Developmental trajectory of individuals with Pelizaeus-Merzbacher Disease (PMD)

Abstract

Introduction:

Pelizaeus-Merzbacher Disease (PMD) is a PLP1-related disorder for which genotype alone is insufficient to reliably determine PMD disease progression. We aim to characterize the developmental outcomes of PMD-affected individuals and evaluate their association with early presentation features.

Methods:

Molecularly confirmed PMD subjects were included. Data on baseline characteristics, developmental, and disease outcomes were abstracted from medical records and, when feasible, supplemented by caregiver interviews. Data were summarized using descriptive statistics.

Results:

The study included 111 subjects with milestone acquisition ranging from 9.9% to 68.5% across gross motor, 41.1% to 85.6% across fine motor, and 30.0% to 66.7% across expressive communication skills. Genotype and milestone were not significantly associated. Presentation with stridor (N = 20) was associated with profound impairments (overall non-attainment ranging from 10.0% to 100%). In the absence of stridor, head control by 12 months was associated with developmental outcomes: severe outcomes (overall non-attainment ranging from 0% to 96.0%) when head control was not acquired by 12 months (N = 50), and moderate outcomes (overall non-attainment ranging from 0% to 78.0%) when head control was attained (N = 41). Median time to PMD-related complications was noted to be earlier in the profoundly impaired subtype, compared to the severe and moderate subtypes.

Conclusions:

PMD impairs skill acquisition, with gross motor milestones impacted more than expressive communication. Stridor and lack of head control by 12 months were associated with poor developmental outcomes and disease progression. These insights will support better characterization of disease course for newly diagnosed families, as well as future trial development.

1. Introduction

Pelizaeus-Merzbacher Disease (PMD) is a rare hypomyelinating leukodystrophy associated with pathogenic variants in the X-linked PLP1 gene [1–4], encoding PLP1 protein and its isoform DM20 [3,5]. PMD has a male predilection [2,4,6,7] with an estimated prevalence of 1.9 per 100,000 male live births in the United States [8,9] and a global incidence of 1 per 90,000 to 750,000 live births [10]. Affected individuals present across a severity continuum with variable neurologic impairments. Within this continuum, distinct subtypes of PMD have been described based on the age at disease onset, and early presenting features. This includes the neonatal-onset connatal form, infantile-onset classic subtype, PLP1 null syndrome and Spastic Paraplegia type 2 (SPG2), including Hypomyelination of Early Myelinating Structures (HEMS) [1,4,7,9,11–13]. Of these subtypes, the connatal PMD is the most severe form and is characterized by neonatal stridor, feeding difficulties, failure to thrive, generalized hypotonia, and profound neuro-developmental delay [1,4,9]. This is followed by the classical PMD subtype with clinical severity described across a broad spectrum and a relatively moderate course compared to the connatal form [9,13–16]. Clinically, most individuals with classical PMD present with nystagmus as the initial symptom [9,13,16,17]. Neurodevelopmentally, infants with classical PMD continue to gain milestones, achieving more than those with connatal PMD, with assisted ambulation reported in a size-able cohort [9,17]. Milder disease phenotypes with onset later in life include the PLP1-null syndrome and SPG2, including HEMS [1,4,7,9,11]. Both these milder phenotypes are characterized by a normal neurodevelopmental course, followed by gross motor difficulties such as gait abnormalities and loss of independent ambulation over time [9]. In addition, nystagmus and autonomic dysfunction have been reported in individuals with the SPG2 phenotype [9].

Most studies on genotype-phenotype correlations in PMD have focused on PLP1 overexpression, which also constitutes approximately 60–70% of the PLP1-related variants [5,6,9,18–20]. Overwhelming evidence suggests that PLP1 overexpression results in classical PMD [9,11,21]. Beyond PLP1 overexpression, such correlations have been challenging to delineate with the other variant groups: Gain-of-Function (GoF) missense variants, Loss-of-Function (LoF) variants, and other splicing variants. These variant groups present across the PMD phenotypic spectrum but often show a tendency to skew toward either one end of the spectrum [1,2,9,11–13,19–24]. For example, most GoF variants have been associated with the severe connatal form, although manifestation of classical PMD and HEMS/SPG2 have also been reported in a small subset [9,11,21,24]. These support the observation that prognostication of PMD subtype or disease progression is difficult based on genotype alone [9,12,13,19,20].

Gene and cell-based therapeutic advancements pave the way for therapeutic disease modification in PMD. Increasing evidence suggests better outcomes from these types of interventions when treatment is initiated early in life [25–29]. As such, timely diagnosis and the ability to determine disease outcomes early in life to identify treatment-eligible subjects are critical. Understanding disease outcomes enables cohort stratification and determines appropriate selection of clinical trial endpoints and outcome measures [9]. Thus, grounding our understanding of PMD prognosis on baseline functions in affected individuals is an urgent need to achieve clinical trial readiness. As PMD is a neurodevelopmental disorder, we hypothesize that exploring early disease features, including early developmental milestones, may enable us to understand disease progression. In this study, we characterize baseline disease presentations, including developmental trajectories of PMD-affected individuals. We evaluate how genotype, age at disease onset and early disease presentations may be associated with developmental outcomes and disease progression in PMD.

