Effectiveness and safety of Risdiplam for types 1–3 spinal muscular atrophy in a single center

Abstract

Background

Risdiplam has demonstrated efficacy in various types of 5q-spinal muscular atrophy (SMA) during clinical trials, yet real-world data remain limited. This study evaluated its effectiveness and safety in Chinese SMA children and explored influencing factors.

Methods

This retrospective study included genetically confirmed SMA patients treated with risdiplam at Children’s Hospital of Fudan University from August 2021 to August 2024. We analyzed data on the age of onset, types, SMN2 copy number, treatment initiation age and duration, respiratory and feeding status, motor function and safety.

Results

A total of 12 SMA patients (4 type 1, 7 type 2, 1 type 3) were included, with a median age of 14.4 months (range: 1.7-161.7 months), was evaluated over a follow-up period of 8–18 months after starting risdiplam. Motor function improvements were well noted in 11 children, with Children’s Hospital of Philadelphia Infant Test of Neuromuscular Disorders scores increasing by 1.73 points monthly. Notably, nine patients attained new World Health Organization motor milestones after an average of 9.8 months post-treatment. Subgroup analysis indicated that younger age and earlier intervention were associated with more pronounced clinical benefits. Adverse events, primarily respiratory infections were the most common adverse event but did not lead to deaths or treatment discontinuations, likely attributable to the underlying disease rather than risdiplam—were observed but did not result in treatment discontinuation or mortality.

Conclusion

Risdiplam treatment is effective and safe in Chinese pediatric SMA patients, enhancing motor function and facilitating new milestones achievement. Early diagnosis and treatment are vital for maximizing benefits.

Introduction

Classic 5q spinal muscular atrophy (SMA) is a progressive motor neuron disorder characterized by muscular atrophy and generalized muscular and bulbar weakness [1, 2]. It is the most common genetic cause of mortality in infants under 2 years old [3] with an incidence of approximately 1 in 11,000 live births [4]. SMA results from homozygous deletions or loss-of-function mutations in SMN1, leading to insufficient expression of survival motor neuron (SMN) protein, which is essential for lower motor neuron function [5, 6]. In China, prior to the availability of disease-modifying therapies (DMTs), the survival probabilities for SMA type 1 at 1, 2, and 5 years were 44.9%, 38.1%, and 29.3%, respectively, whereas type 2 or 3 SMA patients maintained 100% survival through 1, 2, and 5 years [7].

Several DMTs have been developed to increase SMN protein levels. Although SMN2, a paralog of SMN1, can partially compensate for SMN protein deficiency, most SMN2-derived transcripts are nonfunctional due to exon 7 skipping during splicing [8]. Risdiplam, an oral SMN2 splicing modifier that enhances full-length SMN protein production [9], was approved in China for SMA treatment in 2021.

Various factors need to be considered when evaluating efficacy, such as SMA types, disease severity, duration, rehabilitation care, comorbidities, and unknown genetic interactions [10]. This study summarized and analyzed the clinical features of SMA children, who were treated with risdiplam monotherapy for more than 6 months and assessed at least twice, to demonstrate the effectiveness and safety of risdiplam in the real world. Additionally, this study further explored the factors influencing clinical benefits, which may help physicians to form better expectations in different patients.

Methods

This retrospective study included genetically confirmed SMA patients treated with risdiplam at Children’s Hospital of Fudan University from August 2021 and it was conducted with the approval of the hospital’s Ethics Committee.

Inclusion criteria included: (1) confirmed diagnosis of 5q-SMA; (2) monotherapy with risdiplam; (3) participated in clinical assessments for baseline and follow-up. Patients with follow-up < 6 months were excluded.

