Clinical Utility of Rapid Whole-Genome Sequencing in Hospitalized Adults With Unexplained Neurologic Presentations

Abstract

Background and Objectives

Adults with unexplained neurologic presentations often undergo extensive evaluations without timely diagnosis. Evidence supporting the clinical utility of rapid whole-genome sequencing (rWGS) in hospitalized adult populations remains limited. We evaluated the diagnostic yield of rWGS in adults hospitalized for unexplained neurologic manifestations and assessed clinical predictors of a phenotype-concordant genetic diagnosis.

Methods

We performed a retrospective cohort analysis of adult inpatients (≥18 years) undergoing rWGS as part of a structured inpatient clinical genomics implementation at Mayo Clinic between June 2022 and September 2025. Testing was performed after primary team consultation and subsequent assessment by a clinical geneticist. We prespecified a neurologic cohort restricted to patients admitted to the neurology inpatient service in whom presenting neurologic phenotypes were the primary indication for hospitalization and genomics consultation. Patients with non-neurologic primary indications were excluded from this study. The primary outcome was a phenotype-concordant genetic diagnosis on rWGS determined by genotype-phenotype assessment. Analytic objectives included identification of clinical predictors of a phenotype-concordant genetic diagnosis, and a secondary outcome was rWGS-attributable changes in clinical management. Patients with and without phenotype-concordant diagnoses were compared using univariable logistic regression for categorical candidate predictors (odds ratios [ORs] with 95% CIs) and the t test for age.

Results

Among 96 adults who completed rWGS, 57 (59.4%) met criteria for the neurologic cohort (mean age 53.0 ± 18.0 years; 35.1% female). Thirteen of 57 (22.8%) received a phenotype-concordant genetic diagnosis involving IFIH1, CNBP, NOTCH1, C9orf72, FGF14, HUWE1, NLRP12, CCM2, PTPN11, FLNA, HEXA, PRNP, and ATXN8OS. Factors associated with a phenotype-concordant diagnosis included a family history of similar neurologic symptoms in first-degree or second-degree relatives (OR 7.4; 95% CI 1.9–31.5), multisystem involvement (OR 6.9; 95% CI 1.6–29.8), refractory psychiatric symptoms (OR 6.1; 95% CI 1.1–35.7), and unexplained ataxia (OR 4.0; 95% CI 1.1–15.1). rWGS directly altered clinical management in 2 cases, including initiation of immunotherapy for an NLRP12-associated autoinflammatory disorder and enrollment in a gene-therapy trial for adult-onset Tay-Sachs disease.

Discussion

In this tertiary-care inpatient cohort, rWGS identified a phenotype-concordant genetic diagnosis in nearly one-quarter of adults. Limitations include single-center design and preselection through specialized consultation, which may limit generalizability.

Introduction

Undiagnosed medical conditions are characterized by long-standing symptoms that lack clear etiology despite comprehensive clinical investigation.^1,2 These conditions affect an estimated 30 million individuals in the United States, imposing a substantial emotional, financial, and clinical burden on patients and their families.^3,4 Many of these individuals endure a prolonged diagnostic odyssey, often spanning an average of 6 years, during which they undergo multiple specialist evaluations, hospital admissions, and invasive or costly diagnostic procedures without obtaining a definitive diagnosis.⁵ The economic impact of delayed diagnosis is significant, with avoidable health care expenditures reaching nearly half a million dollars per patient.⁵ These conditions frequently arise from rare disorders that are difficult to recognize, atypical presentations of more common diseases, or yet-to-be-described syndromes. Around 80% of undiagnosed diseases stem from genetic causes, underscoring the demand for advanced diagnostics to detect genetic etiologies of complex presentations.⁴

Among individuals with undiagnosed diseases, neurologic presentations constitute a significant proportion. Data from the Undiagnosed Diseases Network involving around 200 individuals with unexplained conditions revealed that nearly one-third exhibited neurologic manifestations.⁶ In both adult outpatient and inpatient settings, unexplained neurologic symptoms are frequently encountered. Approximately one-third of patients seen in neurology clinics and approximately 4% of hospitalized patients present with neurologic findings that remain unexplained after standard evaluation.^7-10 While functional neurologic disorders explain some cases, a subset of patients exhibit progressive symptoms pointing to an underlying organic cause.

Growing evidence suggests that genetic testing provides valuable diagnostic insights for adults presenting with unexplained neurologic symptoms. Several studies have demonstrated the clinical utility of genetic analyses across a wide spectrum of neurologic manifestations, including ataxia,¹¹ dementia,¹² epilepsy,¹³ myopathy,¹⁴ and early-onset Parkinson disease.¹⁵ To date, most studies have relied on whole-exome sequencing (WES), which primarily targets protein-coding regions. By contrast, whole-genome sequencing (WGS) provides broader coverage by detecting variants in both coding and noncoding regions and improving sensitivity for single-nucleotide variants.¹⁶ Rapid WGS (rWGS) modality applies this comprehensive technology through an accelerated workflow, enabling the return of results within days, which offers timely diagnostic information in acute clinical settings. This approach is well studied in inpatient pediatric settings.¹⁷ The aim of our study was to evaluate the diagnostic yield and clinical utility of rWGS in hospitalized adults with unexplained neurologic manifestations.

