Childhood-Onset Neurodegeneration and Brain Atrophy: Defining UBTF-Related Developmental Regression and Progressive Ataxia

Abstract

Background:

The heterozygous variant c.628G>A (p.Glu210Lys) in UBTF causes Childhood-Onset Neurodegeneration with Brain Atrophy (CONDBA) (OMIM # 600673), characterized by early normal or mildly delayed development followed by regression, with individuals frequently experiencing movement disorders. This study defines the natural history of this rare disorder and explores potential biomarkers in humans and mice.

Methods:

Caregivers of individuals with CONDBA completed cross-sectional surveys detailing genetic, developmental, and clinical features. Patients evaluated in a neurogenetics clinic underwent Brief Ataxia Rating Scale (BARS) assessments, compared with remotely collected wrist and ankle accelerometry data. Neurofilament light chain (NFL) levels were assessed in the clinic cohort and in a Ubtf E210K knock-in mouse model.

Results:

All 11 caregiver surveys reported onset of neurodevelopmental regression (median 3.5 years, range 0.5-5 years), at times following anesthesia or illness, and 82% developed ataxia. Motor activity data from 5 participants (median 11.8 years, range 8.1-12.5 years) had high test-retest reliability, correlated with ataxia severity as measured by the BARS, and showed declines across multiple measures over the study period. NFL was abnormally elevated in both humans and the mouse model.

Conclusion:

UBTF-related CONDBA presents with early normal or mildly delayed development followed by regression and progressive ataxia. Wearable accelerometers provide a reliable measure of disease severity, and elevated NFL may serve as a biomarker of neuronal injury in both humans and animal models.

INTRODUCTION

Hereditary ataxias comprise a spectrum of neurogenetic disorders characterized by impaired balance, coordination, and progressive motor decline^1,2. Although movement disorder is the defining feature, regression across cognitive and language domains, epilepsy, and additional neurologic symptoms are observed.

Heterozygous gain-of-function variants in the upstream binding transcription factor ( UBTF) gene have been linked to Childhood-Onset Neurodegeneration with Brain Atrophy (CONDBA), a rare neurodegenerative disorder featuring initially typical or mildly delayed development followed by global regression in the first decade of life^3-11. Affected individuals frequently experience multiple types of movement abnormalities, including cerebellar (ataxia), pyramidal (spasticity), and extrapyramidal (dystonia). Seizures occur in approximately one-third of cases, and microcephaly may develop over time¹². By adolescence, most patients experience profound regression, becoming non-verbal and non-ambulatory with severe intellectual disability (ID).

Upstream Binding Factor (UBF), the protein product of UBTF, is a ubiquitously expressed transcription factor for RNA Polymerase I (RPI) and is necessary for ribosomal transcription in the nucleolus^4,13,14. A recurrent heterozygous variant in UBTF (c.628G>A, p.Glu210Lys) has been identified in all 19 individuals expressing the CONDBA phenotype ^4,8. This substitution of glutamate with lysine in the conserved second high mobility group (HMG)-box homology domain⁴ alters the electrostatic interaction with DNA and disrupts cooperation with Selectivity Factor 1 (SL1), a component of the RPI preinitiation complex, thereby impairing ribosomal transcription and neuronal homeostasis¹³. A novel variant (c.608A>G (p.Gln203Arg)) has been described in one case of more severe neurodevelopmental impairment¹⁵. Recently, 9 individuals with UBTF haploinsufficiency have been identified. Developmental delay/intellectual disability have been described in these individuals but without the neurodevelopmental regression typically seen in CONDBA^16-18.

To delineate the developmental trajectory and disease progression of UBTF-related disorder, we combined caregiver surveys, clinical ataxia ratings, wearable accelerometry data, and serum neurofilament light chain (NFL) analysis in humans and a knock-in mouse model. Our goal was to define the natural history of CONDBA and identify feasible biomarkers for use in future interventional trials.

METHODS

Participants

Participants were recruited from a parent-run, pre-existing, closed Facebook group (“Families with UBTF Genetic Mutation”). The survey link was distributed to all group members via Facebook. Parents/guardians of an individual (age 6 months to 40 years) with a UBTF genetic mutation were eligible. All Facebook group members were English-speaking. Participants needed an active email account to participate.

