Symptom Onset in Classic Rett Syndrome: Analysis of Initial Clinical Severity Scale Entries

ABSTRACT

Objective

We aim to assess the clinical features of Rett syndrome (RTT) at registration into the National Institutes of Health–sponsored natural history study (NHS) using the Clinical Severity Scale (CSS).

Introduction

The CSS was established in 2000 to assess characteristics of individuals with RTT and related disorders. We analyzed the CSS at enrollment into the NHS of all individuals with classic RTT.

Methods

The CSS of 1258 individuals was used to examine three historical items (age at regression, age at onset of hand stereotypies, and head growth) and 10 clinical features at initial enrollment.

Results

Among historical items, age at regression was most prominent after age 12 months with 8% regressing before this, hand stereotypies were most common after 18 months with 30% occurring before, and head growth was substantially lower overall. Hand use, ambulation, and communication skills were reduced in most individuals.

Discussion and Conclusion

These findings reflect core clinical criteria for classic RTT and mirror the top concerns registered independently by parents or caregivers. The CSS is an important analytic tool for the association of relative differences in clinical outcome with specific MECP2 variant groups and has been important in establishing entrance criteria in RTT clinical trials. The range of age‐specific CSS features will continue to inform the RTT natural natural history as well as provide stratification and selection in future clinical trials, especially those involving younger participants.

1. Introduction

Rett syndrome (RTT) is a neurodevelopmental disorder that was first recognized more than 60 years ago. Among the tools used to assess individuals with RTT, the Clinical Severity Scale (CSS) has been a standard measure, used throughout the US natural history study (NHS) from 2003 through 2021. Originally developed by Amir et al. [1], the CSS was subsequently modified before the inception of the NHS to improve the clarity of this 13‐item scale [2] and has been used over the past 21 years to assess disease severity in RTT and related disorders for reports on natural history studies and for stratification in clinical trials [3, 4, 5, 6, 7]. In examining this scale, the first three items (age at onset of regression, onset of stereotypies, and head growth) were based on historical evidence and were not expected to change during repeated assessments. The remaining 10 items, however, were determined at the time of each clinical evaluation. Each item can be assessed using Likert scales of 0–4 or 0–5 with an aggregated scoring range of 0–58, with higher scores reflecting greater severity. Although the CSS has proved to be an excellent measure of clinical severity, it is not an effective outcome measure, in part due to the stability of the three historical items, which represent nearly 25% of the total. This lack of ability to change the score after these milestones have been completed limits the CSS's effectiveness in assessing the clinical outcome over time.

Although the CSS may be limited in its ultimate use as an outcome measure in a clinical trial, the data obtained in the NHS from the CSS can be used to validate the clinical features of Rett and provide an assessment of some individual features in the population.

This present analysis was constructed from the final US NHS data set for classical RTT to provide a comprehensive assessment of the distribution of CSS scores at the time of enrollment.

2. Methods

The US NHS database was used to assess the CSS at the time of enrollment for 1259 individuals with classic RTT. However, as a birth date was not reported for one individual, the total score is based on 1258 participants. The age at enrollment was 10.4 years as those enrolled could have been diagnosed more than 20 years before the start of the NHS. The average age at diagnosis declined from 3.5 to 2.7 years across the duration of the study. The CSS was completed by the responsible physician in each of the NHS clinics across the span of 2006–2021. The demographics for this group were 85% White, 4.5% African American, and 14.5% Hispanic/Latino, as recently published [8]. The head circumference was assessed at each visit and compared with historical records, including the birth head circumference and measurements acquired by the respective physician for each. In a few instances (< 20), historical data were missing due to the adoption or destruction of records. In these instances, the CSS measurement was an extrapolation of the value at the time of enrollment. In the first two funding cycles, assessments occurred at Baylor College of Medicine, the Greenwood Genetic Center, Boston Children's Hospital (Harvard), or the University of Alabama at Birmingham. During this period, travel clinics were also conducted twice yearly by members of the four research teams at Oakland Children's Hospital, Rush Medical Center (Chicago), Robert Wood Johnson Medical School (New Brunswick, NJ), or two separate sites in Florida (Miami and Tampa). During this period, individuals seen at the respective travel clinics accounted for > 40% of enrollment. During the last funding period, assessments continued at the original four sites with additional enrollments at Children's Hospital of Philadelphia (University of Pennsylvania), Vanderbilt University Medical Center, Cincinnati Children's Hospital, St. Louis Children's Hospital (Washington University), Rush Medical Center, Gillette Children's Hospital (St. Paul), University of Colorado School of Medicine (Children's Hospital of Colorado), University of California, San Diego, and University of California, San Francisco (Oakland Children's Hospital).

