incorporated the ClinGen rubric and modified it to apply to the diagnostic workup of a person with CP in a pediatric care setting. We recognized that additional measures of clinical impact would provide added value for practicing clinicians. Accordingly, we developed estimates of untreated outcome severity, safety and practicality (of a particular intervention), and anticipated intervention efficacy. We then calculated an overall clinical utility score by combining these measures. Our overall objective was to determine the potential impact of identifying a genetic etiology on clinical management in CP. Our findings, including curated lists of genes, outcomes, and available interventions, form a starting point for using genetic data and supports the utility of this testing for improved outcomes for individuals with CP.

Methods

This study was reviewed by the Phoenix Children’s Hospital institutional review board. The study was determined to be exempt from human subject research oversight and thus granted a waiver of patient consent.

Cohorts and Gene Selection

An overview of study methodology is described in Figure 1. We identified 4 published research cohorts^7,8,9,10 (aggregate n = 496; 368 trios/quads, 124 duos/singletons) meeting inclusion criteria of at least 10 individuals with a confirmed CP diagnosis receiving exome sequencing and variant pathogenicity classified according to American College of Medical Genetics and Genomics (ACMG) guidelines.¹¹ Research patients were diagnosed based on the 2006 consensus criteria for CP diagnosis and are representative of the clinical population, including individuals with and without known risk factors. We also included a clinical cohort commercially sequenced by GeneDx for exome sequencing with suspected or confirmed CP, including 1009 individuals analyzed as trios and 336 as singletons.¹² For pooled analysis, CP-associated genes from the research cohort were identified from a published gene list¹² to ensure that pathogenic or likely pathogenic variants identified in patients were causally related to CP and not incidental findings.

Figure 1. Actionable Cerebral Palsy (CP) Gene Literature Search Strategy and Workflow.

See the eMethods in Supplement 1 for query terms and tools used for literature review. Publications were eliminated if patients did not have variants listed in the gene or if interventions were not trialed in patients directly, if treatment was part of standard of care, if the potential outcome had no effect on patient function, or if the reported phenotypes were not consistent with the genetic disorder in that gene most closely aligned to CP. Most of the identified literature (104 789 publications) did not meet inclusion criteria for a personalized medicine intervention based on genetic finding. Category assignment criteria and process described in the eFigure, A, in Supplement 1. The clinical utility impact assessment criteria and process described in eFigure, B, in Supplement 1. P/LP indicates pathogenic or likely pathogenic.

Systematic Literature Review and Inclusion Criteria

The Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) reporting guideline was used to guide the literature review. We analyzed the actionability of genes in the pooled cohort harboring likely pathogenic or pathogenic variants. Only relevant CP-associated genes were included.

Actionability Definitions and Categories

We defined a genetic finding as actionable if it was able to be classified as having an intervention that fell into at least 1 of the 3 ClinGen-derived categories defined as follows. Primary disease mechanism was defined as an intervention directly affecting the gene product, a supplement for a primary deficiency, or a specific treatment for a primary cause of pathology. Specific prevention strategies were defined as known, specific triggers associated with the genetic abnormality that worsen the condition and could be decreased through avoidance, early detection, or lifestyle changes. Specific symptom management was defined when it was evidence that a medication or intervention was found to improve at least 1 symptom associated with the disorder.

A gene may have interventions in different categories but was assigned to a single actionability category in rank order: primary disease mechanism, specific prevention strategy, or specific symptom management. Potential for future intervention (an intervention undergoing a phase II or phase III clinical trial listed in ClinicalTrials.gov or included in the Illuminating the Druggable Genome drug-gene interaction database) was selected only if the intervention in question did not meet the criteria for another category. Genes classified as having potential for future intervention were not considered in the overall calculation of actionability but instead reported in this separate category. In the future, such interventions could be reclassified in 1 of the 3 actionability criteria as evidence evolves. Other classes of actionability, such as family planning or referrals to different specialists, were not considered but represent additional important outcomes for individuals with CP and their families.