Patients with UBTF-related disorder were recruited from a neurogenetics clinic at Massachusetts General Hospital (MGH) for longitudinal motor activity assessment. All individuals with a diagnosis of UBTF-related disorder with genetic confirmation were eligible for this study regardless of ambulatory or functional status. Participants completed continuous one-week monitoring sessions using wearable devices, with repeat sessions every three months. Clinical severity was assessed during in-person visits using the Brief Ataxia Rating Scale (BARS)¹⁹. Unaffected siblings of individuals with ataxia-telangiectasia served as controls for the activity measurements. Males with a diagnosis of X-linked Adrenoleukodystrophy but without neurologic involvement (including either cerebral adrenoleukodystrophy or adrenomyeloneuropathy) served as controls for the NFL measures.

Survey Collection

A comprehensive survey was developed at MGH based on previously-published information on UBTF and expertise in related disorders^3,4,7,8. The survey and study description were distributed to families by a parent moderator of the Facebook group. The study was approved by the Mass General Brigham Human Research Committee (Protocol #2021P000518).

The “Survey for Families of Patients Identified as Having UBTF Mutation” comprised 15 sections covering demographics, family history, perinatal history, diagnosis, motor and communication development, neurologic and systemic findings, sleep, pain, hospitalizations, medications, and open-ended comments. The survey administered to participants is provided in supplementary file 1. Surveys could be completed over multiple sessions using a unique study ID and password. Study data were collected and stored in REDCap (Research Electronic Data Capture) hosted at MGH^20,21.

Wearable Device Data Collection

This component was approved by the same ethics committee (Protocol #2019P002752). Participants received GENEActiv Original actigraphy devices by mail and were asked to wear sensors on the dominant wrist and ankle for one week of continuous data collection as described in prior literature²². Devices were then mailed back for data download and reissued every three months.

Accelerometer data were processed to obtain submovement and activity intensity (AI) features as previously reported^22-25. Analysis focused on two validated motor metrics: Entropy of AI, which reflects the range of activity intensity over recording week, and the ALS wrist composite score²⁴, a machine-learning measure sensitive to disease progression. Both metrics were compared with controls and correlated with BARS scores.

Mouse models

The congenic B6J.Cg-Ubtf^tm1.1Ldx/Mmmh mice (referred to as B6J.Ubtf ^E210K or Ubtf E210K knock-in mice) were previously developed by Dr. Mark LeDoux¹³ (MMRRC #50519, MGI:6695077). Genotyping was performed at birth, weaning (P28), and endpoint using the following primers:

Common Forward: 5’CCT TTC CTC TGT CCC TAC CG’3

Common Reverse: 5’CTG CAA TAT GGT GTC CTG CT’3

WT Probe (HEX): 5’CCC GAG AGA TCT CAG GCT A’3

E210K mutant Probe (FAM): 5’CCT ATA CGA AGT TAT AAT ATT CTC AGG C’3

The Jackson Laboratory Animal Care and Use Committee approved all mouse protocols.

Serum collection and NFL analysis (mouse)

Serum collection and NFL analysis were performed as previously described²⁶. Whole blood was obtained by retro-orbital collection under topical anesthesia. Serum was separated by centrifugation and stored at −80°C. NFL levels were quantified using the Simoa HD-X Analyzer (Simoa NF-Light v2 Advantage Kit #104073 by Quanterix) at a 32× dilution with technical replicates per sample.

Serum/Plasma collection and NFL analysis (human)

Serum and/or plasma samples were obtained from individuals with a diagnosis of UBTF-related disorder seen at MGH. During the study, analysis transitioned from plasma NFL (Mayo Clinic Laboratories) to serum NFL (Labcorp) in March 2025 due to a laboratory change.

Statistics

Descriptive statistics were generated using Microsoft Excel. Accelerometer and correlation analyses were conducted in Matlab. Pearson correlation coefficients assessed relationships between wearable-derived motor metrics and BARS scores, excluding controls to avoid inflated correlations. Group differences between UBTF and control participants were analyzed using the Mann-Whitney U-test. Reliability within each one-week recording period was evaluated using intraclass correlation coefficients (ICC, 2-way mixed effect model) by comparing days 1-3 to days 4-6. Sensitivity to change for each motor measure was evaluated by computing the slope of the measure over time for each participant and then computing the mean to standard deviation ratio (MSDR) of the measure’s slopes over the population.