Methods

Study Design

We conducted a retrospective cohort analysis of hospitalized adults who underwent rWGS as part of an institutional initiative integrating inpatient clinical genomics consultation into acute hospital care. The study period spanned from June 1, 2022, and September 15, 2025, at Mayo Clinic.

Study Population and Analytic Cohort

The source population included all hospitalized adult patients aged 18 years and older who were evaluated for rWGS through the inpatient clinical genomics service during the study period. These patients were identified by querying the institutional inpatient clinical genomics consultation registry. Electronic medical records were reviewed using the Epic electronic health record system to confirm admitting service, primary presenting manifestations, rWGS recommendation, and completion status.

From this source cohort, we constructed a prespecified neurologic analytic cohort by restricting inclusion to patients admitted under the neurology inpatient service with primary presenting neurologic phenotype as the reason for hospitalization and genomics consultation who completed rWGS. Neurology service admissions included individuals managed across all levels of care, ranging from intensive care units to general neurology inpatient wards. Patients who declined testing or underwent rWGS for non-neurologic primary indications were excluded from the analyses.

Inpatient Clinical Genomics Consultation Workflow and Eligibility for rWGS

For the purposes of this study, this workflow is described in the context of patients admitted to the neurology inpatient service and applies to the prespecified neurologic analytic cohort described above. Individuals were evaluated for rWGS through a multistep clinical process.

• Step 1. Hospitalized adults admitted to the neurology service underwent standard diagnostic evaluation directed by the primary neurology team. An inpatient clinical genomics consultation was requested when the diagnostic evaluation remained inconclusive after standard inpatient assessment, and a genetic disorder was suspected based on the clinical presentation, disease severity, or atypical features. The diagnostic workup typically included laboratory testing, neuroimaging, electrophysiologic studies, and CSF analysis when clinically indicated.

• Step 2. Patients underwent inpatient bedside evaluation by a board-certified clinical geneticist and a certified genetic counselor. This assessment included detailed phenotype characterization, review of previous diagnostic testing, and collection of personal and family history to assess the likelihood of an underlying genetic condition.

• Step 3. Appropriateness for rWGS was determined by the clinical genomics team based on clinical characteristics, feasibility within the hospitalization, and the potential for a timely molecular diagnosis to inform inpatient management, prognosis, or downstream care planning. Pretest counseling addressed the benefits and limitations of rWGS; possible test outcomes including pathogenic or likely pathogenic variants, variants of uncertain significance (VUS), and negative results; and secondary or incidental findings. rWGS was offered as part of inpatient clinical care without any direct cost to patients at the time of testing.

Variables and Definitions

Clinical and demographic data of included patients, including age, sex, presenting neurologic phenotypes, symptom onset, time from symptom onset to diagnosis for patients with positive rWGS findings, family history of similar symptoms, extraneurologic systemic features (multisystem involvement), presence of refractory psychiatric symptoms, neuroimaging findings, variant classifications, and genotype-phenotype correlations, were systematically collected. All data were securely stored in a Health Insurance Portability and Accountability Act–compliant Research Electronic Data Capture database.

Presenting neurologic phenotypes were defined as the neurologic signs and symptoms prompting hospitalization and inpatient neurology evaluation, which were systematically encoded using Human Phenotype Ontology terms based on admission documentation, neurology notes, and clinical genomics evaluations. Symptom onset was defined as the earliest documented time point at which the patient first experienced symptoms attributable to the presenting neurologic phenotype that ultimately prompted neurology evaluation and inpatient clinical genomics consultation, based on review of admission history, neurology consultation notes, primary care and outpatient documentation, and referring clinician notes within the electronic health record.

Family history of similar symptoms was defined as a documented history of similar neurologic manifestations in first-degree or second-degree relatives, as recorded in the structured family history section of the electronic health record and/or in neurology or clinical genomics consultation notes.

Extraneurologic systemic features (multisystem involvement) were defined as ≥1 clinical finding outside the nervous system accompanying a primary unexplained neurologic presentation, as documented in specialty consultation notes within the electronic health record. These features were described by consulting specialists as atypical, severe, unusual for age, or not fully explained by common chronic adult conditions. Common adult comorbidities, including hypertension, hyperlipidemia, type 2 diabetes mellitus, and obesity, were not counted as extraneurologic systemic features. Refractory psychiatric symptoms were defined as psychiatric manifestations, including mood, psychotic, anxiety, behavioral, or suicidal symptoms, described by the treating neurology or psychiatry teams as persistent or atypical, with lack of meaningful response to previous standard psychiatric interventions.

Neuroimaging findings were defined as the presence of 1 or more abnormalities on brain MRI or CT that were considered clinically relevant to the presenting neurologic syndrome by the treating neurology team, as documented in imaging studies, radiology reports, and clinical notes. These included disproportionate brain atrophy for age, white matter abnormalities, cortical or subcortical signal changes, basal ganglia involvement, or structural abnormalities suggestive of a neurodegenerative, inflammatory, or metabolic process. Incidental or age-related findings without clinical relevance were not counted.

Specific data elements were recorded as missing when not documented in the electronic health record. Overall missingness was minimal (<5% for all variables), and complete-case analysis was performed on a per-variable basis.