Defining Outcomes and Interventions and Scoring Clinical Utility

A total of 101 outcome-intervention pairs were curated. An outcome is a potential or ongoing clinical concern affecting the patient. An intervention was defined as actionable when knowledge of a genetic etiology would be anticipated to prompt a change from standard care. Standard care was defined as the typical treatment for the clinical manifestation or outcome regardless of etiology. For disorders where many drugs are available (ie, antiepileptics), reports of efficacy for a specific drug were considered actionable as precision medicine approaches could include the selection of a specific medication rather than adopting a trial-and-error approach. A list of definitions is provided in eTable 1 in Supplement 1.

Clinical utility metrics were derived from the ClinGen rubric^5,13 and adjusted for relevance in a pediatric neurology or neurodevelopmental care setting using the following grounding scenario. Imagine you are a clinician treating a young child diagnosed with CP. If an actionable genetic finding was identified, how could information about the seriousness of the condition’s natural history (outcome severity), risk and/or burden on the patient/family associated with the actionable intervention (safety/practicality), or potential benefit of the actionable intervention (efficacy) impact your decision-making?

Clinical utility scores were individually generated, and the strength of available evidence was considered a separate category independent of other factors. The relative frequencies of outcomes or adverse events were not explicitly factored into scoring, as we could not determine whether a given outcome was present in individuals from this aggregate cohort. Clinical utility scores were then defined as follows. Outcome severity was defined as the impact of the condition on patient function if left untreated. Safety and practicality was defined as the impact of the intervention including a potential adverse event or disruption to the patient or patient’s family, such as invasiveness, need for lifestyle alterations, or specialized monitoring or management. Efficacy was defined as the degree to which an intervention can prevent or reduce symptoms, improve function, or decrease the likelihood of additional complications or manifestations. Clinical utility was defined as an aggregate score calculated by adding outcome severity, safety and practicality, and efficacy scores. Each metric used a scale from 0 to 3, with higher scores corresponding to more compelling reasons for clinical intervention. Total clinical utility ranged from 0 to 9. Statistical analyses were performed using χ² or Wilcoxon tests with α set at .0.5.

Results

Clinical Actionability

The aggregate cohort in this systematic review and meta-analysis included 1841 individuals, for whom pathogenic or likely pathogenic variants were identified in 502, for an overall diagnostic yield of 27%. This rate agrees with meta-analysis of rates of genetic findings in a clinically verified mixed–risk factor CP population.¹ These variants were distributed among 243 unique genes (eTable 2 in Supplement 1); 94 genes were recurrent, with 2 or more affected individuals with variants meeting ACMG criteria for pathogenic or likely pathogenic consistent with the known inheritance model for each gene.

We found that 140 of 1841 individuals with CP (8%) who underwent exome sequencing had a genetic finding expected to alter their management. Thirteen additional individuals had a genetic result with treatments in clinical trials, and a further 18 had gene findings represented in Illuminating the Druggable Genome, a preclinical development database. Together, 171 of 1841 individuals with CP (9%) who underwent exome sequencing had a genetic finding with precision medicine treatments either currently available or in development (Figure 2A). These patients had pathogenic or likely pathogenic variants in 58 or 243 genes (24%) that we classified as actionable (Table 1). A further 15 (7%) were considered to have the potential for future intervention.

Figure 2. Frequency and Scores for Actionable Findings.

A, Percentage of cohort with pathogenic or likely pathogenic variant calls and with current or future actionable gene findings identified via literature review. B, Mean scores for each outcome-intervention pair for clinical utility metrics of severity, safety/practicality, and efficacy. C, Scores for safety/practicality averaged across all scorers for 61 unique interventions by category, including preventive, dietary changes, or adding over-the-counter supplements, US Food and Drug Administration (FDA)–approved drugs, and biological/surgical interventions, including gene or cellular therapy, enzyme replacement therapy, or surgical interventions. D, Total clinical utility score for each category of intervention. Preventive interventions had a higher clinical utility than interventions targeting the primary disease mechanism or specific symptom management. Boxes represent 25 and 75 percentiles, horizontal lines the medians, and whiskers the range of data. There were 101 total outcome-intervention pairs—18 primary, 36 preventive, and 47 symptom-management interventions. P/LP indicates pathogenic or likely pathogenic.