Mouse serum NFL data were analyzed in GraphPad using two-way ANOVA, and statistical tests were corrected for multiple comparisons via Tukey correction. Quantifications were performed with mice of either sex from two independent mouse cohorts (mix of genotypes and sexes). The presented NFL data reflect the pool of data obtained from both cohorts since significant changes were cohort independent. The number of biological replicates is depicted in the figure.

RESULTS

Participants

Fifty-six caregiver surveys were initiated, of which 45 were not completed. The majority of these incomplete surveys were left entirely or mostly blank (n = 30) or were duplicate entries (n = 8). We suspect that these surveys were generated by respondents who returned to the survey at a later time and started a new survey rather than using the provided password to access and edit the original survey. Eleven complete responses describing 11 unique individuals were included in the analysis. The median age of the affected patient at the time of survey completion by their parent/guardian was 11 years (range 2-20 years), and 55% were male. (Table 1). The median age at genetic diagnosis was 9.9 years (range 1.7-20.8 years). Two subjects (18%) were adopted.

Table 1:

Survey Sample Characteristics (N = 11)

		n	Percentage
Female		5	45
Male		6	55
Age	0-5 years	2	18
	6-10 years	2	18
	11-15 years	4	36
	16-20 years	3	27
Adopted		2	18

Median age 9.9 years at diagnosis (range 1.7- 20.8 years)

Genetics

Eight individuals (73%) provided genetic data; seven carried the recurrent NM_014233.3:c.628G>A (p.Glu210Lys) variant and one a novel NM_014233.3:c.495C>A (p.Phe165Leu) variant, maternally inherited from a mosaic mother with developmental delay. Nine participants (82%) underwent exome sequencing. Four (36%) reported additional gene variants, including one pathogenic SDHA variant (associated with hereditary paraganglioma-pheochromocytoma), one Fragile X premutation, and compound variants in PCDHB10, NUMA1, and POTEF. Only the SDHA variant was considered disease-causing.

Pregnancy and Perinatal Period

Two pregnancies (18%) were complicated by preterm labor, maternal infection, or substance use. Ten infants (91%) were born at term; one was premature at 31 weeks. Two deliveries required cesarean section. Neonatal complications occurred in four cases (36%) and included NICU stay, jaundice, poor feeding, hypoglycemia, neonatal abstinence syndrome, and cleft lip.

Early Symptoms and Neurologic Diagnoses

Most children (64%) initially received alternative diagnoses such as autism spectrum disorder (43%) or developmental delay (29%). Other reported initial diagnoses included cerebral palsy, sensory processing disorder, ADHD, dyspraxia, and epilepsy. Genetic testing was most often prompted by developmental regression, particularly loss of speech and motor skills (reported by 91% of participants). The median age at which caregivers first perceived symptoms, regardless of symptom type, was 2.3 years (range birth-4 years).

Developmental Trajectory

Most subjects achieved early motor milestones on time or with a mild delay, followed by regression in gross and fine motor abilities during early childhood (45%) (Table 2). Regression was defined as a caregiver-reported loss of previously acquired developmental skills. Standing and walking were lost around a median age of 9.4 years (range 3.5–13). Fine motor and language regression occurred later but were universal.

Table 2:

