INTRODUCTION

Spinal muscular atrophy (SMA) is caused by the deletions or pathogenic variants of the survival motor neuron 1 gene (SMN1) (MIM 600354), leading to motor neuron degeneration and resulting in progressive muscle weakness and atrophy. Patients affected by the most common form of SMA, SMA type 1 (SMA1), usually experience symptom onset within the first few weeks of life. A few specific treatments are currently available for SMA, but early, or preferably presymptomatic, initiation of treatment seems to be critical.1,2,3,4,5 We initiated a large-scale population newborn screening program (NBS) for SMA in Taiwan in 2014, and have successfully identified newborns with SMN1 deletion.6 In this report, we assess the impact of the early detection of SMA through this program, and the association between compound motor action potential (CMAP) amplitude and symptom onset and treatment outcomes.

MATERIALS AND METHODS

The SMA NBS has been described previously, with modification of the primers (sequence upon request) since 2018 to avoid the previously described false positives.6 SMN2 copies were confirmed both by a droplet digital polymerase chain reaction (ddPCR) assay using the original screening dried blood spot (DBS) sample and also a multiplex ligation-dependent probe amplification (MLPA) assay using DNA extracted from whole-blood samples of the same patients, as described in our previous publication.6 SMN2 c.859G>C modifier was also checked in genetically diagnosed SMA patients.7 Patients had access to treatment through clinical trials or expanded use programs (NCT02386553, NCT02865109, NCT03837184, NCT03505099, NCT04042025). Infants with two copies of SMN2 were treated immediately after the diagnosis, while those with three copies were followed monthly for symptom onset before starting treatment, and those with four copies were evaluated in outpatient clinics or followed by phone every 4 to 6 months. The evaluations included neurological examinations, developmental milestones, and CMAP, of which the maximal CMAP amplitude of the ulnar nerve was recorded by experienced pediatric neurologists. A minimum of four G1 placements over the hypothenar eminence were used to ensure the measurement of maximum CMAP, as described in previous studies.8 The parents were taught to identify the symptoms of SMA. The phenotype of SMA was classified depending on the age of symptom onset.9 Symptomatic patients also received multidisciplinary standard of care at the specific neuromuscular joint clinic in our hospital.

Ethics statement

This study was approved by the Institutional Review Board of National Taiwan University Hospital (NTUH) (201308058RIN and 201702057RINB), and the parents of all participants signed informed consent forms.

RESULTS

From November 2014 to December 2019, 364,000 newborns were screened for SMA. Among them, 21 were confirmed to have SMN1 dysfunction, including 20 with an SMN1 homozygous deletion (95%) and one who had a false negative screening test with an SMN1 heterozygous deletion (5%). This patient presented with hypotonia since 5 months of age. When she was diagnosed at 12 months of age, she was profoundly hypotonic, could not sit without support, and had difficulty swallowing. A physical examination revealed tongue fasciculation, an absence of deep tendon reflexes. One copy of her SMN1 gene was deleted. She also harbored a c.373G>A (V125M) variant, a novel variant predicting to be damaging, on her SMN gene. The incidence of SMA in our cohort was ~1 in 17,000 (95% CI: 11 350–26,530) newborns. SMN2 copy-number analysis showed two SMN2 copies in nine infants (43%), three copies in six infants (29%), and four copies in six infants (29%) (Table 1). None of these 21 patients had the SMN2 c.859G>C modifier.

Table 1 Clinical and electrophysiological findings, treatment and follow-up of 21 patients diagnosed by newborn screening between November 2014 and May 2020.
Full size table

A total of 17 infants were followed, of whom 12 (70%) developed SMA1 (Figure S1). Nine (75%) of these SMA1 infants (five with two SMN2 copies and four with three SMN2 copies) were treated either with nusinersen or AVXS-101, and the other three (all with two SMN2 copies) died without treatment (Table 1). All eight infants with two SMN2 copies followed became symptomatic before the age of 1 month, while five (63%) showed symptoms at birth. Some of the subtle symptoms at birth improved transiently, followed by rapid deterioration. Five of the six infants with three SMN2 copies became symptomatic before the age of 1 year, of whom four developed SMA1 (80%) defined by age of onset, and one developed SMA type 2a. The other infant (patient 11) was treated presymptomatically, and so the clinical type was undefined. Only one of the four followed infants with four SMN2 copies became symptomatic at the age of 37 months. The age of symptom onset was significantly different among the infants with different SMN2 copies (p < 0.001, Figure S2).

CMAP amplitude values of the ulnar nerve of 12 infants were available, including six who were subsequently treated with nusinersen, four who were not treated, and two before entering an AXVS-101 trial. The initial CMAP amplitudes were much lower in the infants with two SMN2 copies (1.39 ± 0.79 mV) than in those with three copies (3.70 ± 0.59 mV) (Fig. 1). CMAP amplitude decreased rapidly in all SMA1 infants. In the infants with three SMN2 copies (patients 13–15), the ulnar CMAP amplitudes dropped significantly when the muscle power of their upper extremities was still normal. In the infants with type 2 and 3, a decline in CMAP amplitude was also correlated with symptom onset, although the speed of decline was slower than in those with SMA1.

