JEWELFISH: 24-month results from an open-label study in non-treatment-naïve patients with SMA receiving treatment with risdiplam

The study was initiated after institutional review board approvals of the participating study centers, in accordance with the principles of the Declaration of Helsinki and in full conformance with Good Clinical Practice guidelines. The names of these independent ethics committees/institutional review boards have been previously published [17]. An independent Data Monitoring Committee monitored the risdiplam studies on an ongoing basis. The study was registered with ClinicalTrials.gov (NCT03032172). Participants or their legally authorized representatives completed a statement of informed consent. When possible, assent was collected from participants < 18 years of age. The study protocol and statistical analysis plan are available as Online Resource 1 and Online Resource 2, respectively.

JEWELFISH is a multicenter, open-label study of daily risdiplam that includes patients with SMA previously enrolled in the MOONFISH study (NCT02240355) [18], or those treated with nusinersen, onasemnogene abeparvovec, or olesoxime [19]. Patients previously treated with nusinersen were included if they had received ≥ 4 doses of nusinersen, provided that the last dose was received ≥ 90 days prior to screening. Patients previously treated with onasemnogene abeparvovec were included provided that the time of treatment was ≥ 12 months prior to screening. Patients previously treated with olesoxime were included provided that the last dose was received ≤ 18 months and ≥ 90 days prior to screening. Patients aged 6 months to 60 years of age at screening were eligible if they had a confirmed diagnosis of 5q-SMA, including genetic confirmation of a biallelic mutation (homozygous deletion or heterozygosity predictive of loss of function of the SMN1 gene) and clinical history, signs, or symptoms attributable to SMA. Patients were excluded if they met any of the following criteria: participation in any investigational drug or device study with the exception of studies of olesoxime, nusinersen, or onasemnogene abeparvovec ≥ 90 days prior to screening; any history of gene or cell therapy with the exception of onasemnogene abeparvovec; initiation of an oral salbutamol or oral β2-adrenergic agonist within 6 weeks prior to enrollment; any prior use of FMO1 or FMO3 inhibitor or inducer, or OCT-2 and MATE substrates within 2 weeks prior to risdiplam treatment; any prior use of any agents anticipated to increase or decrease muscle strength 90 days prior to enrollment or medications known or suspected of causing retinal toxicity within 1 year prior to enrollment. Patients were excluded if they had a recent history (< 1 year) of ophthalmologic diseases that would interfere with study assessments. Patients aged < 2 years were excluded if they had been hospitalized for a pulmonary event within 2 months prior to screening and pulmonary function had not fully recovered. There were no exclusion criteria based on SMA type, SMN2 copy number, baseline motor function, comorbidities, or respiratory/feeding support. Patients > 2 years of age were considered ambulant if they had the ability to walk unassisted for ≥ 10 m. Full inclusion and exclusion criteria for JEWELFISH have been published [17]. Based upon practical considerations the study size was set at 180 patients [17]. Assuming the underlying adverse event (AE) rate is 1.4%, a study of 180 patients exposed to risdiplam would have a 92% chance of detecting an AE in ≥ 1 patient. Patients received risdiplam orally once daily at the approved dosing regimen based on body weight and age. For patients aged 2–60 years, the dose is 5 mg for patients with a body weight of ≥ 20 kg, and 0.25 mg/kg for a body weight < 20 kg. For patients aged 6 months to < 2 years, the dose is 0.2 mg/kg [17].

Patients were enrolled across 24 different centers in Belgium, France, Germany, Italy, the Netherlands, Poland, Switzerland, the UK, and the USA. After 24 months of treatment, participants were invited to participate in the open-label extension phase to continue treatment for up to an additional 3 years.

Key exploratory objectives included evaluations of the efficacy of treatment with risdiplam in terms of motor and respiratory function. Functional assessments were dependent on the age of the patients. Patients aged 2–60 years were assessed using the 32-item Motor Function Measure (MFM32), the Hammersmith Functional Motor Scale—Expanded (HFMSE), and the Revised Upper Limb Module (RULM). Following the start of the study, the 6-min walk test (6MWT) was added as an exploratory endpoint for ambulant patients 6–60 years of age and therefore not all patients have baseline measurements. Patients aged ≤ 2 years were assessed using the Bayley Scales of Infant and Toddler Development, third edition (BSID-III), and achievement of motor milestones was assessed through the Hammersmith Infant Neurological Examination, Module 2 (HINE-2).