Developmental Milestone Acquisition and Regression

Developmental Milestone	Number Acquired n (%)	Number Regressed n (%)	Acquisition Age (yr) Median (Range)	Regression Age (yr) Median (Range)
Reach for Object	11 (100)	5 (45)	0.7 (0.3-12)	12 (9.9-14.5)
Transfer hand-to-hand	9 (82)	3 (33)	2.8 (0.5-12)	14 (14)
Stand Independently	11 (100)	6 (55)	1.1 (1-2)	9.4 (3.5-13)
Walk without Support	11 (100)	6 (55)	1.3 (1-12)	9.4 (4-13)
Climb Stairs	9 (82)	4 (44)	1.9 (0.7-14)	10.7 (6-13)
Write	5 (45)	7 (100)1	5.5 (2-6)	6.8 (6-8.5)
First Words	10 (91)		1.3 (0.5-2)
Understands most of what
parent says	9 (82)	3 (33)	1.5 (0.7-9)	NR
Laugh when something is funny	10 (91)	3 (30)	1 (0.5-9)	13 (12-18)
Turns to Name	11 (100)	2 (18)	0.8 (0-6)	14 (14-14)
Follows Parent's Voice	11 (100)	2 (18)	0.8 (0.5-14)	15 (15-15)
Listens when Spoken to	11 (100)	2 (18)	0.6 (0.4-4)	NR
Enjoys Stories	9 (82)	0 (0)	0.8 (0.7-14)	NA
Enjoys Music	11 (100)	1 (9)	1 (0.5-14)	NR
Sings Along to a Song	7 (64)	7 (100)	4 (3-6)	6 (4-6)
Waves Bye-Bye	8 (73)	3 (38)	0.8 (0.7-9)	10 (5-10)
Communicates with
Sounds Parent	9 (82)	3 (33)	2.3 (1-9)	6.5 (4-9)
Understands
Social Smile	11 (100)	2 (18)	0.5 (0.3-9)	14.3 (14-14.6)
Sustained Eye Contact	11 (100)	3 (27)	1.3 (0.5-4)	16 (16-16)
5-10 Words	8 (73)	5 (63)	3 (2-6)	7 (4-9)
2-3 Word Phrases	9 (82)	7 (78)	2.5 (2-6)	6 (3.3-10)
Sentences	4 (36)	2 (50)	4 (2-6)	3.5 (2-5)
Tells a story	2 (18)	2 (100)	2 (2-2)	4.5 (4-5)

¹Two subjects were reported to undergo regression in handwriting but were reported to never have acquired the ability to write. NR = not reported; NA = not applicable

Language skills peaked at a median of 3.3 years (range 2–6), with regression beginning at 5 years. Simple social skills (enjoying stories, enjoying music, turning to name, following parent’s voice, listening when spoken to, social smile) were typically preserved, whereas complex skills such as writing, sentence formation, and storytelling were lost.

Participants used multiple communication methods: facial expression (n = 9, 82%), sign language or gestures (n = 6, 55%), vocalizations (n = 6, 55%), augmentative and alternative communication device (n = 4, 36%), and single words (n = 4, 36%). Only one individual used 2-3-word phrases and full sentences. Speech abnormalities were frequent, including dysarthria and high-pitched voice.

Neurologic Findings

The most common neurological features were ataxia (82%), global developmental delay (82%), and intellectual disability (73%) (Table 3). Tone abnormalities, including spasticity, hypotonia, and hypertonia, were common, although the topographic distribution of these abnormalities was not noted. Movement disorders beyond ataxia, including dyskinesias (defined in the survey as “impairment of voluntary movements resulting in jerky motions”) and extrapyramidal movement disorder (described in the survey as “i.e. dystonia, chorea, parkinsonism, or rigidity”), were uncommon (n = 2, 18% and n = 3, 27%, respectively).

Table 3:

Prevalence and Timing of Neurologic Diagnoses

Symptoms	Patients n (%)	Age of Onset (yr) Median (Range)
Ataxia	9 (82)	2 (1-6.5)
Global Developmental Delay	9 (82)	2.5 (03-5.3)
ID	8 (73)	5.2 (2.5-6)4
Apraxia5	7 (64)	4 (1-6)
Spasticity	7 (64)1	11 (2-17.5)
Hypotonia	7 (64)	2 (1-13)
Hypertonia	6 (55)	6.8 (2-11)
Postural Instability	6 (55)	2.5 (0.7-11)
ASD	4 (36)	5 (2-6)
Seizures	4 (36)2	5.8 (4.3-7.3)
Spine-Related Issues6	4 (36)	14 (12-16)
ADHD	3 (27)	4.5 (4-5)
Extrapyramidal Movement Disorder7	3 (27)	11.5 (4-12)
Microcephaly8	3 (27)	2.5 (2.5-2.5)
Dyskinesia9	2 (18)	6.1 (6-6.25)
Obsessive Compulsive Disorder	0 (0)	NA

¹Six diagnosed, one no formal diagnosis but 'obvious in his movements'

²Three definitive diagnoses, one suspicious clinical event but no definitive diagnosis

³Participant reported a diagnosis of global developmental delay ‘since birth,’ interpreted as age 0 for analysis

⁴Note that ID is not typically diagnosed before 5-6 years of age. It is not known whether individuals received an ID diagnosis vs. cognitive delay at younger ages.