^aP = .02.

^bP < 2.5 × 10⁻⁵ by 2-tailed Wilcox test.

^cP < .001.

Table 1. Frequency of Actionable Genes and Patients With Genetic Findings by Categories of Interventions^a.

Category	No. (%)
	Primary disorder	Specific		Future (clinical trials or druggable genome)
		Preventive strategies	Symptom management
Total genes	16/243 (6.6)	16/243 (6.6)	26/243 (10.7)	15/243 (7.8)
Clinical, %	15/224 (6.7)	16/224 (7.1)	24/224 (10.7)	18/224 (8.0)
Research, %	2/50 (4.0)	3/50 (6.0)	6/50 (12.0)	5/50 (10.0)
Total patients	31/502 (6.2)	33/502 (6.6)	76/502 (15.1)	38/502 (7.6)
Clinical patients	28/431 (6.5)	29/431 (6.7)	66/431 (15.3)	33/431 (7.7)
Research patients	3/71 (4.2)	4/71 (5.6)	10/71 (14.1)	5/71 (7.0)

^a Likely pathogenic or pathogenic variants called by authors using American College of Medical Genetics and Genomics guidelines.¹¹ Percentage of patients with a treatable variant in that class was determined by dividing by the number of patients in that group with a pathogenic or likely pathogenic variant. Of 243 genes, 58 (24%) are considered treatable, and 140 of 502 patients (28%) with pathogenic or likely pathogenic findings had currently available treatments (defined as primary, specific symptom, or preventive).

We found that the proportion of actionable findings was similar for research and clinical cohorts. There was no difference in the ratios of actionable genes (11 of 50 vs 55 of 224; χ² P = .66) or individuals with actionable findings (17 of 71 vs 123 of 431; χ² P = .98) between these groups. These findings indicate that actionable genetic findings can be identified in CP populations identified through both clinical referral and assessment by pediatric neurologists during research studies. These datasets were pooled for subsequent analyses.

Clinical Utility

We subsequently scored 101 outcome-intervention pairs for 58 distinct genes for their clinical utility. A complete list of outcome-intervention pairs and assigned scores can be found in the eTable 5 in Supplement 2, with mean scores in Figure 2 and summarized data for each category in Table 2.

Table 2. Summary Scores of Clinical Utility Categories^a.

Score	Severity	Safety/practicality	Efficacy	Combined
0 (No impact)	NA	4	0	NA
1	3	25	38	NA
2	43	48	45	NA
3 (Most impact)	55	24	18	NA
Mean (95% CI)	2.40 (2.30-2.49)	1.94 (1.80-2.08)	1.76 (1.65-1.87)	6.10 (5.89-6.31)

Abbreviation: NA indicates not applicable.

^a Numbers indicate n = 101 total outcome/interventions.

Outcome Severity

We considered the natural history associated with OMIM clinical phenotypes that best matched the disease mechanism for each gene and scored outcomes as mild, moderate, or severe based on rubric criteria (Table 3¹⁴). We considered short- and long-term scenarios to capture day-to-day functional impairments and the potential for worsening function, episodic decompensations, or the possibility of premature death. We only considered outcomes that had evidence of being modified by the intervention. We identified 55 severe outcomes, 43 moderate outcomes, and 3 mild outcomes (Figure 2B). The mean outcome severity score was 2.40 (95% CI, 2.30-2.49) (Table 2) on our scale of 0 to 3. This suggests that outcomes associated with variants in these genes may substantially impair function, supporting the potential benefit of identifying disease-modifying interventions.

Table 3. Rubric for Scoring Impact of Genetic Finding^a.