2. Methods

2.1. Study design

The Global Leukodystrophy Initiative Clinical Trial Network (GLIA-CTN) PMD natural history study (GLIA-PMD) was performed as part of the Children’s Hospital of Philadelphia (CHOP) IRB-approved protocol—Myelin Disease Biorepository Project (MDBP) protocol (IRB #14–011236). The GLIA-PMD natural history study included subjects directly consented into the MDBP protocol and, in addition, de-identified subject-level data collected and contributed by six (6) external international collaborators under local IRB-approved protocols. The international contributors were: the research institute of the McGill University Health Centre, Canada (IRB #11-105-PED, 2019-4972); Ospedale Pediatrico Bambino Gesu, Rome, Italy (IRB # 3002_OPBG_2022); Ospedale Dei Bambini Vittore Buzzi, Milano, Italy (IRB #43792/2018 and 0051756/2018); Vrije Universiteit Medical Center, Amsterdam (IRB #2018.300); National Center of Neurology and Psychiatry, Japan (IRB #A2016–078); Murdoch Children’s Research Institute, Australia (IRB #HREC/61136/RCHM-2019). Certified genetic counsellors determined subject inclusion in the study [30] upon review and confirmation of molecular diagnosis of PMD [31]. Subjects not meeting criteria for diagnostic confirmation were excluded from the study (Supplemental Fig. 1).

2.2. Data collection

Data were collected from medical records, which served as the primary source of data for the study. Preexisting Standardized Operating Procedures (SOPs) on medical record-based natural history studies [30] were used to inform the study’s data collection and verification activities. Trained study personnel reviewed all available medical records to extract data on gender, race/ethnicity, genotype, age at disease onset along with the symptoms, age at diagnosis, age at last evaluation/last available medical record, developmental trajectories, and time to first occurrence of key medical events. Age at disease onset was defined as the age at which the family or the provider noted the first PMD-related concern(s). Data were collected directly into the GLIA-PMD natural history REDCap database hosted at CHOP. REDCap (Research Electronic Data Capture) is a secure, web-based software platform designed to support data capture for research studies, providing 1) an intuitive interface for validated data capture; 2) audit trails for tracking data manipulation and export procedures; 3) automated export procedures for seamless data downloads to common statistical packages; and 4) procedures for data integration and interoperability with external sources [32,33].

2.3. Genotype characterization

At the subject level, the genotype was categorized as “Overexpression” when PLP1 duplication or triplication were noted, and “Others” to encompass GOF and LOF variants as well as splicing, including HEMS-causing variants. The GOF variants included missense and in-frame deletion or duplication variants, resulting in protein mis-folding and toxic gain of function. The LOF included deletions as well as nonsense or frameshift variants.

2.4. Developmental trajectories

Assessment of developmental trajectories included data extraction on 17 developmental milestones, across motor and expressive communication domains, adopted from the Denver Developmental Screening Assessment II [34], previously evaluated in the leukodystrophy population [35–37]. The Denver Developmental Screening Assessment II was used as it captures a comprehensive range of milestones, with availability of normative data for comparison. In addition, information on “assisted walking” was collected in the study and was defined as the ability to walk with walker or assistance (i.e., includes one-hand and two-hand assist) as well as cruising. Milestone acquisition status was captured in the study database as “attained” (age known), “attained (age unknown but appropriate)”, “attained (age unknown but delayed)”, or “never attained”. When milestones were attained, age at acquisition was captured if available from the medical record. If age at acquisition was not specified, but more generally inferable as appropriate or delayed from the text (e.g., “parent reports sitting was delayed”), data were captured accordingly as “attained (age unknown but appropriate)” or “attained (age unknown but delayed)”. When milestone attainment status was not reported in the medical note, the milestone was marked as “unknown/not reported”. Loss of milestones was captured as applicable and when known.

Medical record extraction was supplemented by virtual interviews with caregivers of PMD subjects enrolled in MDBP. Caregivers were invited to participate via email, detailing the purpose, modality, and duration of the interview. Interested caregivers were scheduled for a 30-min virtual interview with a trained member of the study team. The interviews were performed using a structured questionnaire, and the interview responses were annotated in a paper Case Report Form (CRF). Subjects were considered lost to follow-up in these interviews in the absence of a response following three attempts to contact the family by email, each 2 weeks apart, and a final attempt by phone. Upon completion of the caregiver interviews, a clinically trained member of the study team reviewed the data obtained through medical record extraction against that of the caregiver interview CRFs to fill gaps in the developmental data. In the event of discrepancies, data from concurrent medical records were given precedence to minimize recall bias, inherent to interviews in general. The maximum age at collection of developmental data were determined using the highest of either the patient’s age at the last available clinic note or the patient’s age at the time of interview with the caregiver.

2.5. Time to onset of key medical events

Data on the onset of key medical events such as nystagmus, stridor, failure to thrive, gastrostomy tube placement, complete reliance on gastric feeds, aspiration pneumonia, artificial ventilation, loss of urinary function, hip dislocation, scoliosis, and seizures were captured in the study [30]. Events were captured as “noted” when explicitly stated as present in the medical records, “not noted” if an event was explicitly ruled out, and as “unknown/not reported” when there was no mention of the event across the available notes. When an event was noted, the age at the first occurrence was collected.

2.5.1. Statistical approach

Subjects without any information on developmental milestone acquisition/loss from the available records, individuals less than 18 months at last follow-up (as most skills may still be emerging before 18 months), and female carriers (as PMD is an X-linked disorder) were excluded from analysis in the study.

Age at milestone acquisition/loss was not always available from medical records. To maximize the available data and maintain the analytic sample, milestone acquisition was categorized as “Attained appropriate”, “Attained delayed”, and “Never attained”, applying established imputation approaches using normative data [34,36]. Normative data were adopted from the Denver Developmental Screening Assessment II [34]. Delay in milestone acquisition was defined as milestone acquisition attainment by P90 + 2SD on the normative data, where P90 indicated the age by which 90% of typically developing children attained the milestone of interest, and + 2SD established a more stringent criterion for identifying those deviating from a normal developmental course [34,36]. This approach ensured a conservative definition when classifying delayed acquisition of a milestone and has been applied in prior leukodystrophy research [36]. Inversely, appropriate attainment was considered when the age at acquisition was below the set threshold for the delay. Milestone loss was categorized as “Lost” and “Never lost”.