⁵Term defined in the survey text as “difficulty with skilled movement”

⁶Term defined in the survey text as: “eg. Scoliosis, tethering of the spine”

⁷Term defined in the survey text as “impairment of voluntary movements resulting in jerky motions”

⁸Term defined in the survey text as “small head size”

⁹Term described in the survey text as “i.e. dystonia, chorea, parkinsonism, or rigidity”

Three individuals (27%) had microcephaly (defined as caregiver-reported “small head size” without reported OFC Z-scores or centiles), and four (36%) had spinal or postural abnormalities. Cortical visual impairment occurred in two cases; hearing loss was reported in one individual with the novel c.495C>A (p.Phe165Leu) variant.

Nearly half (n = 5, 45%) reported intolerance to heat or cold. Fainting episodes (27%) were occasionally triggered by overheating.

Seizures

Three individuals (27%) had a confirmed diagnosis of epilepsy, while one additional participant experienced paroxysmal events suspicious for seizures but did not meet diagnostic criteria for epilepsy. Although seizure-like events were reported in several participants, only two demonstrated both clinical episodes consistent with epileptic seizure semiology and corresponding epileptiform abnormalities on electroencephalography (EEG). Seizure types included generalized tonic-clonic (endorsed by 3), absence (2-3), and focal (2). Two participants experienced atypical episodes (one with episodic gastric discomfort associated with pallor, sweating, and loss of tone; another with two prolonged episodes of pallor and altered responsiveness in the setting of overheating). At their most frequent, seizures occurred multiple times daily for 2 participants and on most days for the third. Lights, poor sleep, illness, and dehydration were reported triggers. One individual was started on 3 antiseizure medications (valproic acid, topiramate, and clobazam) despite recurrent negative EEGs but experienced worsening seizures, so antiseizure medications were discontinued. He has been off antiseizure medications for more than 7 years at the time of survey completion and continues to experience 3-4 events yearly suspected to be autonomic episodes vs. seizures. Another was treated with lamotrigine for 15 months but did not experience recurrent episodes and was weaned off without recurrence. The third experienced ongoing EEG-confirmed seizures despite concurrent treatment with three antiseizure medications (gabapentin, cannabidiol, and brivaracetam), consistent with medically refractory epilepsy. The last individual was on 3 total lifetime antiseizure medications (lamotrigine, levetiracetam, and valproic acid), with seizure resolution on lamotrigine monotherapy at the time of survey completion. Response to antiseizure medications varied, with 1 reporting complete control, 1 with partial control, and 1 with exacerbation of seizures after starting medication.

The median age at first seizure was 5.3 years (range 3.5-7.3 years), with the most recent seizure occurring at a median of 7.8 years (range 7.5-8 years). Two individuals (18%) had ongoing seizures at the time of survey completion. Notably, one participant who reported generalized tonic-clonic seizures had numerous EEGs performed without electrographic correlate, raising questions as to whether these were truly epileptic seizures. Only two individuals in the cohort experienced clear epileptic seizures.

Associated Features

Nine individuals (82%) reported sleep difficulties, often starting in the first years of life, including difficulty falling asleep (n = 7, 64%), staying asleep (n = 5, 45%) or waking in the morning (n = 2, 18%), excessive nighttime movements (n = 1, 9%), and obstructive sleep apnea (n = 1, 9%). Two (18%) required a sleep aid (auto pap/CPAP). Chronic pain was common (n = 5, 45%) and affected the knees, head, bladder, and full body. Pain management techniques included medication, rest, stretching, therapy, comfort, and massage.

Most subjects (n = 8, 73%) had undergone a procedure requiring anesthesia (median age 5.1 years at first anesthesia exposure, range 1.3-12.1 years), and 3 experienced multiple episodes of anesthesia. Of those who had undergone such a procedure, 3 (38%) experienced subsequent cognitive decline although the exact timing of the decline in relation to the procedure was not reported. Three subjects (27%) experienced regression after an otherwise benign fever or infectious illness. While not explicitly asked in the survey, it was assumed that all survey subjects had experienced such a fever or infectious illness at some point.

Significant injuries, including broken bones and injuries requiring hospitalization, occurred in 4 individuals (36%). One was hospitalized with pneumonia, and 4 (36%) were hospitalized for other indications (including a planned procedure/study, epilepsy/EEG, vomiting/poor intake, urinary retention, ear bleeding/infection, regression, postoperative complications, NICU stay, obstipation, and brain bleed). Of note, the brain bleed was a small, bilateral, frontal subdural hemorrhage without mass effect of unclear etiology, which was felt to be asymptomatic. This was discovered during the evaluation of macrocephaly.