The 13 individual items comprising the CSS were assessed to determine the distribution of scores for each item shown below in the specific tables. The CSS from the first visit to the NHS was analyzed for this article. Description statistics using Microsoft Excel were used to describe the data.

2.1. Human Study Approval

Human study approval was obtained for each participant before entry into the study. A Certificate of Confidentiality was provided by the National Institute of Child Health and Human Development.

3. Results

In total, 1258 individuals with classic RTT completed at least one CSS. The large majority had a known genetic variant at enrollment, although about 4% lacked evidence of an MECP2 variant. Nonetheless, those with classic RTT were included, whether they had a defined variant or not. Most individuals in the NHS were White despite our best efforts to include all racial and ethnic groups. Although some individuals did not meet all criteria at enrollment, they subsequently did meet these criteria, such as demonstrating regression or loss of language skills, and were therefore included. The average age at this first CSS was 10.4 years (median = 10.4 years, standard deviation (SD) = 9.3 years, range = 1.1–66.5 years). The age at enrollment was not the same as the age at diagnosis because the diagnosis could have been more than 20 years before enrollment. The total score for the CSS ranged from 5 to 45 with a mean of 22.5 (SD = 7.5) and a median of 22.0 (interquartile range was 17.0 and 28.0). The results for each of the 13 CSS items are shown in the following tables grouped by (1) the three historical items (Table 1) and (2) the 10 initial exam items (Table 2). These results include the means (SD), medians, interquartile values, and ranges of each item.

Table 1A.

Age of onset of regression.

Regression age	Number	Percent
No regression	1	0.08
> 30 months	114	9.1
18–30 months	579	46.1
12 to < 18 months	455	36.2
6 to < 12 months	90	7.2
< 6 months	15	1.2
Missing	4	0.2
Total	1258	100

Note: Mean = 2.4 (SD = 0.8); median = 2.0; interquartile values = 2 and 3; range is 0–5.

Table 2A.

Somatic growth.

Somatic growth at visit	Number	Percent
BMI 26th to –50th percentile or above	657	52.2
BMI 11th to 25th percentile	253	20.1
BMI 5th to 10th percentile	91	7.2
BMI < 5th percentile	132	10.5
BMI « 5th percentile	119	9.5
Missing	6	0.5
Total	1258	100

Note: Mean = 1.0 (SD = 1.4); median = 0; interquartile values = 0 and 2; range is 0–4.

Abbreviation: BMI, body mass index.

3.1. Historical Items

The three historical items, (1) age at onset of regression, (2) age at onset of stereotypies, and (3) early head circumference, were not expected to change and could possibly be related to later recall bias, over the course of each individual with RTT.

Age at onset of regression (Table 1A) occurred between 12 and 30 months in more than 80% of individuals, with 36.2% occurring between 12 and 18 months of age and 46.1% between 18 and 30 months. Equal numbers (~10%) regressed before 12 months and after 30 months. Stereotypies (Table 1B) began most frequently (55.6%) at 18–36 months, whereas 22.7% developed these from 12 to 18 months. Head growth (Table 1C) was normal in 27.6% and microcephalic by age 2 years in 28.6%. An additional 30.4% had occipital‐frontal circumferences between the 2nd and 10th percentile by 24 months of age. The head circumference data reflect the longitudinal head circumference increases of Tarquinio et al. [9], indicating that nearly 25% of individuals with RTT have head circumferences in the normal range. This finding validated the removal of microcephaly from the more recent diagnostic criteria for RTT [10, 11].

Table 1B.

Age of onset of stereotypies.

Stereotypy age	Number	Percent
> 10 years	8	0.6
36 months to < 10 years	169	13.4
18 to < 36 months	699	55.6
12 to < 18 months	286	22.7
< 12 months	93	7.4
Missing	3	0.2
Total	1258	99.9

Note: Mean = 2.4 (SD = 0.8); median = 2.0; interquartile values = 2 and 3; range is 0–5.