Respiratory function was assessed in patients aged 6–60 years by spirometry tests and patients aged 2–60 years performed the sniff nasal inspiratory pressure test. Patient-reported independence in activities of daily living among patients aged 12–60 years and caregiver-reported independence for patients aged 2–60 years were assessed through the SMA Independence Scale–Upper Limb Module (SMAIS–ULM). This assessment was implemented after the study started and, therefore, not all patients have a baseline assessment.

Safety data were reported for all patients and by previous treatment. Additional methods and exploratory efficacy data are reported in Online Resource 3.

A post hoc safety analysis was conducted to compare the rates of AEs and serious AEs (SAEs) over time in patients who received risdiplam in the JEWELFISH study with patients who were treatment-naïve prior to receiving risdiplam in the SUNFISH (NCT02908685) [20] and FIREFISH (NCT02913482) [21] studies.

All patients who received ≥ 1 dose of risdiplam were included in the safety analysis population. The intent-to-treat (ITT) population was defined as all enrolled patients, regardless of whether they received risdiplam or not. The ITT population was used for all exploratory efficacy analyses.

In this analysis, patients had received risdiplam for ≥ 24 months by the clinical cut-off date (CCOD) of January 31, 2022. Between March 2017 and January 2020, a total of 182 patients were screened for eligibility with 174 participants meeting the recruitment criteria. Of the 174 patients enrolled in the study, one patient previously treated with olesoxime withdrew from the study at baseline prior to receiving risdiplam due to poor venous access as assessed by their physician. A total of 173 patients received risdiplam.

Patients were grouped according to their previous treatment types: 71 patients were previously treated with olesoxime in the OLEOS study (NCT02628743); 13 patients had previously been enrolled in the MOONFISH study, 10 of whom received RG7800 (RO6885237) and three who received placebo; 76 patients were previously treated with nusinersen, three of whom had also previously received olesoxime; 14 patients were previously treated with onasemnogene abeparvovec. One patient in the onasemnogene abeparvovec group had also received subsequent treatment with nusinersen. A total of 153 patients (87.9%) completed 24 months of treatment. Fifteen patients withdrew prior to the completion of 24 months of treatment, five patients did not attend the week 104 visit as of the CCOD and therefore are still represented as being in the open-label treatment period and one patient never received treatment (Fig. 1).

The patient population was broad and heterogeneous, covering a wide range of ages and baseline motor functions, and included patients with a high degree of motor impairment and other SMA comorbidities.

The median age of the JEWELFISH population at enrollment was 14 years (range 1–60); the median age at enrollment for patients who previously received nusinersen or onasemnogene abeparvovec was 12.0 years (range 1–60 and 2.0 years (range 1–5), respectively. The majority of patients had three SMN2 copies (78%). The median duration from symptom onset to first risdiplam administration was 134.5 months for patients who previously received nusinersen and 29 months for patients who previously received onasemnogene abeparvovec. The median time to first risdiplam administration from the last nusinersen dose was 4.6 months (range 3.5–19.3) and from treatment with onasemnogene abeparvovec was 16.8 months (range 13.3–21.4). In the subgroup of patients aged 2–60 years, 83% had scoliosis and 48% had received scoliosis surgery. Additionally, 63% had a baseline total score of <10 on the HFMSE. Detailed baseline characteristics can be found in Table 1. Fifteen patients with Type 3 SMA were ambulatory and had a median age at enrollment of 15 years (range 5–46).

The median duration of exposure to risdiplam was 26.8 months (range 0.9–59) and the median dose intensity (i.e., the number of doses received divided by the expected number of doses) of risdiplam in all patients was 99.7% (range 24.7–100%). Safety results in all patients enrolled in JEWELFISH (n = 174) are summarized in Table 2. Overall, 96% of patients reported ≥ 1 AE, and 20% of patients reported ≥ 1 SAE. In patients previously treated with nusinersen, 96% reported ≥ 1 AE and 21% ≥ 1 SAE. All patients previously treated with onasemnogene abeparvovec experienced ≥ 1 AE, whereas 29% had ≥ 1 SAE. There was one SAE (Grade 2 supraventricular tachycardia) that was considered related to risdiplam treatment in a patient previously treated with olesoxime. There were no treatment-related SAEs in any patients who were previously enrolled in the MOONFISH study or who had previously received treatment with nusinersen or onasemnogene abeparvovec.