Feeding Skills

Difficulty with feeding and growth was common. In the first year of life, 6 individuals (55%) had trouble gaining weight, and 4 (36%) had difficulty learning to eat solid food. Most subjects had ongoing feeding difficulties at the time of survey completion. While 9 individuals (82%) were reported to “always” or “sometimes” like the sensation of eating, 10 (91%) had trouble knowing if they were full, 9 (82%) experienced oral motor difficulties, 7 (64%) ate or drank too fast, and 6 (55%) overstuffed their mouths. Of the 10 individuals who reported some amount of oral intake, all experienced choking either “sometimes” or “always.” Two individuals (18%) had a feeding tube placed.

Free-Living Activity Measurements

Five individuals (median age 11.8 years, range 8.1-12.5 years) were enrolled in the wearable device portion of the study, 4 of whom were included in the caregiver surveys. One individual participated in the wearable device study but was not represented in the caregiver surveys. Four participants were male (80%). Participants completed (5, 4, 4, 4, and 5) ankle and (7, 4, 5, 4, and 5) wrist sessions between December 2021 and December 2023. The sensors were worn on the dominant side, which was right for 4 individuals and left for 1.

Participants with UBTF-related disorder showed a clear difference from controls on the AI Entropy (p < 0.001, Cohen’s d = 2.7) and ALS composite (p < 0.001, Cohen’s d = 2.3) motor metrics in both ankle and wrist accelerometer data. Standardized activity measures were consistently lower in the UBTF cohort and showed a trend to decline over time (Figure 1A-B). High intra-session reliability was seen over 3-day periods in UBTF and control subjects (Figure 1C-D). Notably, this motor activity showed good correlation with ataxia severity (measured by the BARS) (Figure 1E-F). Furthermore, while controls showed stability in motor activity over time, the UBTF cohort had lower scores at baseline and declined over the study period (Figure 1A-B). MSDR for slopes was −1.9 and −4.2 for AI Entropy and ALS composite in the UBTF group and −0.16 and 0.14 in the control group.

Fig. 1.

Selected motor metrics from wearable ankle sensor data (N = 5). Selected activity measures are shown for both UBTF (red) and controls (gray). A-B Standardized activity measures plotted against time from baseline measures showing trajectory over the study period. C-D Reliability of sensory metrics plotting the measures from the first half (days 1-3) and second half (days 4-6) of each data collection period plotted against each other for each participant. The black dashed line represents y = x. Intraclass correlation coefficients (ICC) are shown. E-F Relationship between activity measures and clinical severity as measured by the total BARS score. The black dashed line shows the line of best fit. Pearson correlation coefficients (r) and p-values are shown.

Biomarker Investigation using serum NFL measurements

To better understand the effect of the c.628G>A (p.Glu210Lys) variant on a model organism, the B6J.Ubtf ^E210K mouse was characterized. Interestingly, both heterozygous B6J.Ubtf ^E210K/+ and homozygous B6J.Ubtf ^E210K/E210K mice exhibited elevations in serum NFL compared to littermate controls (Figure 2A-B). Similarly, in humans, elevations in serum (Figure 2C) and plasma (Figure 2D) NFL were seen in all but the oldest individual.

Fig. 2.

Neurofilament Light Chain elevations are seen in a mouse model and in humans. (A-B) Longitudinal serum analysis of neurofilament light chain (NFL) of B6J. Ubtf ^E210K mice. The horizontal line for each group represents the mean. Wild type littermate controls are shown in black. Two-way ANOVA was corrected for multiple comparisons using Tukey method. ns, not significant; **** p ≤ 0.0001. (C) Serum and (D) Plasma NFL levels obtained from individuals with a clinical and genetic diagnosis of UBTF-related disorder (blue) vs. controls (orange). Values below the lower limit of detection of the assay are represented by unfilled circles. A change in clinical laboratory and specimen type occurred during the study period, resulting in both plasma and serum samples being shown.

DISCUSSION

This study delineates the natural history of UBTF-related Childhood-Onset Neurodegeneration with Brain Atrophy (CONDBA), characterized by initially normal or mildly delayed development, followed by regression and progressive ataxia. Clinical severity, captured by the Brief Ataxia Rating Scale (BARS), correlated with motor activity measured by wearable devices and elevations in neurofilament light chain (NFL) in both humans and mouse models. Together, these findings clarify disease trajectory and introduce quantifiable biomarkers for future trials.