Genetic Testing

rWGS was performed by one of several collaborating CLIA-certified clinical laboratories, including Variantyx (Comprehensive Rapid Genome Testing | Variantyx), Rady Children's Institute for Genomic Medicine (Clinical Genome Services - Rapid Whole Genome Sequencing | RCIGM), or Mayo Clinic Laboratories (WGSDX - Overview: Whole Genome Sequencing for Hereditary Disorders, Varies).

Genomic results were issued as finalized clinical reports and interpreted by the reporting laboratories in accordance with American College of Medical Genetics and Genomics and Association for Molecular Pathology guidelines.¹⁸ Variant confirmation and quality assurance procedures, including orthogonal validation when indicated, were performed according to each laboratory's standard operating procedures before clinical reporting. Reported variants were subsequently reviewed by the inpatient neurology and clinical genomics teams to assess phenotype-genotype concordance. This study analyzed finalized clinical reports and did not include independent bioinformatic reanalysis, experimental validation, or variant reclassification.

Outcomes

The primary outcome was a phenotype-concordant genetic diagnosis identified by rWGS. A phenotype-concordant diagnosis was defined as a genomic finding that was judged by the inpatient clinical genomics and neurology teams to plausibly explain the patient's primary neurologic presentation after clinical correlation and structured genotype-phenotype assessment. This determination incorporated review of the rWGS report in the context of the patient's presenting neurologic features, disease course, and previous diagnostic evaluation.

The primary analytic objective was to identify clinical predictors associated with a phenotype-concordant genetic diagnosis. Candidate predictors were prespecified and included age, sex, presenting neurologic phenotype, family history of similar symptoms, neuroimaging findings, presence of refractory psychiatric symptoms, and extraneurologic systemic features accompanying the primary neurologic presentation.

The secondary outcome included the clinical impact of rWGS through structured chart review. A change in clinical management was defined as any rWGS-attributable modification to inpatient or postdischarge care, including initiation, escalation, or discontinuation of disease-specific therapy; referral to subspecialty services; changes in prognostic counseling or goals-of-care discussions; or enrollment in a clinical trial. Management changes were considered attributable to rWGS when documentation explicitly linked the decision to the genetic result.

Statistical Analysis

Continuous variables were presented as mean ± SD, and categorical variables were reported as frequencies and percentages. Comparisons were performed between patients with a phenotype-concordant genetic diagnosis and those without a phenotype-concordant diagnosis using an unpaired t test for age. Univariable binary logistic regression was performed for each categorical predictor. Odds ratios (ORs) with 95% CIs were calculated using Wald estimates. A 2-tailed p < 0.05 was considered statistically significant. All statistical analyses were performed in GraphPad Prism version 10.2.1 (GraphPad Software, San Diego, CA). Figures were generated using GraphPad Prism and BioRender (BioRender.com) as appropriate.

Standard Protocol Approvals, Registrations, and Patient Consents

This study was reviewed and approved by the Mayo Clinic Institutional Review Board (IRB #24-008081). The study involved a retrospective analysis of clinical genomic testing performed as part of clinical care. Written informed consent for rWGS was obtained from all participants or their legally authorized representatives before testing. The Institutional Review Board determined that additional consent for retrospective analysis was not required. No identifiable images, videos, or personal data of participants are included in this publication.

Data Availability

Deidentified individual participant data underlying the results reported in this article cannot be publicly shared because of ethical and legal restrictions related to patient privacy and genomic data protection. Aggregate data supporting the findings of this study are included within the article and its supplementary materials. Additional summary-level data may be made available from the corresponding author on reasonable request, contingent on institutional approvals and execution of an appropriate data use agreement. Study protocols and analytic code are available on request.

Results

A total of 99 hospitalized adult patients (61 men and 38 women) with undiagnosed conditions were evaluated by the clinical genomics team between June 1, 2022, and September 15, 2025. Three male patients declined rWGS because of patient preference. Our review did not identify a primary neurologic phenotype in any of these cases. One declined because of testing fatigue after extensive previous evaluations, and 2 declined citing a perceived lack of impact on clinical management despite discussion of potential benefits of testing. Among the 96 adults who completed rWGS, neurologic manifestations were the most prevalent primary presenting indications, identified in 57 patients (57/96; 59.4%), followed by cardiac involvement in 21 patients s(21/96; 21.9%) (Figure 1A). These 57 patients comprised the primary neurologic cohort for subsequent diagnostic yield and predictor analyses. Of these individuals, 37 (37/57; 64.9%) were male and 20 (20/57; 35.1%) were female, with a mean age of 53.0 ± 18.0 years.

Figure 1. Study Cohort and rWGS Diagnostic Outcomes.

(A) Diagram of study flow. (B) Clinical outcomes of rWGS in the neurologic inpatient cohort. rWGS = rapid whole-genome sequencing; VUS = variants of uncertain significance.

Among the 57 patients in the neurologic cohort, rWGS was performed predominantly as proband-only testing (42/57; 73.4%), with duo-based sequencing in 8 patients and trio-based sequencing in 7 patients. Limited availability of parental samples reflected the inpatient, often high-acuity clinical context, in which rapid clinical decision making and logistical constraints frequently precluded collection of additional family member samples.