Table 1C.

Head growth.

Head circumference	Number	Percent
No/minimal deceleration	347	27.6
Deceleration > 2 centiles‐OFC > 10th after 24 months	163	13.0
2nd to 10th centile after 24 months	242	19.2
2nd to 10th centile before 24 months	141	11.2
< 2nd centile by 24 months	360	28.6
Missing	5	0.4
Total	1258	100

Note: Mean = 2.0 (SD = 1.6); median = 2.0; interquartile values = 0 and 4; range is 0–4.

Abbreviation: OFC, occipital‐frontal circumference.

3.2. Initial Exam Items

Initial somatic growth (Table 2A) and independent sitting (Table 2B) scores indicated that the majority of individuals had normal values, although significant levels of abnormality were evident. In terms of somatic growth, 21% of individuals had somatic growth values < 5th percentile. For independent sitting, significant delays occurred in > 27% and were completely lost, after initial acquisition, in another 11.5%.

Table 2B.

Independent sitting.

Sitting ability	Number	Percent
Sits alone < 8 months	688	54.7
Sits delayed > 8 months	293	23.3
Sits delayed > 18 months	34	2.7
Sits delayed > 30 months	19	1.5
Lost	145	11.5
Never acquired	76	6.0
Missing	3	0.2
Total	1258	99.9

Note: Mean = 1.1 (SD = 1.6); median = 0; interquartile values = 0 and 1; range is 0–5.

Conversely, more than 70% of individuals had delays in ambulation (Table 2C) and nearly 100% had abnormalities in hand function (Table 2D), both in keeping with two of the four main diagnostic criteria for classic RTT. Although the ability to ambulate occurred in 27%, this scale does not capture the quality of walking, which would be expected to be abnormal. More than 31% had delayed acquisition, and 8% had lost the previously acquired ambulation skill. In addition, nearly 34% had never acquired the ability to ambulate.

Table 2C.

Ambulation.

Walking ability	Number	Percent
Acquired < 18 months	340	27.0
18 to < 30 months alone	226	18.0
> 30 months alone	45	3.6
> 50 months with help	120	9.5
Lost	101	8.0
Never acquired	423	33.6
Missing	3	0.2
Total	1258	99.9

Note: Mean = 2.5 (SD = 2.1); median = 3.0; interquartile values = 0 and 5; range is 0–5.

Table 2D.

Hand use.

Hand use ability	Number	Percent
Acquired and conserved	81	6.4
Acquired on time and partially conserved	467	37.1
Acquired late and partially conserved	157	12.5
Holding objects acquired and lost	520	41.3
Never acquired	30	2.4
Missing	3	0.2
Total	1258	99.9

Note: Mean = 1.96 (SD = 1.1); median = 2.0; interquartile values = 1 and 3; range is 0–4.

Although scoliosis (Table 2E) was absent in greater than 50%, the appearance of abnormal curvature in nearly 35% as well as the evidence of surgical correction in 11.8% indicated nearly half having significant difficulties with spinal deformity.

Table 2E.

Scoliosis.

Presence of scoliosis	Number	Percent
None	672	53.4
1° to < 20°	241	19.2
20° to < 40°	133	10.6
40°to < 60°	44	3.5
≥ 60°	17	1.4
Surgery	148	11.8
Missing	3	0.2
Total	1258	100.1

Note: Mean = 1.15 (SD = 1.7); median = 0; interquartile values = 0 and 2; range is 0–5.

Nearly all participants had delays in language development (Table 2F) and 90% had abnormalities in nonverbal communication (Table 2G), again reflecting significant difficulties in one of the main diagnostic features for classic RTT. The inability to communicate accurately with parents, siblings, or other caregivers is the most significant parental concern as determined in the NHS study [12].

Table 2F.

Language.

Language ability	Number	Percent
Preserved, contextual	3	0.2
Short phrases only	15	1.2
Single words	147	11.7
Vocalization, babbling	864	68.7
Screaming, no utterances	226	18.0
Missing	3	0.2
Total	1258	100

Note: Mean = 3.0 (SD = 0.6); median = 3.0; interquartile values = 3 and 3; range is 0–4.