The following data were collected up to August 2024, including onset and diagnosis age, SMN2 copy number, risdiplam initiation age, respiratory and feeding conditions, motor function and safety. Children’s Hospital of Philadelphia Infant Test of Neuromuscular Disorders (CHOP-INTEND) [11], Hammersmith Functional Motor Scale Expanded (HFMSE) [12], the motor milestones portion of the Hammersmith Infant Neurological Exam - Part 2 (HINE-2) [13] and World Health Organization (WHO) motor milestones [14] were used to assess motor function during baseline and follow-up. To better analyze the effectiveness of risdiplam treatment, the patients were divided into ‘responders’ and ‘non-responders’ according to the following criteria: 1) An increase of at least 4 points in the CHOP-INTEND score [15, 16] is considered as response; 2) HINE-2 response was defined as improvement in at least 1 category and more categories with improvement than categories with worsening [15] ; 3) The HFMSE response definition utilized minimal clinically important difference (MCID) improvement values reported by G Coratti et al., where the MCID for improvement was 1.5 points for SMA type 2, 2.4 points for SMA type 3) [17]. Given the integer-based HFMSE scoring system, these MCID thresholds were conservatively rounded upwards to define response as an improvement of ≥ 2 points (SMA type 2) and ≥ 3 points (SMA type 3). Safety data included adverse events reporting, physical examinations, vital signs, and laboratory tests.

Descriptive analyses were used to report baseline characteristics, effectiveness, and safety. Continuous variables were summarized by median and range. Categorical variables were summarized by the total number of patients and corresponding percentages in each category. In addition, spaghetti plots were produced to describe the trajectory of motor measures at baseline and thereafter, regardless of the measuring time post-index date. The generalized estimating equations model (GEE) was used to evaluate the effectiveness of risdiplam [18]. Statistical analysis was conducted with a 95% confidence interval (CI) and a value of p < 0.05. All analyses were conducted using R and Prism version 9.0 software.

Results

Baseline characteristics

A total of 12 patients were included (Fig. 1). Their baseline characteristics were summarized (Table 1). There were 4 cases of Type 1, 7 cases of Type 2, and 1 case of Type 3 SMA according to the best milestone achieved and the age of onset. Three patients required feeding support, two patients had scoliosis and one of them was accompanied with hip dislocation. The interval between treatment initiation and last visit was 13.3 (8.0-18.8) months.

Fig. 1.

Inclusion flowchart

Table 1.

Characteristics of the patients at baseline

Baseline Characteristics	Risdiplam (n = 12)
Gender
Male	2 (16.7%)
Female	10 (83.3%)
SMN2 copy number
2	4 (33.3%)
3	8 (66.7%)
SMA Type
1	4 (33.3%)
2	7 (58.3%)
3	1 (8.3%)
Age at symptom onset (months)	7.0 (0.7–23)
Age at diagnosis (months)	13.5 (1.7–26.4)
Symptom onset to enrollment (months)	3.5 (0.4-153.7 §)
Age at enrollment (months)	14.4 (1.7-161.7 §)
Treatment duration (months)	13.3 (8.0–18.0)
CHOP-INTEND score (n = 11*)	30.0 (9–58)
HINE-2 score	6.5 (1–25)
HFMSE score (n = 6$)	7.0 (2–30)
Maximal milestone achieved
Ability to sit without support	5(41.7%)
Ability to stand with support / independently	2 (16.6%)
Ability to walk independently	1 (8.3%)
Need for respiratory support	2 (16.7%)
Need for feeding support	3 (25.0%)
Scoliosis	2 (16.7%)
Hip dislocation (n = 6ǂ)	1 (16.7%)

Data are n (%), or median (range)

SMN2 Survival Motor Neuron 2, CHOP-INTEND Children’s Hospital of Philadelphia Infant Test of Neuromuscular Disorders, HINE-2 Hammersmith Infant Neurological Exam-Part 2, HFMSE Hammersmith Functional Motor Scale Expanded

^§Diagnosed in 2012, but no DMT available

* CHOP-INTEND was not performed in the type 3 patient

$ indicated patients who were able to cooperate with HFMSE evaluation

ǂ Indicated patients who received a pelvic X-ray examination before baseline among included patients

Motor function improvement

In this study, 12 patients were followed up ranging from 8 to 18 months after treatment initiation. Among them, 6 patients were followed for over 15 months, 4 patients for 10 to 12 months, and 2 patients for 8 months. A total of 11 completed the CHOP-INTEND assessment, and the spaghetti plot illustrated CHOP-INTEND score changes for each patient (indicated by different colors) (Fig. 2A), with 9 patients showing improvements. GEE model indicated that for each additional month of risdiplam treatment, the CHOP-INTEND score was expected to increase by 1.73 points. Similarly, the HINE-2 and HFMSE were assessed during follow-up in 12 and 9 patients, respectively (Fig. 2B-C). For each additional month of risdiplam treatment, the HINE-2 score was expected to increase by 0.87 points, and the HFMSE score by 1.01 points.