Baseline data were summarized using counts and proportions for categorical variables and mean, standard deviation (SD), median and range (total and inter-quartile range, IQR) for continuous variables. Chisquare tests and Mann-Whitney tests were used to determine the association of categorical and continuous variables, respectively. Overall milestone acquisition during follow-up was summarized. Differences in overall milestone acquisition were evaluated across genotype categories as well as the presence/absence of presenting symptoms (i.e., clinical features noted at the time of disease onset in the GLIA-PMD natural history cohort), using the Mann-Whitney test, to identify the early presenting features related to milestone acquisition in PMD. The milestones were then grouped by domain and ordered based on the timeline of acquisition in typically developing children. Percent milestone attainment was calculated for each milestone. When a marked decrease in percent attainment was observed across two consequent milestones within a domain, the earlier of the two milestones was selected as a key milestone for initial evaluation during cohort stratification. To identify strata, we summarized the percent attainment at the well-visit timepoints (1, 2, 3, 4, 6, 9, 12, 15, 18, and 24 months), based on the related early presenting features, as well as head control (the earliest milestone expected to be achieved). The timepoint by which head control, if attained, would be clinically relevant to developmental outcomes in PMD was selected based on the well-visit timepoint with the greatest decline in the key milestones (sit without support, thumb-finger grasp, and speak 1 word) from the previous timepoint (Supplemental Fig. 2). Following this, we expanded the calculation of percent milestone attainment by the strata to the remaining 13 milestones (Supplemental Table 5).

A horizontal stacked bar chart depicted the overall acquisition and loss of milestones in PMD. A box-and-whiskers plot depicted the median age at milestone acquisition and was overlaid with normative data adopted from the Denver developmental scale. Kaplan-Meier curves were used to depict the time from birth to the occurrence of key medical events, with individuals censored at age of last follow-up, and log-rank test was used for comparisons across the overall cohort, genotype, and identified subtypes.

3. Results

3.1. Study cohort

The GLIA-PMD study included a total of 144 subjects with a molecularly confirmed diagnosis of PMD: 81 MDBP-enrolled and 63 contributed by international collaborators. Of these, 33 subjects were excluded from the study, 30 for lack of information on developmental milestones (including 10 individuals less than 18 months at last follow-up and deemed too young to assess milestones) and 3 female carriers (Supplemental Fig. 1). Baseline characteristics was compared between the included cohort and the cohort excluded due to lack of information on developmental milestones (Supplemental Table 1). The median age at last follow-up was significantly different between the included and the excluded cohort: 5.8 (1.30, 12.0) years in the excluded cohort and 7.60 (IQR: 4.54, 16.40) years in the included cohort; p = 0.009). Also, residence was significantly different between the included and excluded cohorts; p < 0.001.

Baseline characteristics of the included cohort are summarized in Table 1. A total of 111 (78.7%) subjects were included for further analysis. The most common genetic change was overexpression of the PLP1 gene (N = 51/111; 45.9%). The median age at disease onset was 0.10 (Inter-quartile range, IQR: 0.00, 0.33) years, and the median age at diagnosis was 1.54 (IQR: 0.90, 2.93) years.

Table 1.

Baseline characteristics of included subjects.

Baseline Characteristics	Cohort (N = 111)
Race, n (%)
Asian	10 (9.0%)
Bi–/Multiracial	2 (1.8%)
Black/African-Amcrican	8 (7.2%)
Caucasian/White	73 (65.8%)
Hispanic/Latino	4 (3.6%)
Other	3 (2.7%)
Unknown/Not Reported	11 (9.9%)
Residence, n (%)
International	67 (60.4%)
United States of America	42 (37.8%)
Unknown/Not Reported	2 (1.8%)
Variant Category, n (%)
Overexpression	51 (45.9%)
Other	60 (54.1%)
Gain of Function- Missense	36 (60.0%)
HEMS/Splicing	17 (28.3%)
Loss of Function	7 (11.7%)
Age at Onset (years)
Mean (SD)	0.27 (0.70)
Median [Q1.Q3]	0.10 [0.00, 0.33]
[Min, Max]	[0, 7.00]
Age at Diagnosis (years)
Mean (SD)	3.71 (5.83)
Median [Q1, Q3]	1.54 [0.90, 2.93]
[Min, Max]	[0, 32.0]
Missing	7 (6.3%)
Age at Last Evaluation (years)
Mean (SD)	10.9 (8.47)
Median [Q1, Q3]	7.80 [4.54, 16.4]
[Min, Max]	[1.58, 38.6]

Of the eligible MDBP-enrolled subjects contacted for the interview, 26 participated. There were no significant differences in the baseline demographics between the interviewed and non-interviewed cohorts (Supplemental Table 2). None of the international subjects contributed by our outside sites underwent an interview.