Fig. 1: Changes in compound motor action potential (CMAP) amplitude (amp) of the ulnar nerve against age in 12 infants with spinal muscular atrophy (SMA) identified through a newborn screening program.
figure 1

The infants are grouped into those with two (red), three (blue), and four (green) copies of SMN2. Changes in the pretreatment periods are connected by dashed lines, and changes after nusinersen treatment are connected by solid lines. The upper right insert is an enlargement of the infants with two copies of SMN2 at age 0–6 months, to better show the rapid decline in CMAP amplitude in the first month of age.

Full size image

CMAP data from six infants who were treated with nusinersen were available. The pretreatment CMAP amplitudes were 0.89 ± 0.76 mV and 2.36 ± 1.71 mV in the patients with two and three SMN2 copies, respectively, compared with 1.33 ± 0.22 mV and 3.69 ± 2.84 mV 6 months after treatment. In those with a pretreatment CMAP amplitude ≥2 mV (patients 14 and 15), CMAP increased significantly after treatment, accompanied by normal motor development. In those with a pretreatment CMAP amplitude <2 mV (patients 3, 4, and 13), the increase in CMAP after treatment was small and slow, although still accompanied by functional improvements. The only outlier was patient 7 (with two SMN2 copies) whose pretreatment CMAP amplitude was <1 mV and who showed significant improvements in CMAP after treatment. He had mild motor developmental delay, while he could sit independently at the age of 9 months and walk holding on at the age of 15 months.

DISCUSSION

The incidence of SMA in our population was around 1 in 17,000, which is lower than expected because of carrier screening and prenatal diagnosis.10,11 Very recently, a study of population-based newborn screening showed that the SMA incidence in New York State, due to the uptake of carrier screening, is also low with 1 in 28,137.12 In this study we prospectively observed the onset of symptoms in SMA patients identified by newborn screening. The high percentage (70%) of SMA1 in our cohort suggests that compared with historical experience with type 1, fewer patients died early and that the symptoms of type 1c were probably detected earlier than others.13 Studies of SMA newborn screening from Australian and German groups also revealed that all patients with two SMN2 copies had early symptom onset at the first visit or before the age of 3 months if presymptomatic therapies were not applied.14,15 All patients with two or three SMN2 copies in our cohort, who have not received treatment or before treatment, developed type 1 (n = 13) or 2a (n = 1); thus, our results further support that newborn babies with two or three SMN2 copies should be treated immediately.16 With regard to newborns with four copies, immediate presymptomatic treatment has not been used in Taiwan or in Australia. Therefore, those cases with four copies were not reported as screening positives in one Australia research pilot study.14

Definitions of the clinical types of SMA may need to be revised when prospective symptom monitoring and disease-modifying treatments both become common.17 For example, the symptom onset time described in our prospective study was defined by neurologists, which should be earlier than that detected by parents. In addition, because of the effective treatment, we could only use the disease onset time for classification but not the maximal motor function achieved. For example, our patient 14 was type 1c but can now walk, suggesting that interventions may modify the phenotypes of patients. Moreover, interventions may lead to a new phenotype instead of changing from type 1 to 2 or 3, which should further remind clinicians to carefully describe the phenotypes in treated patients. Taken together, these findings suggest that the scenario of complication management and standard of care in SMA will need to be changed in the future.18

In this report we provide evidence that CMAP amplitude can serve as a sensitive method to earlier detect the onset of disease from presymptomatic to symptomatic. Our results revealed that the CMAP values were lower and decreased faster in the infants with two SMN2 copies than in those with three or four copies, similar to the findings of Swoboda et al. with regard to the natural history of denervation in SMA.8 We also demonstrated the rapid decrease of CMAP amplitude in the very beginning of clinical or subclinical symptoms and signs in our type 1 patients, both two and three copies, suggesting the use of rapid decline in CMAP amplitude to earlier detect infantile-onset SMA for prompt treatment.17

We also correlated CMAP to treatment. The NURTURE study demonstrated a trend regarding higher baseline CMAP values, post-treatment D64 CMAP values, and better prognosis.3 In our patients with two copies, although patient 7 had even lower pretreatment CMAP values, he still demonstrated a late increase in CMAP values 9 months after treatment. Interestingly, patients with three copies of SMN2 (patients 14 and 15) exhibited a rapid decline in CMAP upon symptom onset, but had a rapid recovery of CMAP after the initiation of treatment. It is likely that, under the current follow-up protocol, motor neurons can still be rescued even after the occurrence of symptoms. Our results further clearly indicated that pretreatment CMAP and a rapid reversal or increase in CMAP after treatment were closely related to the developmental outcomes. Therefore, CMAP appears to have great potential in clinical application and in the evaluation of therapeutics for infantile-onset SMA patients. Still, the application of CMAP in evaluating therapeutics for later-onset SMA patients with four SMN2 copies needs further investigations.

In conclusion, this report demonstrates the incidence of SMA and its types, and the importance of pretreatment CMAP amplitude as well as the change of post-treatment CMAP with regard to disease presentation and also treatment outcomes. CMAP may have great potential in clinical application and in the evaluation of therapeutics in SMA patients. However, there are several limitations to this study, including the small sample size of treated patients and a lack of detailed clinical and CMAP information of the patients with four SMN2 copies. Considering the latest treatment recommendations,19 further studies regarding the natural history and application of CMAP amplitude in treatment, especially in patients with four SMN2 copies, are needed.