Fig. 2.

Motor function improvement after Risdiplam treatment. A Spaghetti plot of CHOP-INTEND motor function test for individual patients over days of treatment (n = 11). B Spaghetti plot of HINE-2 motor milestones for individual patients over days of treatment (n = 12). C Spaghetti plot of HFMSE motor function test for individual patients over days of treatment (n = 9). Each patient is indicated by a different color

Incremental improvements in the WHO motor milestones were observed in 9/11 patients (except for 1 patient who had achieved all milestones before treatment) at the last visit (Table 2) and 4/9 patients did not reach any milestones before treatment. Following treatment, 1 patient achieved 3 new motor milestones (sitting without support, standing with assistance and walking with assistance), 1 achieved 2 new milestones (sitting without support and standing with assistance), and 2 achieved a new milestone (sitting without support). The other 5 patients, 2 had already lost all milestones at baseline but achieved a new one (sitting without support) after treatment. Two patients reached sitting without support milestone at baseline and achieved the crawling milestone after treatment, with 1 also reaching standing with assistance milestone. One patient reached sitting without support and standing with assistance milestones at baseline and achieved crawling milestone after treatment.

Table 2.

Proportion of children achieving new milestones after treatment start

WHO motor milestone	Proportion of children achieved new milestone after treatment start	Age of achievement for motor milestones (months)	Time from enrollment to new milestone achievement (months)
Sitting without support	5/11	12.1 (10.1–45.9)	9.8 (8.2–10.4)
Standing with assistance	2/11	14.0 (13.2–14.9)	12.2 (11.2–13.3)
Hands & knee crawling	2/11	29.2 (21.7–36.6)	8.7 (7.1–10.2)
Walking with assistance	1/11	32.0	10.1
Standing alone	2/11	18.8 (17.9–19.8)	10.7 (5.1–16.2)

Data is the number of patients, or median (range)

WHO New World Health Organization

Sub-group analysis

According to the definition of responders and non-responders, 11 patients were divided into ‘responders’ (CHOP-INTEND score increase at least 4 points) and ‘non-responders’ (< 4 points) groups, and the characteristics of each group were analyzed (Table 3). The results showed that two patients in the ‘non-responders’ group were both female and with 3 copies of SMN2. Compared to the ‘responders’ group, both patients were older when diagnosed and treated, and the time from disease onset to treatment was significantly longer. Based on these differences, we conducted correlation tests between CHOP-INTEND improvements and various baseline parameters for all patients (Fig. 3A-F). The results showed that CHOP-INTEND changes were significantly negatively correlated with onset age (R = 0.8527, p = 0.0015), diagnosis age (R = 0.8118, p = 0.0011), treatment initiation age (R = 0.8795, p = 0.0008), disease duration (R = 0.6400, p = 0.0463), time from onset to treatment (R = 0.6573, p = 0.0389), and baseline CHOP-INTEND scores (R = 0.6676, p = 0.0349). In other words, the result revealed greater improvements in younger and receiving treatment earlier children. Additionally, the worse the baseline motor function, the more significant the improvement. Furthermore, we also compared the improvement in CHOP-INTEND scores among patients with different SMN2 copies (Fig. 3G). The results showed that patients with 2 SMN2 copies showed significantly greater improvement in CHOP-INTEND scores compared to those with 3 copies (p = 0.0202, unpaired t test with Welch’s correction).

Table 3.