3.2. Milestone acquisition and loss in PMD

Milestone acquisition and/or loss status and the corresponding age are summarized in Supplemental Table 3 & Fig. 1. Among individuals with known milestone attainment status, percent acquisition was overall low, ranging between 9.9% and 68.5% of individuals in the gross motor domain, 41.1% to 85.6% in the fine motor domain, and 30.0% to 66.7% in the expressive communication domain. Overall, a gradual decrease in percent attainment with increasing complexity of developmental skills was noted (Fig. 1). When grouped by functional domains (Supplemental Table 3), this was particularly evident within the gross motor domain (percent attainment: 10.0% with independent ambulation, 41.1% with thumb-finger grasp, and 52.3% with speak 1-word). By order of acquisition, the first milestone, by domain, with notable decrease in percent acquisition were independent sitting (attainment of ‘roll over’ in 68.5% to ‘sitting’ in 34.9%), thumb-finger grasp (attainment of ‘place hands together’ in 71.9% to thumb-finger grasp in 41.1%) and speak 1-word (attainment of ‘non-specific speech’ in 66.7% to ‘speak 1-word’ in 52.3%) (Supplemental Table 3). Finally, despite low milestone acquisition, loss of acquired skills was noted in a small percentage of the cohort, mostly affecting gross motor skills. This was most notable among individuals who were delayed in attainment, in particular complex motor skills. (Fig. 1).

Fig. 1.

Acquisition and loss of developmental milestones in PMD. Milestone acquisition was overall low in the cohort, with most individuals failing to achieve milestones of higher complexity. Loss of milestones was noted in a subset of individuals and impacted higher motor milestones despite low acquisition. The horizontal stacked bar graphs depict the frequency of milestone acquisition (solid blue shaded bars) and loss (striped or dotted bars) in the cohort. Milestones, sorted by age normative attainment age, are presented on the y-axis, and the number of subjects is presented on the x-axis. When a milestone was known to have been acquired, the milestone acquisition was further categorized as ‘attained appropriate’ (dark blue) or ‘attained delayed’ (sky blue) or ‘attained age unknown’ (light blue). Dark grey bars indicate ‘never attained,’ and light grey bars indicate ‘unknown’ status of milestone acquisition. When a milestone was attained, loss of the milestone was indicated as ‘striped regions’ within each attainment category, and when loss status was unknown, this was indicated as ‘dotted regions’. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

The median age at milestone acquisition was computed in the subset of individuals with known milestone attainment status as well as age. Acquisition was noted to be delayed in most individuals, ranging between 55.1% and 100.0% across gross motor skills, 64.9% and 72.2% across fine motor skills, and 32.3% and 68.7% across expressive communication skills. The median age at acquisition was greater than P50 and P75 for all 17 milestones. Delay in milestone acquisition (>P90 + 2SD) was noted to be more pronounced in the motor domains, compared to expressive communication. Delay was noted in a median of 82.3% (IQR: 72.3%, 87.4%) of individuals within the gross motor domain (highest in ‘walk up steps’ milestone: N = 3, 100%), 65.2% (IQR: 65.0%, 68.7%) in the fine motor domain (highest in ‘thumb-finger grasp’ milestone: N = 13, 72.2%) and 47.4% (IQR: 34.9%, 61.5%) in the expressive communication domain (highest in ‘speak 6 words’ domain: N = 11, 68.7%). In addition, a subset of individuals acquired milestones at a much older age than expected, for example, the acquisition of head control at 60 months. The median age (and IQR) at acquisition of each milestone relative to the normative data is depicted in Fig. 2 (and Supplemental Table 4).

Fig. 2.

Milestone acquisition in PMD by age. When known, age at milestone acquisition was summarized using the box plot and compared with normative data from the Denver Developmental Screening tool, 2nd edition. In most subjects, the age at acquisition was largely beyond the threshold of what is considered delayed acquisition. Delayed acquisition was notably pronounced in motor domains compared to expressive communication. The box and whiskers plot depicts the median age of acquisition of each of the milestones (vertical black line within the box) along with inter-quartile range (Q25 and Q75, respectively, indicated by the left and right sides of the box). Within each milestone, the age at attainment at the subject-level is indicated by solid grey circles, with darker shades indicating multiple patients with the same or close age of attainment. With the exception of walk-assisted, for each milestone, the expected target attainment age (P50) is indicated using a dark blue line, and the range P25-P75 is indicated by the light blue shaded area. The dark yellow shaded area shows the range P75-P90, the light-yellow area indicates P90-P90 + 2SD, and the threshold for delayed acquisition (defined as P90 + 2SD) is indicated by the dark orange line. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

3.3. Percent milestone attainment by genotype and presentation

There was no significant difference in overall percent attainment by genotype: 37.5% (IQR: 22.2, 58.3) in overexpression vs 35.3% (IQR: 16.7, 83.3) in others; p = 0.636). Age at disease onset did not differ significantly by genotype: overexpression: 0.10 (IQR: 0.00, 0.25) years vs Others: 0.13 (IQR: 0.00, 0.38) years; p = 0.308.

Disease onset was noted in the neonatal period (defined as onset <1 month of age) in 41.4% of the cohort (N = 46/111). On average, milestone attainment was 31.25% (IQR: 14.2%, 55.6%) among individuals with neonatal disease onset, compared to 42.9% (IQR: 27.8%, 83.3%) in those with onset after the first month of life; p = 0.027 (Table 2).

Table 2.

Summary of early disease features by percent milestone attainment and age or onset Percent milestone attainment was not associated with genotype. Milestone attainment was significantly low in individuals with neonatal disease and stridor at presentation.