The proportion of patients with Grade 3–5 AEs in the overall safety population was 21%. When separated by patients previously treated with nusinersen or onasemnogene abeparvovec, the proportion was 25% and 21%, respectively. Six Grade 4 AEs occurred in three patients, which were unrelated to risdiplam and resolved without dose modification, and there were no patients with Grade 5 AEs.

Across all patients, the most common AEs were pyrexia (24%), upper respiratory tract infection (URTI, 21%), and headache (18%). Patients previously treated with nusinersen experienced pyrexia (30%), diarrhea, headache, and URTI (22% for each), whereas patients previously treated with onasemnogene abeparvovec experienced pyrexia (43%), nasopharyngitis (36%), and URTI (29%) as common AEs. Pneumonia (3%) was the most reported SAE for all JEWELFISH patients, including those previously treated with nusinersen (4%) or onasemnogene abeparvovec (7%).

When adjusted for patient-years (PY) at risk, the overall AE and SAE rates for all patients during the total treatment period were 334.66 events per 100 PY (95% confidence interval [CI] 316.91–353.15) and 19.14 events per 100 PY (95% CI 15.08–23.95), respectively. The rate of AEs and SAEs decreased over the duration of the JEWELFISH study (Fig. 2). During the first 6 months of treatment, the AE rate was 653.86 events per 100 PY (95% CI 600.77–710.38; total PY at risk: 85.5 years) and remained stable for the following three 6-month periods. Over the period of 0–12 months of treatment, the AE rate was 475.51 events per 100 PY (95% CI 443.17–509.59; total PY at risk: 168.7 years) and the SAE rate was 25.50 events per PY (95% CI 18.45–34.34). Comparing the 0–12 months treatment period with the 12–24 months treatment period, a 50% decrease in the rate of AEs was observed, with 225.65 AEs per 100 PY (95% CI 203.04–250.10; total PY at risk: 160.9 years) and 9.95 SAEs per 100 PY (95% CI 5.69–16.15).

The decline in the overall rate of AEs was reflected in decreases in the rates of the most common AEs. Comparing the rates of the common AEs per 100 PY between the first and fourth 6-month treatment periods (0 to ≤ 6 months vs. > 18 to ≤ 24 months) showed declines in pyrexia (31.58 to 13.93), URTI (37.43 to 13.93), and headache (44.45 to 7.60); the rate of the most common SAE of pneumonia also decreased (4.68 to 1.27). There were no treatment-related safety findings that led to a withdrawal.

A review of all available safety laboratory results, vital signs, electrocardiograms, and ophthalmologic assessments did not show any clinically significant adverse findings compared with baseline. As of the CCOD, there have been no safety findings in patients reflective of potential risks previously identified from non-clinical toxicology studies (effects on epithelial tissues, retinal toxicity, or hematologic effects).

JEWELFISH was an open-label study; there was no control group and all efficacy endpoints were exploratory. Therefore, caution should be taken in making conclusions from the efficacy data.

Potentially, the broad inclusion criteria of JEWELFISH, which enrolled patients who had previously received other SMA treatments, may have been biased towards a patient population that had not optimally responded to previous treatment and/or who subjectively felt they needed more treatment efficacy.

Furthermore, caution should be taken when interpreting the HFMSE score in weak patients due to floor effects [32], and for the JEWELFISH population, other motor function scores such as the MFM32 or RULM may be more informative. Caution should be also taken with the interpretation of the subgroup data analyses which were conducted with a small sample of patients. Also, differences in the demographics between the subgroups, in particular, those previously treated with nusinersen or onasemnogene abeparvovec, make comparisons of efficacy difficult to interpret. There is a lack of a single universal functional outcome that can sensitively measure functional changes among this broad and heterogeneous patient group, thus making the interpretation of efficacy challenging.

Some assessments may have been affected due to travel restrictions during the COVID-19 pandemic. Although these restrictions may have contributed to lower rates of AEs between months 6 and 12 of the study, when patients were required to spend more time indoors with limited in-person socialization, the exact impact cannot be determined.