Metric description	Score and definition	Example
Severity: short-term impact (active clinical management)	1 = Impairments present, but do not substantially impact adaptive function or self-help skills (mild)	Joint stress
	2 = Impairments limit adaptive function and self-help skills (moderate)	NKX2-1–associated chorea
	3 = Complete lack of adaptive function and self-help skills (severe)	PANK2-associated dystonia
Severity: long-term impact (outcome or risk of severe events if untreated)	1 = Transient decompensations (mild)	Paroxysmal dyskinesia
	2 = Decompensations or manifestations can worsen function, but may be small changes or not permanent (moderate)	ATP1A3-associated episodic decompensation
	3 = Decompensations or manifestations can permanently worsen function or are a serious event (severe)	Sudden cardiac death
Safety: potential risk of the intervention	0 = Life-threatening outcomes; urgent intervention indicated, or death related to adverse event (CTCAE grade 4-5)	Stem cell transplantation (organ failure, sepsis)
	1 = Severe or medically serious but not immediately life threatening; hospitalization or prolongation of hospitalization indicated; disabling; limiting self-care activities of daily living (CTCAE grade 3)	Deep brain stimulation surgery (perilesional edema, mild intracerebral hemorrhage, infection)
	2 = Moderate; minimal, local or noninvasive intervention indicated; limiting age-appropriate instrumental activities of daily living (CTCAE grade 2)	Levodopa (vomiting)
	3 = Mild; asymptomatic or mild symptoms; clinical or diagnostic observations only; intervention not indicated (CTCAE grade 1)	Leucine supplements (clinically silent hypoglycemia)
Practicality: impact of the intervention on the patient/family	0 = Highly invasive
	1 = Invasive, serious impact on lifestyle, or specialized management required	Low valine/protein diet
	2 = Potential for burden, but reasonable for most patients; treatment does not require intensive oversight	Ketogenic diet
	3 = Minimum burden or oversight	Avoid contact sports
Efficacy: improvement of the outcome with the intervention	1 = Small improvement in condition or decrease in risk of acquiring potential outcome	Levodopa for CTNNB1-associated dystonia
	2 = Moderate improvement in condition or decreased risk in acquiring potential outcome	Caffeine for ADCY5-associated dyskinesia
	3 = Considerable improvements, complete avoidance of potential outcome	Levodopa for SPR-associated dopa responsive dystonia; avoid metronidazole to prevent ERRC6-related hepatic failure

Abbreviation: CTCAE, Common Terminology Criteria for Adverse Events.

^a Severity evaluated disease-related outcomes for (1) short-term impact based on outcome effect on comfort, development, or adaptive function and (2) long-term impact based on potential functional decline, decompensations, or risk of serious events. The single most appropriate severity category was assigned for each outcome. Outcomes that had no impact on function (0 on the 0-3 scale) were not considered in this study. Risk assessed severity of a potential adverse event, as it was not possible to reliably ascertain absolute risk due to missing frequency data. Risk classifications also used CTCAE version 5.0 from US Department of Health and Human Services.¹⁴ Practicality considered examples from the Multimorbidity Treatment Burden Questionnaire to assess the impact of treatment on patients’ lifestyle. Efficacy evaluated the extent to which an intervention can prevent or considerably reduce symptoms, improve function, or decrease risk of additional complications or manifestations. Interventions with no efficacy (0 on the 0-3 scale) were not considered in this study.

Safety and Practicality

We assessed the safety and practicality of potential interventions (Table 3). We evaluated safety as the risk posed by possible adverse events and practicality as the degree of burden of the intervention. We evaluated safety based on the type of interventions. We found that preventive measures, dietary changes, and US Food and Drug Administration (FDA)–approved compounds generally scored well. In contrast, experimental interventions or those requiring special monitoring, such as gene-replacement and surgical therapies, were generally scored as riskier (Figure 2C). We identified 24 no- or low-risk, 48 mild-risk, 25 moderate-risk, and 4 high-risk interventions (Figure 2B). Overall, many available treatments feature relatively acceptable risk profiles and moderate burden for individuals, with a mean score of 1.94 (95% CI, 1.80-2.08) (Table 2).