Within the neurologic cohort, 25 patients (25/57; 43.9%) had reportable genomic findings identified on rWGS, including pathogenic/likely pathogenic variants (n = 15) and VUS (n = 10). After a detailed genotype-phenotype assessment, 13 individuals (13/57; 22.8%), including 8 men and 5 women, were found to have variants consistent with their primary neurologic phenotypes (Table 1). The remaining patients had either reportable but non–phenotype-concordant findings (n = 12) or negative rWGS results (n = 32) (Figure 1B), with additional details provided in eTable 1.

Table 1.

Demographic, Clinical and Imaging Features of Patients in the Neurologic Cohort With Phenotype-Concordant Genetic Diagnosis

Sex	Age group (y)	Time to diagnosis	Family history	Neurologic phenotype (clinical)	Corresponding HPO terms	Extraneurologic manifestations	Imaging findings	Gene/variant
M	60–69	7 mo	Y	Progressive cognitive decline; gait instability; dysarthria	Ataxia (HP:0001251); cognitive decline (HP:0001268); dysarthria (HP:0001260)	None reported	Normal brain imaging	IFIH1 c.2465G>A p.Arg822Gln
F	50–59	12 mo	Y	Generalized muscle weakness; spastic paraparesis; gait impairment; refractory major depressive disorder	Ataxia (HP:0001251); depression (HP:0000716); muscle weakness (HP:0001324); spastic paraparesis (HP:0001258)	None reported	Brain MRI: progressive pontine and cerebellar degeneration	CNBP CCTG repeat expansion (1 normal allele and 1 fully penetrant expanded allele with 174–514 repeats)
M	50–59	27 mo	N	Stroke	Stroke (HP:0001297)	Type A and B aortic dissection; thoracoabdominal aortic aneurysm, bicuspid aortic valve	Head and neck CTA: dissection in the bilateral carotid arteries Brain MRI: multifocal areas of enhancements in the cerebral hemispheres and left basal ganglia in favor of infarcts	NOTCH1 c.2804C>T p.Pro935Leu
F	40–49	126 mo	Y	Rapid progressive cognitive decline; cortical visual impairment; gait instability; refractory major depressive disorder	Ataxia (HP:0001251); cognitive decline (HP:0001268); cortical visual impairment (HP:0100704); depression (HP:0000716)	None reported	Brain MRI: moderate-to-advanced brain parenchymal volume loss which is disproportionate for her age	C9orf72 GGGGCC repeat >100
M	70–79	78 mo	N	Progressive gait instability; dysarthria; previous right frontal lobe hemorrhage in the context of metastatic melanoma	Ataxia (HP:0001251); dysarthria (HP:0001260)	Metastatic melanoma	Brain MRI: T1-hyperintense area compatible with blood products in the right frontal lobe and multifocal white matter hyperintensities involving the periventricular, deep, and subcortical white matter	FGF14 GAA repeat expansion (1 normal allele and 1 reduced to fully penetrant expanded allele with 257–270 repeats)
F	20–29	43 mo	Y	Seizures; behavioral dysregulation; agitation; aggression; suicidal ideation	Aggressive behavior (HP:0000718); agitation (HP:0000713); behavioral abnormality (HP:0000708); seizures (HP:0001250); suicidal ideation (HP:0000749)	None reported	Normal brain imaging	HUWE1 c.3481G>A p.Gly1161Arg
M	30–39	7 mo	N	Right-sided hemiparesis	Hemiparesis (HP:0001269)	Recurrent mucocutaneous ulcers; neutrophilic dermatosis	Brain MRI: T2 hyperintensity in the left deep gray nuclei, brainstem, and left frontal lobe encephalomalacia/gliosis with mild patchy enhancement	NLRP12 c.2761G>C p.Gly921Arg
M	70–79	8 mo	N	Dizziness; delirium; functional and cognitive decline	Cognitive decline (HP:0001268); delirium (HP:0000739)	History of esophageal cancer	Brain MRI: numerous areas of hemosiderin deposition in both cortical, subcortical, and deep gray matter areas, as well as involvement of the brainstem and cerebellum in favor of cavernous malformation	CCM2 c.612del p.Ala205LeufsTer
M	40–49	47 y	N	Seizures; bilateral sensorineural hearing loss	Seizures (HP:0001250); sensorineural hearing loss (HP:0000407)	Multiple lentigines; congenital heart malformation; undescended testis	No recorded brain imaging	PTPN11 c.1403C>T p.Thr468Met
F	40–49	63 mo	Y	Chronic severe headaches attacks	Headache (HP:0002315)	Spontaneous pneumothorax; enlarged aortic arch; coronary artery aneurysm	Brain MRI: periventricular nodular heterotopia	FLNA c.5184C>T p.Gly1728=
M	20–29	3 mo	N	Slurred speech; gait impairment; auditory hallucinations; disorganized thinking	Ataxia (HP:0001251); auditory hallucinations (HP:0008765); dysarthria (HP:0001260)	None reported	Brain MRI: significant cerebellar hemisphere and vermian atrophy	HEXA c.805G>A p.Gly269Ser (paternal) and c.677C>T p.Ser226Phe (maternal)
F	50–59	2 mo	Y	Rapidly progressive dementia; gait instability; myoclonus; abnormal posturing; dysarthria	Ataxia (HP:0001251); dementia (HP:0000726); dysarthria (HP:0001260); myoclonus (HP:0001336)	None reported	Brain MRI: cortical restricted diffusion of the right postcentral gyrus and FLAIR hyperintensity involving the right parietal lobe near the vertex	PRNP c.598G>A p.Glu200Lys
M	60–69	48 mo	Y	Progressive upper extremity weakness; gait instability; Parkinson disease; cerebral amyloid angiopathy	Ataxia (HP:0001251); muscle weakness (HP:0001324); parkinsonism (HP:0001300)	None reported	Brain MRI: several small foci of susceptibility throughout the supratentorial brain most consistent with advanced cerebral amyloid angiopathy. Single focus of susceptibility in the right cerebral peduncle of the midbrain	ATXN8OS CTG repeat expansion (1 normal allele and 1 expanded allele with 106–111 repeats)

Abbreviations: CTA = CT angiography; FLAIR = fluid-attenuated inversion recovery; HPO = Human Phenotype Ontology.