Characteristics of the CHOP-INTEND responders and non-responders

Characteristics	Responders(n = 9)	Non-responders(n = 2)	R ǂ
Female	8 (88.9%)	2 (100%)	-
SMN2 copy number
2	4 (44.4%)	0 (0%)	-
3	5(55.6%)	2 (100%)	-
SMA Type
1	4 (44.4%)	0 (0%)	-
2	5 (55.6%)	2 (100%)	-
Age at symptom onset (months)	2 (0.7–12)	8 (8–8)	0.8527 **
Age at diagnosis (months)	12.7 (1.6–15.6)	14.7 (11.6–17.7)	0.8118 **
Age at first treatment (months)	13.4 (1.7–21.7)	99.2 (36.6-161.7 *)	0.8795 ***
Symptom onset to enrollment (months)	1.4 (0.3–12.3)	91.2 (28.6-153.7 *)	0.6573 *
Disease duration (months)	24.3 (12.9–46.8)	113.6 (65.6-161.7 *)	0.6400 *
CHOP-INTEND score	27 (9–58)	29 (28–30)	0.6676 *

Data are n (%), or median (range)

ǂ R value of the correlation between ΔCHOP-INTEND score and these characteristics

*p < 0.05, **p < 0.01, ***p < 0.001

Fig. 3.

Change from Baseline to the End-of-trial in Individual CHOP-INTEND Scores According to Different Baseline Characteristics. A-F Shown is the change from baseline to the end-of-trial in each patient’s CHOP-INTEND score according to age at symptom onset (Panel A), age at diagnosis (Panel B), age at first treatment (Panel C), time from symptom onset to first treatment (Panel D), disease duration (Panel E) and CHOP-INTEND score at baseline (Panel F). Disease duration is a child’s age at screening minus the age at symptom onset. The analyses included children in the intention-to-treat population who did not have missing data for CHOP-INTEND assessment, 1 patient who diagnosed SMA in 2010 but had no available treatment was excluded. G Shown is the change from baseline to the end-of-trial in each patient’s CHOP-INTEND score with different SMN2 copies

Based on the response threshold of the HINE-2 score, 12 patients were divided into ‘responders’ and ‘non-responders’ groups, and the characteristics of each group were analyzed (Table 4). The results indicated that only one patient born in 2010 did not improve, with disease onset at 8 months of age, and was diagnosed at 17.7 months, accompanied by severe scoliosis and hip dislocation. This patient often experienced chest tightness and required respiratory support and had a poor appetite. At the age of 13, she began taking risdiplam, 12 years after her diagnosis. After 1–2 months treatment, improvements in her appetite, respiratory function, and swallowing ability were reported from this patient and caregivers.

Table 4.

Characteristics of the HINE-2 responders and non-responders

Characteristics	Responders (n = 11)	Non-responders(n = 1)
Age at symptom onset (months)	6 (0.7–23)	8
Age at diagnosis (months)	12.7 (1.6–26.4)	17.7
Symptom onset to enrollment (months)	3.5 (0.3–28.6)	153.7 §
Disease duration (months)	24.3 (12.9–65.6)	161.7 §
Need for respiratory support	1/9 (11.1%)	1 (100%)
Need for feeding support	3/10 (33.3%)	0 (0%)
Scoliosis	1/4 (25.0%)	1 (100%)
Hip dislocation	0/5 (0%)	1 (100%)

^§Diagnosis in 2012, but no DMT available

Table 5 presents individual patient characteristics and HFMSE scores, categorizing each patient as a responder or non-responder according to pre-defined MCID thresholds. The cohort included 7 patients with SMA type 2 and 1 patient with SMA type 3. Based on the MCID criteria, 6 patients (75%) were classified as HFMSE responders, while 2 patients (25%) were classified as non-responders. Notably, these two HFMSE non-responders are the same patients who were non-responders on the CHOP-INTEND scale.

Table 5.

Characteristics of HFMSE responders and non-responders

Patient No.	SMA Type	Onset age(month)	Treatment initiation age(month)	HFMSE score after treatment	HFMSE Response
1	3	23	26.5	Baseline: 30/ 4-month: 30/ 10-month: 38	Responders
2	2	11	21.7	Baseline: 13/ 3-month: 14/ 10-month: 23
3	2	6	16.1	Baseline: 4/ 3-month: 6/ 9-month: 16
4	2	11	14.5	Baseline: 17/ 5-month: 28/ 9.5-month: 34/ 15-month: 36
5	2	2	14.3	Baseline: NAǂ/ 7-month: 22/ 13-month: 29
6	2	12	13.4	Baseline: NAǂ/ 6.5-month: 4/ 13-month: 13
7	2	8	161.7	Baseline: 2/ 6-month: 2	Non-responders
8	2	8	36.6	Baseline: 3/ 6-month: 3