Presentation feature	Total	Status	N (%)	Percent Milestone Attainment			Age of symptom onset (years)
				median [Q1, Q3]	[min, max]	P-Value	median [Q1, Q3]	[min, max]	P-value
Feeding Concerns (nutrition)	111	yes	18 (16.22%)	34.31 [17.92, 55.56]	[5.56, 100.00]	0.514	0.00 [0.00, 0.01]	[0.00, 0.50]	<0.001
		no	93 (83.78%)	37.50 [18.75, 72.22]	[0.00, 100.00]		0.16 [0.01, 0.33]	[0.00, 7.00]
Ataxia (movement abnormalities)	111	yes	17 [15.32%]	35.29 [16.67, 54.55]	[6.67, 100.00]	0.583	0.10 [0.01, 0.50]	[0.00, 2.00]	0.609
		no	94 (84.68%)	38.20 [19.62, 72.22]	[0.00, 100.00]		0.10 (0.00, 0.33]	[0.00, 7.00]
Nystagmus	111	yes	81 (72.97%)	37.50 [18.75, 61.11]	[0.00, 100.00]	0.541	0.12 [0.01, 0.30]	[0.00, 1.00]	0.164
		no	30 (27.03%)	42.64 [13.54, 83.33]	[5.56, 100.00]		0.01 [0.00, 0.47]	[0.00, 7.00]
Stridor	111	yes	20 (18.02%]	15.00 [5.56, 29.87]	[0.00, 55.36]	<0.001	0.00 [0.00, 0.03]	[0.00, 0.20]	<0.001
		no	91 (81.98%)	50.00 [27.78, 77.78]	[5.56, 100.00]		0.17 [0.01, 0.36]	[0.00, 7.00]
Tone abnormalities	111	yes	32 (28.83%)	33.33 [18.48, 77.78]	[0.00, 100.00]	0.979	0.10 [0.00, 0.35]	[0.00, 2.00]	0.806
		no	79 (71.17%)	38.89 [18.20, 62.91]	[5.56, 100.00]		0.10 [0.00, 0.33]	[0.00, 7.00]
Onset <1 mo	111	yes	46 (41.44%)	31.25 [14.17, 55.56]	[5.56, 100.00]	0.027
		no	65 (58.56%)	42.86 [27.78, 83.33]	[0.00, 100.00]
Over expression	111	yes	51 (45.95%)	37.50 [22.22 58.34]	[5.56, 100.00]	0.636	0.10 [0.00, 0.24]	[0.00, 0.75]	0.308
		no	60 (54.05%)	35.29 [16.67, 83.33]	[0.00, 100.00]		0.16 [0.00, 0.44]	[0.00, 7.00]

Nystagmus (N = 81, 72.3%) was the most common presenting symptom. This was followed by tone abnormalities (N = 32, 28.8%), stridor (N = 20, 18.02%), feeding concerns such as dysphagia, feeding intolerance, or failure to thrive (N = 18, 16.2%), and ataxia (N = 17, 15.3%). Compared to individuals presenting without stridor (median = 50.0%; IQR: 27.8%, 77.8%), those with stridor at presentation were noted to have a significantly lower percentage attainment of developmental milestones (median = 15.0%; IQR: 5.6%, 29.9%); p < 0.001. Such statistically significant differences in percent milestone attainment were not observed with other features at presentation. Median percent attainment by genotype (overexpression vs others) and presentation is described in Table 2.

Age at disease onset was significantly associated with stridor at presentation: disease onset ranged between 0 and 2.4 months in subjects with stridor at presentation, compared to 0 to 7 years in those presenting without stridor. Similar association was also noted with feeding concerns, with disease onset ranging between 0 and 6 months in subjects with features of feeding concern at presentation, compared to 0 to 7 years in those without feeding concerns. No such associations were noted between the age at disease onset and other presenting features (Table 2).

3.4. Stridor and head control acquisition is associated with developmental trajectories in PMD

Stridor at presentation (due to its significant association with overall percent milestone attainment) and head control acquisition (selected as it is the earliest gross motor milestone and a key prerequisite for achieving higher milestones) were evaluated to determine their association with developmental trajectories in PMD. Independent sitting, thumb-finger grasp, and speak 1-word were selected for initial evaluation, given the marked decrease in percent attainment noted on these milestones (Supplemental Table 3). Percent attainment of independent sitting, thumb-finger grasp, and speak 1-word, by stridor at presentation and head control acquisition, was computed for each well-visit timepoint up to 2 years of age (from 3-month visit onward for head control, as it is typically achieved by this age) (Table 3) and expanded to all other milestones (Supplemental Table 5). Strikingly, none of the patients with stridor at presentation attained independent sitting across any well-visit timepoint. Among individuals who presented with stridor at presentation, thumb-finger grasp was achieved in N = 1/18 individuals and one-word speech in N = 5/20 individuals. Head control was attained in 18.3% (N = 20/109) of the cohort by 3 months of age, with a gradual increase in the proportion of individuals attaining head control up to 9 months of age (N = 26/109, 23.9%), followed by a spike at the 12-month visit (N = 41/109, 37.6%). Between the 9- and 12-month visits, 15/63 individuals were noted to have gained head control. With each subsequent well-visit time point, percent milestone acquisition increased with the attainment of head control and decreased when head control was not attained (Table 3, Supplemental Table 5).

Table 3.

Comparison of key milestone acquisition percentages by stridor presentation and head control acquisitions status at standard well visit timepoints.