Table 2G.

Nonverbal communication.

Nonverbal skill level	Number	Percent
Preserves/propositive	127	10.1
Good eye contact > 30 s	201	16.0
Intermittent eye contact 5 to < 30 s	676	53.7
Infrequent eye contact < 5 s	241	19.2
Lost and not regained	6	0.5
None	4	0.3
Missing	3	0.2
Total	1258	100

Note: Mean = 1.8 (SD = 0.9); median = 2.0; interquartile values = 1 and 2; range is 0–5.

In terms of respiratory function (Table 2H), nearly 50% presented evidence of significant breath‐holding or hyperventilation (or both), and more than 25% had some degree of abnormality. This is a critical issue for parents and caregivers as well, as it can interfere significantly with mealtimes and therapy in the school or home settings.

Table 2H.

Respiratory dysfunction.

Hyperventilation/apnea	Number	Percent
None	351	27.9
Minimal < 10% of time	346	27.5
Intermittent 50% of time	350	27.8
Constant 100% of time	194	15.4
Constant with cyanosis	14	1.1
Missing	3	0.2
Total	1258	99.9

Note: Mean = 1.3 (SD = 1.1); median = 1.0; interquartile values = 0 and 2; range is 0–4.

About 60% had abnormal autonomic function (Table 2I) with cool or cold hands and/or feet. Although this characteristic did not seem to bother individuals with RTT, it remains a matter of some concern for their parents or other caregivers.

Table 2I.

Autonomic features of hands and feet.

Autonomic features	Number	Percent
None	500	39.7
Pink but cool	486	38.6
Mottled and cold	179	14.2
Blue‐purple and cold hands or feet	82	6.5
Blue‐purple and cold hands and feet	8	0.6
Missing	3	0.2
Total	1258	99.8

Note: Mean = 0.89 (SD = 0.9); median = 1.0; interquartile values = 0 and 2; range is 0–4.

Lastly, epilepsy (Table 2J) was present in slightly less than 40%, which differs from the overall expectancy of epilepsy reaching 90% throughout life [13, 14]. These data indicate that the point prevalence of epilepsy is about 40%. Nonetheless, these initial findings support previous reports that the occurrence of seizures becomes more prominent in RTT after age 5 years [13, 14, 15, 16].

Table 2J.

Epilepsy/seizures.

Presence of epilepsy	Number	Percent
Absent	791	62.9
< monthly	161	12.8
< weekly to monthly	97	7.7
Weekly	69	5.5
More than weekly	99	7.9
Daily (intractable)	38	3.0
Missing	3	0.2
Total	1258	100

Note: Mean = 0.91 (SD = 1.5); median = 1.0; interquartile range = 0 and 1; range is 0–5.

4. Discussion

The utilization of the CSS in studies of RTT and related disorders was amplified dramatically with the establishment of the RTT NHS in 2003 (enrollment beginning in 2006). With the termination of the NHS in 2021, analysis of the CSS in 1258 individuals with classic RTT allowed a retrospective look at the range of scores for each of the 13 items. This analysis indicated that the CSS results mapped closely to the top five caregiver concerns reported in 2024 [12]. These top five concerns were lack of effective communication, seizures, walking/balance issues, lack of hand use, and constipation. The top four are components of the CSS, including both verbal and nonverbal communication, epilepsy, ambulation, and hand use. Four of the remaining eight CSS items (repetitive hand movements, breathing issues, scoliosis, and poor weight gain) are included in the top 16 parental concerns. The other four items (regression, head growth, sitting, and autonomic changes) were not included in the list of potential items that parents could select. Two of these (regression and head growth) are not expected to be areas of ongoing concern.

Overall, the CSS provided important indicators of specific involvement that correlate with clinical outcomes. As RTT is associated with MECP2 variants in > 96% of those fulfilling the clinical criteria, a specific relationship between the CSS level and individual MECP2 variants has been documented [2, 17, 18]. Cuddapah and colleagues [18] reported a slow increase in CSS values over time in terms of both the eight common point variants and specific categories of variants such as deletions, insertions, and 3′ truncations, and the subsequent analyses of Merritt and colleagues [19] reinforced these findings.

More recently, the CSS became an important qualifier for entry into various clinical trials, including the most recent phase 3 trial for trofinetide, which led to FDA approval of this agent for individuals with RTT [20].