ǂNA: Missing data due to the patient being too young or unable to cooperate with the assessment

Safety

During follow-up, 11 patients experienced at least one adverse event(AE). Two patients experienced serious adverse events (SAEs), both pneumonia (Table 6). Mild AEs comprised 18 upper respiratory tract infections, and 1 instance each of skin discoloration, acute gastroenteritis, constipation, and abnormal liver function. Acute gastroenteritis occurred in 1 patient one year after treatment, presenting with nausea and diarrhea; it was suspected to be related to viral infection and resolved following symptomatic treatment. Constipation persisted for 2 months and improved by symptomatic treatment. One patient developed abnormal liver function tests (ALT 103U/L, AST 134 U/L) 10 days post-treatment. Levels decreased to ALT 76.2 U/L and AST 84.5 U/L by day 35 and returned to normal by month 5. Some AEs, particularly respiratory infections (common in infants with SMA), were assessed as unrelated or unlikely related to risdiplam. No rash, headache, deaths, or treatment discontinuations were reported.

Table 6.

Summary of adverse events

Events	Risdiplam (n = 12)
Total number of adverse events	24
Patients with ≥ 1 adverse event	11 (91.7%)
Patients with ≥ 1 severe adverse event	2 (16.7%)
Adverse events with the highest incidence
Upper respiratory tract infection	18 (75.0%)
Pneumonia	2 (8.3%)
Skin discoloration	1 (4.2%)
Acute gastroenteritis	1 (4.2%)
Constipation	1 (4.2%)
ALT/AST increase	1 (4.2%)
Any adverse event leading to treatment discontinuation	0

Discussion

Risdiplam’s oral formulation facilitates convenient home-based treatment; however, long-term follow-up data from large cohorts remain limited. Herein, in a study with a median follow-up duration of 13.3 months, risdiplam demonstrated favorable tolerability and efficacy. Critically, the overall treatment effect was characterized by stabilization or improvement in motor function. Clinically meaningful improvements in CHOP-INTEND and HINE-2 scores were observed in 81.8% (9/11) and 91.7% (11/12) of patients, respectively. Importantly, the remaining two patients, who did not achieve improvement on CHOP-INTEND and HFMSE scores, exhibited no loss of motor function during the observation period. This maintenance of function is a clinically significant outcome, as it stands in stark contrast to the progressive motor decline consistently documented in the natural history of untreated SMA [19, 20]. Indeed, spaghetti plot analysis from this study revealed no functional decline across any of the three scales evaluated. These findings collectively demonstrate that risdiplam treatment effectively prevents the functional deterioration expected in the natural course of the disease, representing a clear therapeutic benefit even in the absence of measurable motor score gains for some individuals.

Of note, the oldest patient (13 years of age) showed no motor function improvement after 6 months of treatment but experienced rapid multisystemic benefits, including alleviated respiratory distress, improved swallowing function, increased appetite, and weight gain. This underscores risdiplam’s potential to alleviate non-motor manifestations even in advanced SMA, further supporting the notion that therapeutic benefit, particularly for chronic patients with limited neuromuscular recovery potential, extends beyond traditional motor scale improvements and crucially includes the stabilization of function.

In this study, we employed the GEE model to estimate the extent of motor function improvement per unit of time with risdiplam treatment. We found that for each additional month of risdiplam treatment, the CHOP-INTEND, HINE-2 and HFMSE score increased by an average of 1.73, 0.87 and 1.01 points, respectively. Given that this study involves repeated measurements over a period, with a small sample size and some missing data, the GEE model was the most suitable choice.