Visit (months)	Group	Attainment of Sitting		Attainment of Thumb-finger grasp		Attainment of Speak 1-word
		Total	N (%)	Total	N (%)	Total	N (%)
1	Stridor (presentation)	15	0 (0.0)	13	1 (7.7)	15	4 (26.7)
	No stridor	94	38 (40.4)	77	36 (46.8)	94	53 (56.4)
2	Stridor (presentation)	19	0 (0.0)	17	1 (5.9)	19	5 (26.3)
	No stridor	90	38 (42.2)	73	36 (49.3)	90	52 (57.8)
3	Stridor (presentation)	20	0 (0.0)	18	1 (5.6)	20	5 (25.0)
	No stridor, no head control	69	20 (29.0)	52	19 (36.5)	69	34 (49.3)
	No stridor, head control	20	18 (90)	20	17 (85.0)	20	18 (90.0)
4	Stridor (presentation)	20	0 (0.0)	18	1 (5.6)	20	5 (25.0)
	No stridor, no head control	67	18 (26.9)	50	17 (34.0)	67	32 (47.8)
	No stridor, head control	22	20 (90.9)	22	19 (86.4)	22	20 (90.9)
6	Stridor (presentation)	20	0 (0.0)	18	1 (5.6)	20	5 (25.0)
	No stridor, no head control	65	17 (26.2)	48	16 (33.3)	65	30 (46.2)
	No stridor, head control	24	21 (87.5)	24	20 (85.3)	24	22 (91.7)
9	Stridor (presentation)	20	0 (0.0)	18	1 (5.6)	20	5 (25.0)
	No stridor, no head control	64	16 (25.0)	46	15 (32.6)	63	29 (46.0)
	No stridor, head control	25	22 (88.0)	26	21 (80.8)	26	23 (88.5)
12	Stridor (presentation)	20	0 (0.0)	18	1 (5.6)	20	5 (25.0)
	No stridor, no head control	49	10 (20.4)	33	7 (21.2)	45	18 (37.5)
	No stridor, head control	40	28 (70.0)	39	29 (74.4)	41	34 (82.9)
15	Stridor (presentation)	20	0 (0.0)	18	1 (5.6)	20	5 (25.0)
	No stridor, no head control	48	9 (18.8)	32	6 (18.8)	47	17 (36.2)
	No stridor, head control	41	29 (70.7)	40	30 (75)	42	35 (83.3)
18	Stridor (presentation)	20	0 (0.0)	18	1 (5.6)	20	5 (25.0)
	No stridor, no head control	46	8 (17.4)	30	6 (20.0)	45	15 (33.3)
	No stridor, head control	43	30 (69.8)	42	30 (71.4)	44	37 (84.1)
24	Stridor (presentation)	20	0 (0.0)	18	1 (5.6)	20	5 (25.0)
	No stridor, no head control	45	7 (15.6)	29	6 (20.7)	44	15 (34.1)
	No stridor, head control	44	31 (70.5)	43	30 (69.8)	45	37 (82.2)

3.5. Developmentally coherent cohort stratification

Stridor at presentation and head control acquisition status were associated with developmental milestone acquisition in PMD. Stridor at presentation, in isolation, was associated with the most severe developmental outcomes in PMD, with percent non-attainment ranging between 10.0% and 100.0% across the milestones. Non-attainment was particularly high (70.0% to 100.0%) for gross motor milestones; when attained, milestones were always delayed when stridor was the presenting feature. In the absence of stridor, head control acquisition status was noted to be associated with acquisition of higher milestones. The largest decrease in percent milestone acquisition was observed between the 9-month and 12-month visits: independent sitting (25.0% at 9-months to 20.4% at 12-months), thumb-finger grasp (32.6% at 9-months to 21.2% at 12-months), and speak one-word (46.0% at 9-months to 37.5% at 12-months) (Table 3, Supplemental Fig. 2). In the absence of stridor, the percentage of patients achieving higher milestones was lowest when head control was not acquired by the 12-month well visit (Supplemental Table 5). Lastly, at least one acquired skill was lost in a subset of individuals (N = 11/40, 5 unknown) who did not present with stridor and achieved head control by 12 months of age (Fig. 3).

Fig. 3.

Evolution of developmental milestones by groups. Stridor at presentation, in isolation, was associated with the most severe developmental outcomes. In the absence of stridor, head control acquisition status by 12 months of age was associated with acquisition of higher milestones. The horizontal stacked bar graphs depict the frequency of milestone acquisition (solid blue shaded bars) and loss (striped or dotted bars) in phenotypic subtypes: stridor at presentation, no head control by 12 months, and head control by 12 months (no stridor at presentation). Milestones, sorted by age normative attainment age, are presented on the y-axis, and the number of subjects is presented on the x-axis. When a milestone was known to have been acquired, the milestone acquisition was further categorized as ‘attained appropriate’ (dark blue) or ‘attained delayed’ (sky blue), or ‘attained age unknown’ (light blue). Dark grey bars indicate ‘never attained,’ and light grey bars indicate ‘unknown’ status of milestone acquisition. When a milestone was attained, loss of the milestone was indicated as ‘striped regions’ within each attainment category, and when loss status was unknown, this was indicated as ‘dotted regions’. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Based on this, we defined three developmentally coherent PMD subtypes: a ‘profoundly impaired’ subtype consisting of individuals “presenting with stridor” (N = 20), a ‘severely impaired’ subtype with “no stridor at presentation and no head control by 12 months of age” (N = 50) and a ‘moderately impaired’ subtype with “no stridor at presentation and head control attained by 12 months of age” (N = 41). The distribution of PLP1 genotypic variants within the identified PMD subtypes was tabulated (as shown in Table 4). PLP1 overexpression was noted in the majority of the severely (N = 30/50; 60.0%) and moderately (N = 20/41; 48.8%) impaired cohorts. Also, 12.0% of (N = 6/50) subjects with severe impairments had a loss-of-function variant. Further, 85.0% (N = 17/20) of the profoundly impaired cohort were classified with a GOF missense variant.

Table 4.

Distribution of the PLP1 genotypic variations within the developmentally coherent PMD subtypes.