The mean age at diagnosis among 13 cases with phenotype-concordant genomic findings was 51.7 ± 17.6 years. The time from initial symptom onset to diagnosis ranged widely from 2 months to 47 years. On average, patients experienced a diagnostic delay of 6.3 ± 12.6 years. All diagnoses represented monogenic disorders, with clinically relevant variants identified in IFIH1, CNBP, NOTCH1, C9orf72, FGF14, HUWE1, NLRP12, CCM2, PTPN11, FLNA, HEXA, PRNP, and ATXN8OS (Table 1).

Among patients with phenotype-concordant genomic findings, 10 patients (10/13; 76.9%) harbored pathogenic/likely pathogenic variants and 3 had VUS. Most identified disorders followed autosomal dominant inheritance (10/13; 76.9%), followed by 2 X-linked disorders (HUWE1, FLNA) and 1 autosomal recessive condition due to compound heterozygous HEXA variants. Identified variant types included single-nucleotide variants (IFIH1, NOTCH1, HUWE1, NLRP12, PTPN11, PRNP, and HEXA) and pathogenic repeat expansions (CNBP, C9orf72, FGF14, and ATXN8OS). Underlying gene-disease mechanisms were inferred from established literature. A comprehensive summary of all genetic findings is provided in Table 2.

Table 2.

Genomic Characteristics of Identified Variants

Gene/variant	Classification	Sequencing approach	Variant type	Proposed mechanism	Inheritance	Associated OMIM phenotype
IFIH1 c.2465G>A p.Arg822Gln	Pathogenic	Singleton	Missense	Gain-of-function	AD (unknown)	Aicardi-Goutières syndrome 7 (#615846)
CNBP CCTG repeat expansion (1 normal allele and 1 fully penetrant expanded allele with 174–514 repeats)	Pathogenic	Singleton	Repeat expansion	Toxic RNA–mediated	AD (unknown)	Myotonic dystrophy type 2 (#602668)
NOTCH1 c.2804C>T p.Pro935Leu	VUS (related to patient phenotype)	Singleton	Missense	Uncertain (NOTCH1 variant not functionally established)	AD (unknown)	Aortic valve disease 1 (#109730)
C9orf72 GGGGCC repeat >100	Pathogenic	Trio	Repeat expansion	Toxic RNA–mediated	AD (inherited from mother)	Frontotemporal dementia/ALS (#105550)
FGF14 GAA repeat expansion (1 normal allele and 1 reduced to fully penetrant expanded allele with 257–270 repeats)	Likely pathogenic	Singleton	Repeat expansion	Loss-of-function	AD (unknown)	Spinocerebellar ataxia type 27 (#193003)
HUWE1 c.3481G>A p.Gly1161Arg	VUS (related to patient phenotype)	Trio	Missense	Loss-of-function (variant-level impact uncertain)	X-linked (inherited from mother)	Intellectual developmental disorder, X-linked syndromic, Turner type (#309590)
NLRP12 c.2761G>C p.Gly921Arg	VUS (related to patient phenotype)	Singleton	Missense	Loss-of-function (variant-level impact uncertain)	AD (unknown)	Familial cold autoinflammatory syndrome 2 (#611762)
CCM2 c.612del p.Ala205LeufsTer	Pathogenic	Singleton	Frameshift	Loss-of-function (haploinsufficiency)	AD (unknown)	Cerebral cavernous malformations 2 (#607929)
PTPN11 c.1403C>T p.Thr468Met	Pathogenic	Singleton	Missense	Gain-of-function (RAS/MAPK hyperactivation)	AD (unknown)	Noonan syndrome 1 (#163950)
FLNA c.5184C>T p.Gly1728=	Likely pathogenic	Singleton	Synonymous	Loss-of-function (splice disruption)	X-linked (unknown)	Periventricular nodular heterotopia (#300049)
HEXA c.805G>A p.Gly269Ser (paternal) and c.677C>T p.Ser226Phe (maternal)	Pathogenic	Trio	Missense (compound heterozygous)	Loss-of-function (HexA enzyme deficiency)	AR (inherited from mother and father)	Tay-Sachs disease (#272800)
PRNP c.598G>A p.Glu200Lys	Pathogenic	Duo (mother)	Missense	Gain-of-function (misfolded prion protein)	AD (unknown)	Creutzfeldt-Jakob disease (#123400)
ATXN8OS CTG repeat expansion (1 normal allele and 1 expanded allele with 106–111 repeats)	Pathogenic	Singleton	Repeat expansion	Toxic RNA–mediated	AD (unknown)	Spinocerebellar ataxia type 8 (#608768)

Abbreviation: AD = autosomal dominant; ALS = amyotrophic lateral sclerosis; AR = autosomal recessive; MAPK = mitogen-activated protein kinase OMI = online mendelian inheritance in man; RAS in RAS/MAPK pathway = rat sarcoma / mitogen-activated protein kinase pathway; VUS = variants of uncertain significance.