Most participants carried the recurrent c.628G>A (p.Glu210Lys) variant, confirming its strong association with the CONDBA phenotype. One individual harbored a novel c.495C>A (p.Phe165Leu) variant, presenting with earlier regression (6 months) and absence of ataxia, supporting earlier observations that non- c.628G>A (p.Glu210Lys) variants may produce distinct or more severe phenotypes (Table 4)^15,18. This aligns with gnomAD data indicating UBTF’s intolerance to both missense (o/e 0.46, Z = 6.27) and predicted loss-of-function (o/e 0.07, pLI = 1) variants²⁷, underscoring its functional constraint.

Table 4:

Review of UBTF variants reported in the literature¹

Coding	Variant	Zygosity	Phenotype	Source
c.628G>A	p.Glu210Lys2	Heterozygous	CONDBA	Multiple3-11
c.495C>A	p.Phe165Leu	Heterozygous	Developmental delay and early regression	Current manuscript
c.608A>G	p.Gln203Arg	Heterozygous	Developmental delay, hypomyelination	Tinker et al. 202215
c.1327C>T	p.Arg443Ter	Heterozygous	Developmental delay, facial dysmorphism	Wang et al. 202418
17q21.31 (42248053_42294092)×1		Heterozygous	Developmental delay
17q21.31(42239218_42345522)×1		Heterozygous	Developmental delay, cardiac anomalies
c.905 + 1G>A (splice-site variant)3		Heterozygous	Developmental delay/Intellectual disability	Chiriatti et al. 202628
c.2104del (p.Ser702Profs*83)		Heterozygous	Developmental delay/Intellectual disability	Yi et al. 202617
c.1199del (p.Gly400Alafs*38)		Heterozygous	Developmental delay/Intellectual disability

¹Unless otherwise noted, each variant was described in a single individual

²Reported in 19 individuals prior to this current manuscript

³Reported in 4 individuals

Several participants reported additional genetic findings, but these were generally considered noncontributory to the neurodevelopmental phenotype, except for one pathogenic SDHA variant causing hereditary paraganglioma-pheochromocytoma. Importantly, all subjects with additional variants also carried UBTF mutations, emphasizing UBTF’s central pathogenic role.

Regression was frequently reported following anesthesia or illness, consistent with previous case reports suggesting that physiological stressors may precipitate neurologic decline. Nearly half reported adverse responses to extreme temperatures, and several experienced fainting or autonomic-type episodes, suggesting autonomic instability as a potential feature. Notably, only two individuals exhibited clear epileptic seizures, while others had events possibly representing dysautonomic paroxysms rather than epilepsy. The similar frequency of “seizures” between this and earlier reports (35%)¹² supports this interpretation. Future autonomic testing may help distinguish epileptic from nonepileptic paroxysms, as autonomic dysfunction is common in both genetic ataxias and disorders such as Rett syndrome^29-32.

Wearable devices provided a sensitive and non-invasive measure of motor decline. Accelerometry-derived metrics (AI entropy and ALS composite) not only differentiated affected individuals from controls but also correlated with clinical ataxia severity and demonstrated progressive deterioration over one year. While not tailored specifically to the UBTF population, the ALS composite score was chosen given its ability to detect within-subject change over time in a neurodegenerative disease and to generalize to other populations, including ataxias²⁵. In the ALS study²⁴, it was shown that rate of change of the ALS composite score correlated well with the rate of change of the ALS functional rating scale (ALSFRS-R). This supported that the score was meaningful in ALS and enabled assignment of clinical value to changes in the sensor-based score. In this study, we showed that the ALS composite score and AI entropy correlate with clinical ratings (BARS) in UBTF and have high intraclass correlation coefficients, supporting meaningfulness and reliability in this disorder. However, the small sample size and limited longitudinal data do not allow for precise estimates of minimum detectable change and minimal clinically important difference, which is a limitation of this study. In the interventional trial context, we envision using these metrics to establish the individual’s pre-treatment trajectory and subsequently evaluate whether treatment alters the trajectory toward that of the control population.