Genotype	Profoundly impaired (stridor at presentation) N = 20	Severely impaired (no stridor, no head control) N = 50	Moderate phenotype (no stridor, head control) N = 41
Overexpression	1 (5.0%)	30 (60.0%)	20 (48.8%)
Other
GOF Missense	17 (85.0%)	7 (14.0%)	12 (29.3%)
Loss of function	0 (0.0%)	6 (12.0%)	1 (2.4%)
HEMS/Splicing	1 (5.0%)	7 (14.0%)	8 (19.5%)

3.6. Time to event

The time to the occurrence of key medical events related to disease progression in PMD was computed for the overall cohort (Supplemental Table 6) and by genotype (Supplemental Fig. 3). On average, the subjects were followed for 10.9 (Total range: 1.58, 38.60) years since the first medical encounter. Distribution of the common PMD-related complications were nystagmus in 91.5% (N = 97/106), failure to thrive in 63.3% (N = 38/60), feeding tube placement in 49.4% (N = 44/89), stridor in 46.3% (N = 25/54), scoliosis in 45.2% (N = 38/84), hip dislocation in 45.3% (N = 34/75), incontinence in 40.9% (N = 27/66), artificial ventilation in 24.3% (N = 18/74), seizures in 14.0% (N = 12/86), and aspiration pneumonia in 11.9% (N = 8/67). The overall time to key medical events is reported in Supplemental Table 6. Nystagmus was the earliest event noted in the cohort (median years = 0.2; 95% CI: 0.1, 0.2). This was followed by failure to thrive (median years = 4.4; 95% CI: 1.6, 13.9) and hip dislocation (median years = 10.5; 95% CI: 8.0, 14.6) in the first decade of life, and feeding tube placement (median years = 13.0; 95% CI: 8.0, 20.0), scoliosis (median years = 13.0; 95% CI: 10.8, 14.2) and incontinence (median years = 18.6; 95% CI: 15.0, NA) in the second decade. On stratifying by genotype (overexpression vs others), significant differences were noted in the distribution of time to occurrence of nystagmus (log-rank test; p = 0.005), aspiration pneumonia (log-rank test; p = 0.004), and artificial ventilation (log-rank test; p = 0.036). This was not evident for any other key medical events in the PMD cohort (Supplemental Fig. 3).

Comparison of the distribution in time to the occurrence of key medical events by the PMD developmental subtypes showed statistically significant differences in time to seizure (log-rank test; p = 0.005), scoliosis (log-rank test; p = 0.001), failure to thrive (log-rank test; p = 0.001), feeding tube placement (log-rank test; p < 0.001), and artificial ventilation (log-rank test; p < 0.001), and time to nystagmus (log-rank test; p = 0.043). There were no significant differences in the time to hip dislocation (log-rank test; p = 0.374), or incontinence (log-rank test; p = 0.841) by the PMD developmental subtypes. Feeding tube placement (median years = 0.8; 95% CI: 0.3, NA) was noted to occur before the first note of failure to thrive in the profoundly impaired subtype (median years = 1.0; 95% CI: 0.2, NA). In contrast, in the severely impaired and moderately impaired cohorts, respectively, feeding tube placement (median years = 13.0 (95% CI: 7.0, NA) and 20.0 (95% CI: 13.0, NA)) followed failure to thrive (median years = 7.0 (95% CI: 1.4, NA) and 13.9 (95% CI: 10.6, NA)). Scoliosis was noted in the profoundly impaired cohort at 9.0 (95% CI: 2.5, NA) years, compared to 13.0 (95% CI: 10.0, NA) years in the moderate and 13.8 (95% CI: 11.5, NA) years in the severely impaired cohort. Less than 50% of the subjects classified as having severe impairments developed seizures, aspiration pneumonia, or required artificial ventilation, and among those classified as moderately impaired, less than 50% developed aspiration pneumonia. Time to medical events by subtype is illustrated in Fig. 4.

Fig. 4.

Survival plots depicting the time to onset of key medical events by the developmentally coherent PMD subtypes: profoundly impaired (stridor at presentation), severely impaired (no stridor at presentation and non-attainment of head control by 12 months), and moderately impaired (no stridor at presentation and head control attainment by 12 months).

4. Discussion

PMD is a phenotypically heterogeneous continuum disorder [1,4,9], in which variability in disease severity and trajectory cannot be explained by genotype alone [9,12,13,19,20]. Therefore, in PMD, there is a need for alternative tools to stratify disease cohorts and characterize disease trajectories. There are ongoing studies [38] (NCT05659901) to evaluate biomarker-based approaches to characterize disease progression in PMD-affected individuals. While biomarker-based approaches are critical, their validation as a prognostic marker requires a complete understanding of disease pathophysiology, including the potential for genetic and environmental variability [39,40], which is not yet fully understood [9,12]. A phenotype-based approach enables the evaluation of the full spectrum of PMD manifestations and is more reflective of the patient’s lived experiences, and can advance clinical trial readiness for PMD. The current study is the first to explore a phenotype-based approach to stratify disease cohort and characterize disease outcomes. This work builds on existing evidence on developmental outcomes of PMD and explores associations between early disease features and developmental outcomes as well as disease progression in PMD.