Comparisons between individuals with rWGS findings concordant with their primary neurologic phenotypes (n = 13) and those without phenotype-concordant findings (n = 44) revealed multiple significant differences. Individuals with phenotype-concordant genomic findings had higher odds of multisystem involvement (46.1% vs 11.4%; OR 6.9, 95% CI 1.6–29.8, p < 0.01), refractory psychiatric symptoms (30.8% vs 6.8%; OR 6.1, 95% CI 1.1–35.7, p: 0.03), a family history of similar neurologic symptoms (53.8% vs 13.6%; OR 7.4, 95% CI 1.9–31.5, p < 0.01), and unexplained ataxia (53.8% vs 22.7%; OR 4.0, 95% CI 1.1–15.1, p = 0.04) (Figure 2). By contrast, age, sex, and neuroimaging findings did not significantly differ between groups (Table 3).

Figure 2. Comparison of Clinical Predictors Between Participants With and Without Phenotype-Concordant Genetic Diagnosis in the Neurologic Inpatient Cohort.

Participants with phenotype-concordant genetic diagnosis had significantly higher odds of extraneurologic systemic features (multisystem involvement) (46.1% vs 11.4%; odds ratio [OR] 6.9, 95% CI 1.6–29.8), refractory psychiatric conditions (30.8% vs 6.8%; OR 6.1, 95% CI 1.1–35.7), similar neurologic manifestations in first-degree or second-degree relatives (53.8% vs 13.6%; OR 7.4, 95% CI 1.9–31.5), and unexplained ataxia (53.8% vs 22.7%; OR 4.0; 95% CI 1.1–15.1). *p < 0.05, **p < 0.01.

Table 3.

Comparison of Demographic and Clinical Features Between Groups

Variable	Genetically diagnosed (n = 13)	No genetic diagnosis (n = 44)	Model	Effect size (95% CI)	p Value
Age (y), mean ± SD	51.7 ± 17.6	53.3 ± 18.3	Unpaired t test	Mean difference 1.6 (−9.8 to 13.4)	0.77
Sex (male %), n (%)	8 (61.5)	29 (65.9)	Logistic regression	OR 0.8 (0.2–3.1)	0.77
Family history, n (%)	7 (53.8)	6 (13.6)	Logistic regression	OR 7.4 (1.9–31.5)	<0.01
Multisystem involvement, n (%)	6 (46.1)	5 (11.4)	Logistic regression	OR 6.9 (1.6–29.8)	<0.01
Refractory psychiatric symptoms, n (%)	4 (30.8)	3 (6.8)	Logistic regression	OR 6.1 (1.1–35.7)	0.03
Ataxia, n (%)	7 (53.8)	10 (22.7)	Logistic regression	OR 4.0 (1.1–15.1)	0.04
Epilepsy, n (%)	2 (15.3)	7 (15.9)	Logistic regression	OR 0.9 (0.1–4.7)	0.90
Brain atrophy, n (%)	3 (23.0)	4 (9.1)	Logistic regression	OR 3.0 (0.2–15.9)	0.20

Abbreviation: OR = odds ratio.

Management decisions were influenced by both the underlying genetic diagnosis and the time interval between symptom onset and rWGS confirmation. In 2 patients, the genetic findings led to substantial immediate changes in treatment and care planning. The first involved a patient with a NLRP12 variant, who had presented with recurrent mucocutaneous ulcers, neutrophilic dermatosis, and right-sided hemiparesis associated with left frontal inflammatory lesions. After the genetic diagnosis, established 7 months after symptom onset, immunotherapy with infliximab was initiated at our institution, resulting in marked improvement in neurologic manifestations and prevention of further inflammatory episodes. The second case involved a patient with biallelic HEXA variants causing Tay-Sachs disease, which was diagnosed 3 months after initial symptoms of slurred speech, auditory hallucinations, and gait impairment. Identification of the underlying disorder enabled referral to an external center for enrollment in a gene-therapy clinical trial and facilitated disease-specific counseling, psychiatric medication adjustment to avoid agents with higher neurologic risk, and early diagnostic evaluation of an affected twin sibling.

In additional cases, rWGS contributed to clinically meaningful management decisions beyond immediate therapeutic intervention. In a patient with familial cerebral cavernous malformation, genetic diagnosis led to initiation of propranolol as part of anticipatory management. In another case, identification of Creutzfeldt-Jakob disease clarified prognosis and directly guided goals-of-care discussions with the patient's family.

Discussion

Genomic sequencing has been studied in pediatric populations more extensively than in adults. A recent meta-analysis of over 100 studies reported that genome sequencing yields a molecular diagnosis in approximately 30% of pediatric patients with suspected genetic disorders.¹⁷ This diagnostic rate reaches around 50% in critically ill neonates and infants admitted to intensive care units.¹⁹ By contrast, evidence in adult populations remains comparatively limited. Most published adult studies have focused on disease-specific cohorts including hereditary ataxias, inherited neuropathies, dementia syndromes, or neuromuscular disorders.^20-23 Data on the use of rWGS in hospitalized adults presenting with unexplained neurologic manifestations remain limited. The diagnostic yield within the neurologic cohort is around 23%, which shows the potential value of genome-wide evaluation in undiagnosed, complex neurologic presentations in adults.