The study further investigated how baseline characteristics influenced motor function improvement. Consistent with prior evidence [15, 21], the most substantial gains occurred in younger children and those treated soon after diagnosis, reinforcing the critical window for early intervention. Notably, patients with SMA type 1 or 2 SMN2 copies achieved significantly greater CHOP-INTEND score improvements, likely due to their earlier symptom onset (typically ≤ 6 months) and lower baseline function—a population for whom risdiplam’s therapeutic ceiling appears highest. This observed trend aligns with the biological collinearity between SMA subtypes, SMN2 copy number, and disease severity: patients with earlier-onset, low-functioning disease (e.g., Type 1) inherently possess greater potential for measurable improvement. However, the CHOP-INTEND scale’s ceiling effect [11] may underestimate gains in milder subtypes (Type 2/3), necessitating complementary endpoints for comprehensive assessment. Remarkably, even in one non-responder by motor metrics, risdiplam improved respiratory/feeding outcomes, underscoring its multidimensional benefits. The ongoing expansion of newborn screening programs in China [22] is thus pivotal to enable timely therapy initiation and maximize outcomes. Future studies should integrate individualized functional assessments to capture subtler but clinically meaningful improvements beyond standardized scales.

In this study, the definition of an HFMSE responder for patients with Type 2 and 3 SMA was established based on the latest evidence regarding the MCID [17]. This international collaborative study determined the MCID for HFMSE change in treatment-naïve patients with Type 2 and 3 SMA. We adopted these evidence-based MCID values as the new response threshold, replacing the HFMSE improvement of ≥ 3 points commonly used in previous studies [23]. This was done to more precisely identify patients achieving clinically meaningful functional improvements using this criterion.

The most commonly observed adverse events during follow-up were upper respiratory tract infections, which were likely attributable to the underlying disease pathology. Two patients experienced severe pneumonia (one episode each). Among potentially treatment-related adverse events, one case exhibited skin yellowing, a phenomenon also reported in a prior Chinese study [24]. Notably, this manifestation has not been documented in non-Asian populations, suggesting potential racial variations in drug response. Importantly, the child with skin discoloration remained asymptomatic otherwise. One patient exhibited mild, transient elevations in ALT/AST levels which completely normalized within five months. This phenomenon was also reported previously by a study based on real-world data obtained from the US Food and Drug Administration Adverse Event Reporting System database [25]. Nevertheless, this cannot conclusively confirm that the elevation was directly caused by the drug risdiplam itself. Previous studies have reported that SMA patients may develop fatty liver disease, which can also lead to increased liver enzymes [26]. Therefore, continuous monitoring is necessary during clinical treatment to ensure the long-term health of SMA patients. No fatalities or treatment discontinuation occurred, further supporting the favorable safety profile of risdiplam in this cohort.

Limitations: As a retrospective study, it is prone to issue with incomplete data. This study enrolled 12 children diagnosed with SMA. There were variations in the frequency and time points of follow-up, and hematological examination data were incomplete. To address these limitations, it is imperative to establish detailed and feasible follow-up systems or multi-centers monitoring programs. These initiatives aim to enhance follow-up rate and enlarge the sample size, thereby enabling the acquisition of more comprehensive real-world data regarding risdiplam treatment in Chinese SMA patients.

Conclusion

In summary, risdiplam treatment in Chinese pediatric patients with various SMA subtypes led to stabilization or improvement in motor function, with a substantial proportion of patients achieving new motor milestones and others maintaining function, thereby preventing the progressive decline characteristic of the natural history of SMA. The therapy maintained a favorable safety profile. Notably, motor function gains showed a robust inverse correlation with age of onset, age at diagnosis/treatment initiation, and pre-treatment disease duration. These results strongly support the clinical benefits of early intervention, with particularly pronounced therapeutic effects observed in motor function outcomes. Importantly, for patients with longer disease duration or later treatment initiation, risdiplam effectively stabilized function, representing a critical therapeutic benefit in halting disease progression.

Authors’ contributions

XZ prepared and drafted this manuscript. MW helped with data collection. HL and CH helped reviewing and analysis of the data. SZ and YW helped revise the manuscript. WL designed the study, revised the manuscript and approved the submission. All authors agreed to the submission.

Funding

This work was supported by the National Key R&D Program of China(No.2024YFC3406700, 2024YFC3406701).

Data availability

Data can be accessed and reviewed upon reasonable request to the corresponding author.

Declarations

Ethics approval and consent to participate

This study was approved by the Ethical Committee of the Children’s Hospital of Fudan University (No: (2021)307). The study was conducted in accordance with the principles of the Declaration of Helsinki. Written informed consent to participate in this study was provided by the participants’ legal guardian. No potentially identifiable human images or data is presented in this study.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.