To this end, we described the acquisition and loss of developmental milestones across the phenotypic spectrum captured in the GLIA-PMD natural history study. Overall, milestone acquisition was low in the cohort, and there was a notable decrease in the proportion of individuals acquiring skills with increasing complexity of execution [17]. Motor milestones were relatively more impacted than social or expressive communication. Within the motor domain, gross motor milestones were more severely impacted and often delayed when acquired. Relative to gross motor milestones, more individuals gained expressive communication skills, which were also preserved in most upon attainment. These findings bolster existing anecdotal evidence of severe neuro-developmental impairments in PMD [9,13,17,19]. However, in contrast to the existing literature, which describes a moderate disease course in the majority of subjects with PLP1 overexpression/duplication [9,10,41], the study reports milestone non-attainment in nearly 50% of this cohort. This may be associated with between-subject variations in the number, size, and location of duplication mutations [9,19], and should be examined comprehensively. Next, this study reports a loss of acquired skills in a subset of individuals, a finding that is sparingly described in PMD [13,14]. This finding confirms regression as a feature of PMD and warrants further research, including as it relates to age and concomitant tone abnormalities. Furthermore, social milestones (i.e., the ability to smile responsively) appear to be relatively preserved in the cohort, raising a critical question about the extent to which receptive communication may be preserved in PMD and should be assessed in a prospective study.

This study is the first to investigate the associations between early disease features and disease progression in PMD and define phenotypically coherent subtypes. Presentation with stridor and head control acquisition status by 12 months were associated with developmental outcomes in PMD. In particular, stridor at presentation was the most important determinant of milestone acquisition and was associated with profound neurodevelopmental impairments. While still impaired, expressive communication was relatively less impacted than the motor milestones, despite its direct clinical relevance to stridor. The mechanism by which stridor at presentation impacts milestones acquisition, particularly the motor milestones, is unknown. We hypothesize that this may in part be related to the more severe nature of this subtype, which may also explain the premature need for feeding tube placement at 7–8 months (and as early as 4 months in some subjects) of life with subsequent onset of failure to thrive despite initiation of feeding tube (Supplemental Table 6). In the absence of stridor, our study identified head control acquisition status by 12 months as the next most important determinant of neurodevelopmental outcomes in PMD. The lack of head control by 12 months was associated with a lower percent attainment of higher milestones across the domains. In addition, compared to the moderate subtype, the onset of hip dislocation, failure to thrive, feeding tube placement, and urinary incontinence was also observed to occur early in the severe subtype, and may be partly explained by the limited mobility of these subjects. Nevertheless, given that these manifestations are multifactorial, it is essential to investigate this within the framework of a more extensive future study. Lastly, our study showed loss of function variants in over 10% of the severe subtype, which is in contrast to existing literature which report a mild disease course [9], warranting further study.

Our study approach has inherent limitations. First, the data collected in the study were primarily sourced from medical records, which are not purpose-built for research. Therefore, heterogeneity in the duration of medical records and data missingness pose limitations for the study and may lead to variability of the data available for analysis. To reduce the data gap, we adhered to rigorous standardized protocols for data collection from medical records [30] and, when feasible, supplemented the collection of milestone data through caregiver interviews. Still, the study findings must be interpreted in the context of this limitation. For example, despite efforts to close the gap in missing data, the limited availability of datapoints on loss of skills precludes our ability to study neuromotor regression in PMD comprehensively. To address these limitations, the study findings have been presented and interpreted with caution in the context of data missingness. Second, the average follow-up duration of the study cohort was relatively short (median: 7.80 (IQR 4.54, 16.40) years). This limits our ability to more reliably estimate the ranges of the upper or lower confidence intervals in the survival analysis of some clinical events, particularly those anticipated to occur later in the disease course. Shorter follow-up duration also limited our ability to reliably collect data on mortality in the cohort. Finally, compared to the included cohort, follow-up duration was also noted to be shorter in the excluded cohort. Collectively, these limitations underscore the need for longer follow-up to fully characterize PMD disease progression. Third, gaps in data points related to milestones were addressed through interviews in a subset of subjects and may be subject to recall bias. To address this, only trained individuals administered the interviews, and as needed, additional probing questions were asked to establish temporality with other milestones and events were used to reconfirm the acquisition/loss age. In addition, ascertainment bias is possible, as only about a third of the cohort participated in the interview. Therefore, we assessed systematic differences in the baseline characteristics between the interviewed and non-interviewed cohorts that could potentially impact the data collected through interviews. This did not reveal any significant differences. Fourth, data on receptive communication could not be collected in the context of this study, limiting the ability to fully study the developmental outcomes within the communication domain. A prospective study is needed to characterize receptive communication. Fifth, the number and distribution of subjects within each identified subtype were small, limiting our ability to fully understand and explain the phenotypic patterns of disease progression across subtypes, necessitating a larger study with even distribution by subtype. Finally, the categorization of variants for genotyping and the pathogenetic mechanism by which variants result in the disease is yet to be fully understood in PMD [9,12,13,19,20]. Hence, in this study, the analysis of genotypes was binarized as ‘overexpression’ for duplications and ‘other’ to include GoF missense, loss of function, and splicing or HEMS-causing variants. A future study of developmental outcomes and disease progression in the context of variant type and location of variant within the PLP1 gene is needed.

5. Conclusions

This is the first study to systematically evaluate neurodevelopmental outcomes and explore associations between early disease features and disease progression in PMD in a geographically diverse cohort. By utilizing approaches widely applied in multiple other leukodystrophies [35–37,42–45], the study enabled access to and analysis of pre- and post-diagnostic data to identify developmentally coherent cohorts based on early disease features that were associated with developmental outcomes in PMD. This study is a crucial advancement toward clinical trial readiness in PMD and informs the determination of treatment eligibility and the selection of appropriate clinical trial endpoints in future PMD trials.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.ymgme.2026.110154.

Data availability

Data supporting the study’s findings are not publicly available due to privacy and ethical concerns with rare diseases.