Different adult genomic studies have reported the nervous system as the most commonly affected system among patients with undiagnosed conditions, which is in line with our findings. In 4 large-scale studies neurologic presentations accounted for more than half of all referrals, ranging from approximately 55% to 80% of enrolled participants.^24-27 Among adults with neurologic phenotypes, reported diagnostic yields of genetic tests ranged from 20% to 30%. Further studies focusing exclusively on adults with unexplained neurologic disorders have shown clinically meaningful diagnostic yields between 10% and 30%.^28,29 Additional data on previous studies are listed in eTable 2. The variability in diagnostic yield across studies likely reflects methodological differences, including heterogeneity in patient selection, clinical setting, and the type of conducted genomic testing.

A recent large retrospective study evaluated exclusion-based WES in critically ill adults admitted to the intensive care unit, in which testing was applied broadly to eligible patients without preselection based on genetics consultation. This study reported a diagnostic yield of approximately one-quarter of patients, with all diagnostic findings representing monogenic disorders.³⁰ Despite fundamental differences in cohort composition, clinical context, and implementation strategy, this diagnostic yield is highly similar to that observed in our inpatient genetics-evaluated cohort. Together, these findings suggest that monogenic disease contributes meaningfully to adult inpatient morbidity across diverse hospital settings, and that comparable diagnostic yields can be achieved through distinct implementation models.

Repeat-expansion disorders represented a substantial proportion of diagnoses in our cohort (4 of 13 cases, 31%), including pathogenic expansions in CNBP, ATXN8OS, C9orf72, and FGF14. These conditions exemplify a key advantage of rWGS over WES. Most pathogenic repeat expansions reside in intronic or other noncoding regions that are not captured by exome enrichment, such as the C9orf72 hexanucleotide repeat located in the first intron.³¹ In addition, the high GC content, length, and repetitive nature of short tandem repeat loci cause poor capture efficiency and misalignment in WES, which prevents accurate sizing of expansions that exceed short-read length.^32,33 Notably, our case of spinocerebellar ataxia type 8 (SCA8) involved a CTG repeat expansion in ATXN8OS, which is particularly interesting as this gene encodes a noncoding RNA.³⁴ This pathogenic expansion causes disease through RNA-mediated toxicity rather than protein dysfunction. Expansion of the ATXN8OS transcript leads to the formation of toxic RNA foci, sequestration of RNA-binding proteins, and widespread dysregulation of alternative splicing, which are hallmark mechanisms in SCA8.³⁵ This case further shows the diagnostic power of rWGS in identifying pathogenic alterations that are inaccessible to WES.

A positive family history of similar neurologic manifestations in a first-degree or second-degree relative was the strongest predictor of a relevant genetic diagnosis in our cohort, with an odds ratio exceeding 7. Previous studies also demonstrated substantially higher diagnostic yields among individuals with familial clustering of similar neurologic symptoms.^24,26 In our study, unexplained ataxia was the most common neurologic presentation, a finding that is consistent with previous genomic studies.²⁵ Multiple investigations of adults with unexplained or progressive ataxia, using different testing modalities have consistently reported diagnostic yields exceeding 30% and in some cohorts reaching 50%.^20,36,37 By contrast, studies focusing on adults with other unexplained neurologic presentations, such as neuromuscular disorders, dementia, epilepsy, or neuropathies, have generally demonstrated substantially lower yields, typically below 15%.^21-23,38 Our study showed that unexplained ataxia is a strong predictor of identifying a clinically relevant genetic diagnosis.

Unexplained extraneurologic systemic features including cardiac manifestations, malignancy, and cutaneous lesions, as well as refractory psychiatric symptoms such as mood disorders, hallucinations, and suicidal ideation, were significantly more frequent among our cases with phenotype-concordant genomic findings than among those without a relevant rWGS diagnosis. This pattern is consistent with large sequencing cohorts showing that adults with multiorgan involvement are substantially more likely to receive a molecular diagnosis,²⁴ highlighting the importance of considering a genetic etiology when neurologic symptoms coexist with systemic involvement. In addition, a recent scoping review reported that rare genetic disorders frequently present with prominent neuropsychiatric features, including hallucinations, severe mood symptoms, and suicidality.³⁹ The review also reported that individuals with these presentations often experience considerable diagnostic delays, with an average time to diagnosis of nearly 8 years, along with substantial health care–related economic burden.³⁹ In our cohort, the time to diagnosis among the 4 patients with refractory psychiatric manifestations ranged from 3 months to more than 10 years. Early rWGS in one of these patients led to the diagnosis of adult-onset Tay-Sachs disease caused by pathogenic variants in the HEXA gene only 3 months after symptom onset. Two of the 4 patients with refractory neuropsychiatric presentations carried repeat-expansion disorders involving the CNBP and C9orf72 genes. Although the mechanisms linking repeat expansions with psychiatric symptoms are not fully understood, this association has been frequently described in the literature.^40-42