Similar to findings in Ataxia-Telangiectasia (A-T), reduced motor activity in UBTF-related disease likely reflects a combination of ataxia, hypotonia, and weakness³³. Cerebellar atrophy observed in advanced disease stages supports the cerebellar contribution to these deficits^11,12. The consistency and sensitivity of wearable data suggest that wrist and ankle accelerometers could serve as an objective motor biomarker, suitable for remote monitoring in interventional trials.

NFL has emerged as a promising biomarker of neuroaxonal health in numerous neurological conditions, including genetic ataxias^26,34-39. In both humans and the Ubtf E210K mouse model, NFL was markedly elevated despite the absence of overt neuropathology in mice. This suggests that the c.628G>A (p.Glu210Lys) variant may induce subclinical neuronal stress or dysfunction before symptomatic onset. Such findings parallel other neurodegenerative conditions where NFL elevations precede measurable clinical decline, including spinocerebellar ataxia type 3⁴⁰. It is interesting that NFL is inversely correlated with age in this cohort, which may indicate that these levels decrease as the disease advances. However, further work will need to be done to examine this potential correlation, given the limited sample size. These results support the potential of NFL as an early and cross-species indicator of neuronal injury in UBTF-related disease.

The concurrent use of wearable accelerometry and NFL quantification represents a significant methodological advance. Although the cohort size in this current study was too small to establish a relationship between these measures, studies in A-T demonstrate a positive correlation between NFL levels and ataxia severity⁴¹. Future longitudinal studies in larger cohorts may clarify whether combined biomarker analysis can provide a multidimensional view of disease progression, integrating motor function and molecular pathology.

A key limitation of this study is the small sample size that was chosen as a convenience sample, which limits the generalizability of the results, and reliance on caregiver-reported data, which may introduce recall bias and variability in reporting (e.g. “microcephaly” was defined as “small head size” but not quantified so the degree of microcephaly in the cohort is not known). Parental interpretation of survey questions and medical terms (e.g., “dystonia” and “parkinsonism”) may differ from standard clinical interpretations, representing an additional limitation of the study. Two individuals in the small cohort were adopted, which may affect the reliability of the early developmental history. Lack of longitudinal NFL sampling precludes assessment of temporal dynamics. Additionally, while most participants had confirmed c.628G>A (p.Glu210Lys) variants and one reported the novel variant c.495C>A (p.Phe165Leu), a minority (n = 3) did not report their genetic variants, limiting our ability to confirm the diagnosis of UBTF-related disorder. While neuroimaging features of this disorder are clinically important and relevant to biomarker interpretation, the survey-based nature of the study limited the ability to collect this data and thus was not included in this current study. Despite these limitations, the data remain internally consistent and reinforce existing knowledge of the UBTF disease spectrum.

In conclusion, UBTF-related disorder manifests as mild developmental delay followed by global regression and progressive ataxia. Wearable accelerometers offer an accessible, quantitative measure of disease severity that correlates with clinical function, while elevated NFL levels serve as a biochemical marker of neuronal injury across humans and mice. Together, these biomarkers hold promise for future therapeutic trials and early detection of disease activity.

What is already known on this topic –

The neuroregressive phenotype of CONDBA has been described in several case studies and small case series totaling 19 individuals (along with 9 individuals with haploinsufficiency leading to developmental delay and 1 with more severe neurodevelopmental impairment); however, a systematic natural history has not been performed. Biomarkers for use in potential future clinical trials have not yet been identified.

What this study adds –

This study adds a broader description of the natural history of UBTF-related disease and introduces both motor activity obtained from wearable devices and blood NFL levels as potential biomarkers.

How this study might affect research, practice or policy –

This study raises awareness of UBTF-related disorder as an “ataxia plus” phenotype and opens up the opportunity for further validation of these potential biomarkers.

List of abbreviations

ADHD

Attention Deficit Hyperactivity Disorder

ASD

Autism Spectrum Disorder

A-T

Ataxia-Telangiectasia

BARS

Brief Ataxia Rating Scale

bMRI

Brain Magnetic Resonance Imaging

CONDBA

Childhood-Onset Neurodegeneration with Brain Atrophy

EEG

Electroencephalogram

HMG

High Mobility Group

ID

Intellectual Disability

MGH

Massachusetts General Hospital

NFL

Neurofilament Light Chain

RPI

RNA Polymerase I

SL1

Selectivity Factor 1

UBF

Upstream Binding Factor

UBTF

Upstream Binding Transcription Factor