Potential limitations of this study relate to the number of different clinicians completing the CSS. Specific efforts were conducted to clarify and standardize CSS input, but this could be a possible limitation. A second potential issue is the large age range of enrollees, resulting in the potential effect of recall errors on the completion of the CSS historical items (age at regression, onset of stereotypies, and head growth). Maximal effort was made to obtain medical records and to relate specific events of importance (birthdays, holidays, etc.) to improve accuracy and recall of these specific historical items. In addition, with the baseline visits occurring at the wide age range reported, the individual CSS score for each item could vary significantly with respect to the cross‐sectional analyses. A third limitation is the lack of comparison with other registries, namely, the European and Australian registries. In the early years of the NHS, we endeavored to harmonize the three registries to address these CSS items as well as other issues collected. However, data collection in each registry differed significantly, and the severity measures used were not the same. As a result, it was impossible to generate a direct comparison.

5. Conclusion

The CSS has been used consistently in the US NHS as an indicator of clinical involvement in RTT and related disorders. The initial assessment in 1258 individuals with classic RTT provided an important measure of the level and range of involvement and continues to be important in correlating clinical outcome with MECP2 variants as well as providing a useful measure for entry into clinical trials.

Author Contributions

Alan K. Percy: conceptualization, investigation, funding acquisition, writing – original draft, methodology, validation, writing – review and editing, formal analysis, project administration, supervision. Jeffrey L. Neul: investigation, methodology, validation, writing – review and editing, formal analysis. Amitha Ananth: investigation, methodology, validation, writing – review and editing. Timothy A. Benke: investigation, methodology, validation, writing – review and editing, formal analysis. Eric D. Marsh: investigation, methodology, validation, writing – review and editing, formal analysis.

Conflicts of Interest

Dr. Percy has received research support from the NIH and has been a site PI for Acadia Pharmaceuticals. He is a consultant for Acadia Pharmaceuticals, Taysha Gene Therapies, and Neurogene. He has prepared educational materials for WebMD, Medscape, Pharmacy Times Continuing Education, Prime Inc., and the CME Institute. Dr. Neul has received research funding from the National Institutes of Health, the International Rett Syndrome Foundation (IRSF), and Rett Syndrome Research Trust (RSRT); clinical trials with Acadia Pharmaceuticals, GW Pharmaceuticals, Ionis Pharmaceuticals, Neurogene, Neuren Pharmaceuticals, Newron Pharmaceuticals, and Taysha Gene Therapies; personal consultancy for Acadia Pharmaceuticals, Alcyone Lifesciences, AveXis, GW Pharmaceuticals, Hoffmann‐La Roche, Ionis Pharmaceuticals, Jazz Pharmaceuticals, Myrtelle, Neurogene, Newron Pharmaceuticals, Ovid Therapeutics, and Taysha Gene Therapies; participation as a member of data safety monitoring boards for clinical trials conducted by Ovid Therapeutics and Ultragenyx. Dr. Ananth was a site PI for Acadia Pharmaceuticals. Dr. Benke has received research funding from GRIN2B Foundation, the International Foundation for CDKL5 Research, Loulou Foundation, the National Institutes of Health, and Simons Foundation; consultancy for Alcyone, AveXis, GRIN Therapeutics, GW Pharmaceuticals, the International Rett Syndrome Foundation, Marinus Pharmaceuticals, Neurogene, Ovid Therapeutics, and Takeda Pharmaceutical Company Limited; clinical trials with Acadia Pharmaceuticals Inc., GW Pharmaceuticals, Ionis, Marinus Pharmaceuticals, Neurogene, Ovid Therapeutics, and Rett Syndrome Research Trust; all remuneration has been made to his department. Dr. Marsh is site PI for UCB Pharmaceuticals, Takeda Pharmaceuticals, Epygenix Pharmaceuticals, Ionis Pharmaceuticals, and SK Life Pharmaceuticals. He has received research support from NIH, Penn Orphan Disease Center, International Rett Syndrome Foundation, International Foundation for CDKL5 Research, and Curaleaf Inc., is a consultant for Acadia Pharmaceuticals and Mahzi Pharmaceuticals, and has prepared an educational program for Medscape and France Foundation.