Efficacy

We then assessed the ability of a successful intervention to reduce symptoms or avoid an unwanted outcome (Table 3). This score describes the relative impact of a successful prevention or treatment. It could be based on a quantitative measurement, such as a clinical trial outcome, or a qualitative statement by the author describing the change to the patient outcome. Intervention efficacy was not weighted by the strength of the evidence supporting its use (see below). We identified 18 highly effective, 45 moderately effective, and 38 mildly effective interventions (Figure 2B). The overall efficacy was 1.76 (95% CI, 1.65-1.87) (Table 2) suggesting that these interventions can modify the severity of the potential outcome.

Clinical Utility Overall Score

We found a mean overall clinical utility score of 6.10 (95% CI, 5.89-6.31) (Table 2). The range was 3.38-9.00 from a possible range of 0 to 9. We compared clinical utility scores between different categories of interventions and found that preventive interventions were associated with higher clinical utility scores than targets of the primary mechanism of disease (6.65 vs 5.94; Wilcox P = .02) and symptomatic management strategies (6.65 vs 5.75; Wilcox P < .001) (Figure 2D). There was no difference in clinical utility for genetic findings with primary disease targeting interventions and symptom management (5.94 vs 5.57; Wilcox P = .40). This suggests that the overall clinical utility of identifying a genetic etiology and implementing precision medicine strategies is moderate to high for CP.

Interrater Reliability

We calculated the percent agreement between scorers for severity, safety and practicality, and efficacy based on the number of scorers contributing to the mean score (eTable 3 in Supplement 1). We identified a mean of 85.5% agreement (95% CI, 81.0-89.6) (eTable 3 in Supplement X).

Levels of Evidence

Evidence was weighted on an alphabetical ranking using modified ClinGen criteria drawn from Oxford Centre for Evidence-Based Medicine guidelines (eTable 4 in Supplement 1). Notably, 56 of 101 outcome-intervention pairs (55.4%) had a C or greater level of evidence supporting a given prevention or treatment.

Discussion

In this systematic review and meta-analysis, we found that 8% of the individuals included harbored a clinically actionable genetic finding. This is comparable to the 6% identification rate of clinically actionable findings reported in individuals with epilepsy (121 of 2008).³ Other studies have evaluated broader definitions of actionability in patients with CP which included medication, dietary changes, access to clinical trials, and clinical referrals.^15,16,17,18 These studies found rates of actionability ranging from 12% to 17% of sequenced patients, suggesting our findings are quite conservative. However, those studies did not agree on definitions of personalized medicine or consider clinical impact. Therefore, this study provides a more rigorous approach, curates a starting list of genetic findings to consider, and expects benefits for patients. This has implications for the value of genetic testing in this patient population to identify personalized medicine approaches and improve patient outcomes.

In addition to the methodology used to assess the actionability of genetic findings in epilepsy, we also developed tools to evaluate the clinical utility of actionable interventions that could be used in future studies. With these tools, we were able to quantitatively assess clinical utility to help further capture the impact that personalized medicine strategies could have on ongoing clinical care. Our results indicate that the interventions target relevant aspects of the disorder, are reasonably safe and acceptable, and have the potential for clinically significant impact. Preventive strategies yielded higher safety and clinical utility scores, suggesting that early identification of genetic etiologies will be necessary for optimal outcomes.

Despite the potential clinical utility of actionable findings, most interventions are typically off label. Although this is not unusual in pediatric neurology practice,¹⁹ it is notable that the evidence base for 44.6% of the actionable interventions is limited, usually derived from small cohort studies, defined as class D or lower (eTable 4 in Supplement 1). This highlights a need for additional studies of potential precision medicine approaches in CP using rigorous and reproducible designs, especially including pediatric cohorts. This is particularly important given that additional CP-associated genes are being discovered rapidly^1,20 and that the list of potentially actionable genes is expected to grow in parallel. N-of-1 treatment trials have recently been proposed²¹ as one possible solution to the evidence gap for neurological disorders, particularly individually rare neurodevelopmental disorders.²² Earlier treatment for neurodevelopmental disorders such as CP may be more impactful than later interventions.