In our cohort, 12 patients had reportable but non–phenotype-concordant rWGS findings, including both pathogenic/likely pathogenic variants and VUS. These results were addressed through inpatient or postdischarge genetics follow-ups, conducted either at the bedside or through virtual visits with patients and/or their families. During these discussions, results were interpreted in clinical context, with attention to patient understanding and preferences, including review of the phenotypic spectrum typically associated with the identified variant and explicit explanation of why the patient's presentation was not considered concordant with the genetic finding. For pathogenic/likely pathogenic non–phenotype-concordant variants, counseling focused on potential longer-term implications and appropriate outpatient follow-up when relevant. VUS were discussed with particular emphasis on uncertainty, lack of immediate actionability, and the possibility of future reinterpretation. This approach reflects the ethical balance required in inpatient genomic testing between transparent disclosure and avoidance of unnecessary anxiety during acute illness.

An important implementation question in adult inpatient genomics is whether rapid sequencing should be directly ordered by neurologists vs routed through genetics consultation. While protocolized ordering may improve scalability, our experience supports maintaining inpatient or virtual clinical genomics involvement to guide test selection, detailed phenotyping, consent, and result interpretation. In adults with complex neurologic presentations, genetics expertise is essential to avoid inappropriate testing and misinterpretation of uncertain findings. Workflow bottlenecks may be mitigated by embedding clinical genomics within inpatient teams or by leveraging tele-genetics models.

There are several limitations to this study. First, this was a single-center study conducted at a tertiary referral institution with a dedicated inpatient clinical genomics consultation service. Patients underwent rWGS only after a multistep evaluation by neurology and clinical genomics teams, resulting in a highly selected cohort enriched for diagnostically challenging cases. As such, the diagnostic yield and predictors identified here may not be generalizable to unselected adult inpatients or to settings without specialized inpatient genomics infrastructure. Second, although rWGS was implemented through a structured inpatient clinical genomics program, our analysis was retrospective and, therefore, subject to inherent limitations, including potential selection bias and reliance on documentation from routine clinical care. Third, our sample size, particularly within the phenotype-concordant genomic subgroup was small, which reduces the power to detect additional predictors of positive findings. Fourth, our analytic cohort was restricted to patients admitted under the neurology service with a primary neurologic indication for hospitalization and genomic evaluation. Adults who underwent rWGS under other admitting services were excluded. Review of these cases did not identify unexplained primary neurologic phenotypes. Accordingly, our findings are most generalizable to inpatient neurology populations and may not directly extend to all hospitalized patients with neurologic symptoms. Finally, a negative rWGS result does not exclude an underlying genetic disorder. Several factors contribute to false-negative findings, including limited detection of deep intronic variants, low-level mosaicism, epigenetic disorders, small structural variants, and pathogenic variants in genes not yet associated with human disease. In addition, the interpretation of novel or rare variants remains constrained by incomplete knowledge of gene-phenotype relationships. Therefore, continued clinical follow-up and periodic reanalysis of genomic data is needed, even when rWGS is nondiagnostic.

Future studies should explore the integration of rWGS earlier in the diagnostic pathway for adults with unexplained neurologic presentations, particularly those with a positive family history, ataxia, presence of refractory psychiatric symptoms, or multisystem involvement. Earlier genomic diagnosis may enable timely access to emerging precision therapies, including individualized gene-based treatments such as DNA replacement technologies and gene augmentation, as well as RNA-targeted strategies such as antisense oligonucleotide therapies. These personalized molecular approaches have shown promise across a range of neurogenetic disorders and are likely to be most effective when initiated before irreversible neurologic injury occurs.

Another important future consideration is the cost and accessibility of rWGS. Although sequencing costs have decreased substantially over the past decade, the price of clinical rWGS remains considerable, typically ranging from $3,500 to $9,000 per test.^43-45 Limited insurance coverage for adult patients further constrains implementation. As sequencing becomes more affordable and reimbursement policies evolve, earlier use of rWGS has the potential to reduce overall healthcare expenditures by shortening diagnostic odysseys, avoiding unnecessary investigations, and facilitating timely initiation of targeted treatments. Prospective, multicenter studies will be crucial for determining how earlier genomic testing can be optimized to improve clinical outcomes and health-system efficiency in adults with rare neurologic disorders.

These findings support the incorporation of rWGS into the diagnostic evaluation of adults with complex neurologic presentations and highlight the potential for genomics to shorten the diagnostic odyssey and direct patients toward emerging precision therapies.

Glossary

OR

odds ratio

WES

whole-exome sequencing

WGS

whole-genome sequencing

rWGS

rapid WGS

SCA8

spinocerebellar ataxia type 8

VUS

variants of uncertain significance

Author Contributions

M. Amanat: drafting/revision of the manuscript for content, including medical writing for content; analysis or interpretation of data. M. Toledano: major role in the acquisition of data. L.A. Schimmenti: major role in the acquisition of data. P.N. Pichurin: major role in the acquisition of data. B.C. Lanpher: major role in the acquisition of data. D.R. Deyle: major role in the acquisition of data. Q.K.G. Tan: major role in the acquisition of data. G. Iverson: major role in the acquisition of data. A. Cera: major role in the acquisition of data. R. Gavrilova: major role in the acquisition of data. R. Dhamija: major role in the acquisition of data; study concept or design.

Study Funding

The authors report no targeted funding.

Disclosure

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