Limitations

There are several important limitations to this work. First, this approach likely missed potential interventions reported in the literature that did not meet our search criteria and those undergoing approval at the time of analysis (eg, ganaxolone for seizures due to variants in CDKL5).²³ Periodically updating analyses like ours will likely increase the count of actionable findings. We also did not consider other changes to clinical management, such as referral to specialists. Thus, our findings likely underestimate the number of individuals with CP who would benefit from genetic testing. Second, clinical presentations can vary considerably from one individual to the next despite a common genetic etiology. This can affect the applicability of actionable findings for a given gene. The severity of clinical symptoms may also vary from patient to patient. We recognized this inherent challenge and sought to emphasize typical presentations and symptom severity whenever possible and to strike a balance when considering possibilities rather than focusing on extremes. We used a modified Delphi process, using frequent consensus-building discussions to clarify points of contention or discrepancies in scoring. Accordingly, we did not conduct a risk-benefit analysis for each outcome-intervention pair but evaluated metrics of clinical utility instead.

We included variants in genes that may result in phenotypes that are not associated with CP, including progressive or regressive phenotypes and nonneurodevelopmental features. Many of these disorders can be difficult to differentiate based on early clinical features and these genes are often included in genetic panels to assist in differential diagnosis. Given that such cases could be incompatible with the international consensus definition of CP,²⁴ doing so may invite criticism that such individuals do not truly have CP. Conversely, atypical clinical phenotypes resembling CP within a broader physical neurodisability umbrella are increasingly recognized in related genetic disorders, such as hereditary spastic paraplegia.²⁵ Considering such factors, we have retained these cases in our analysis because they were identified in research cohorts clinically ascertained as having CP or a clinical cohort referred based on suspicion of CP. This reflects real-world clinical practice where incomplete histories, longitudinal follow-up, or other factors can prompt a reevaluation.

Exome sequencing or genome sequencing has recently been recommended by the American College of Medical Genetics and Genomics as a first- or second-tier test for individuals with congenital anomalies, (global) developmental delay, or intellectual disability, based on the Grading of Recommendations Assessment, Development and Evaluation (GRADE) evidence framework.²⁶ In the future, the methods and classification scheme used in this work could be readily extrapolated to conduct a similar analysis. We anticipate actionability will improve over time as the degree of evidence improves with additional case studies and clinical trials for interventions based on genetic etiologies. Thus, this study provides both a framework and initial data to support such future efforts to guide providers caring for individuals with CP.

Conclusions

There is an increasing emphasis on early diagnosis and intervention for CP for optimal outcomes.²⁷ Although early intervention generally has an occupational, physical, or speech therapy connotation,²⁸ early intervention may increasingly reflect precision medicine–based treatments as well. A heuristic view of CP as a nongenetic disorder could prevent patients from accessing genetic testing. Our findings build on prior work that indicates that identifying specific etiologies of CP may change clinical care.^10,29 Early use of genetic sequencing could lead to optimal clinical outcomes by offering therapies tailored to the individual before complications develop, potentially reducing lifetime care costs for families, payers, and society, which can be considerable.³⁰ In addition, genetic diagnoses have potential value beyond precision medicine, with impacts that include a sense of closure, family planning, and access to the rare disorders community.

Supplement 1.

eMethods

eAppendix. Additional web resources

eFigure

eTable 1. Definitions used in study

eTable 2. List of 243 genes and number of patients with P/LP variants in 1841 individuals sequenced in either clinical or research setting included in the analysis

eTable 3. Interrater reliability

eTable 4. Grade of evidence

Supplement 2.

eTable 5. Outcome-interventions

eTable 6. Summary statistics

eTable 7. Gene categories

eTable 8. Clinical trials underway

eTable 9. Druggable genome

Supplement 3.

Data sharing statement