Introduction

Spinal muscular atrophy (SMA) is an autosomal recessive neuromuscular disorder that affects approximately 1 in 10,000 live births1,2,3. SMA is a leading cause of infant mortality worldwide4. SMA is an autosomal recessive disorder due to biallelic mutations in the SMN1 gene in chromosome 5q13, which encodes the survival motor neuron protein5. SMA has been categorized into three primary types that manifest in childhood (types I-III) and two rarer types (antenatal type 0 and adult-onset type IV). This classification system is primarily determined by the age of onset and the best motor function achieved6. Infants with type I SMA cannot sit, with the disease manifesting at 0 to 6 months of age; those with type II disease cannot walk and are diagnosed at 6 to 18 months of age; and those with type III disease can walk but manifest symptoms after 18 months of age.

The SMN2 gene is a centromeric copy of the SMN1 gene, but differs by a C-to-T substitution in exon 7. This single nucleotide change causes exon 7 to be skipped during SMN2 pre-mRNA splicing, resulting in the production of a mostly nonfunctional SMN protein. Only 10–15% of the SMN2 transcripts produce a full-length, functional protein. In patients with SMA, an inverse correlation between SMN2 copy number and the severity of the phenotype has been established. Increasing the full-length SMN protein level by modifying the SMN2 function is the main target of SMA treatment. Nusinersen (Spinraza®; Biogen Inc, Boston, MA) achieved a significant milestone as the first drug to successfully complete randomized, sham-controlled clinical trials in early-infantile and late-infantile onset SMA, leading to its Food and Drug Administration approval in 2016. It is a splice-switching antisense oligonucleotide that facilitates the incorporation of exon-7 into the SMN2 gene transcript, thereby augmenting the quantity of biologically active SMN protein7,8,9,10. Nusinersen is a disease-modifying drug that reduces patient mortality rates, lowers the need for permanent ventilator support, and improves bulbar and motor function11.

Owing to its high molecular weight, nusinersen cannot traverse the blood–brain barrier, necessitating intrathecal injections12. Various methods have been employed for intrathecal injection, including blind lumbar puncture, ultrasound-guided injection, fluoroscopy-guided injection, and computed tomography (CT)-guided injection. In patients with intermediate- and late-onset SMA (types II–IV), scoliosis frequently develops because of neuromuscular weakness, often requiring spinal instrumentation to prevent the progression of deformities13. Intrathecal drug delivery can pose challenges in these patients and conventional lumbar puncture techniques are often insufficient to achieve intrathecal access.

CT-guided transforaminal injection has been reported to be an effective method for drug delivery in patients with severe scoliosis or in those who have undergone fusion surgery14. However, nusinersen treatment requires short-term four loading doses therapy and periodic maintenance therapy15, necessitating multiple invasive procedures. Along with radiation exposure, various side effects have been reported, including post-lumbar puncture headaches, radiculopathy, and pain at the lumbar puncture site. Patients with SMA often complain of chronic and persistent pain and fatigue; however, its exact cause remains unclear16. Some researchers argue that an imbalance in the body and increased muscle tension may be contributing factors17. Regardless of the underlying reasons, repeated procedures and secondary pain resulting from such interventions can be burdensome for patients who consistently experience pain and fatigue. Therefore, factors associated with postprocedural pain should be determined, as identifying potential avenues to mitigate such pain is crucial. Thus, we hypothesized that aspects directly related to the procedure, such as needle angle, needle insertion depth, and needle tip position, could be associated with postprocedural pain in patients.

Therefore, our study aimed to identify the factors associated with postprocedural pain following repeated CT-guided nusinersen therapy in non-ambulatory patients with SMA.

Results

Patient and procedure data

Our research involved 34 patients who underwent a total of 290 CT-guided transforaminal injections, with an equal distribution of 17 men and 17 women, and an average age of 21.8 ± 9.3 years (Tables 1 and 2). The median follow-up period for the entire cohort was 33.1 months, with an interquartile range (IQR) of 20.4 to 42.6 months. Most patients (88.2%) were diagnosed with type II SMA. The mean SMN2 copy number in all patients included in this study was 3. The baseline Expanded Hammersmith Functional Motor Scale (HFMSE), Hammersmith Infant Neurological Examination (HINE), and Children’s Hospital of Philadelphia Infant Test of Neuromuscular Disorders (CHOP-INTEND) scores were 3.7 ± 6.6, 5.7 ± 4.2, and 21.1 ± 12.4, respectively (Table 1). Of the 34 patients, 30 experienced at least mild pain, 17 with moderate to severe pain, 17 with headaches after at least one procedure, 9 with radicular pain, and 2 with nausea and fever. Of the 290 procedures conducted, 49.3% had postprocedural pain and 13.8%with moderate-to-severe pain. Specific discomforts included headache (11.0%, 32 of 290), radicular pain (6.2%, 18 of 290), nausea (0.7%, 2 of 290), and fever (1.0%, 3 of 290). The average radiation dose was 207.2 ± 120.5 mGy, with an average of 3.9 ± 2.2 CT scans performed per procedure. The mean procedural duration was 23.5 ± 9.1 min (Table 2).

Table 1 Patient demographic characteristics at baseline.
Full size table
Table 2 Details of CT-guided transforaminal injection procedure.
Full size table

Factors affecting post-procedural pain

In the multivariate analysis of postprocedural pain (Table 3), the only factor significantly associated with the presence of post-procedural pain was the pre-procedure HFMSE score (odds ratio [OR], 0.998; 95% confidence interval [CI], 0.988–0.998; p = 0.0101). In the subgroup analysis conducted on patients experiencing postprocedural pain, the needle angle with vertical (V) and horizontal (H) lines and the presence of prophylactic pain control emerged as significant variables in the multivariable analysis. The coefficients (standard error [SE]) of these variables were 0.034 (0.011) and 1.013 (0.290), respectively (p < 0.05, both).

Table 3 Factors affecting the presence and severity of post-procedural pain.
Full size table

Factors affecting radiation dose

In the multivariate analysis of radiation dose (Table 4), significant correlations were observed with injection age, CT-guided injection number, performer experience, needle length, needle angle with V and H lines, and CT scan number (all p < 0.05). The respective coefficients (SE) for these variables were 3.730 (1.147), −8.070 (1.571), −25.637 (10.113), 0.802 (0.317), 1.037 (0.397), and 34.140 (1.995), respectively.

Table 4 Factors affecting radiation dose and scan time.
Full size table

Factors affecting procedure time

In the multivariate analysis of procedure time (Table 4), significant correlations were observed between needle angles with V and H lines and CT scan number (all p < 0.05). The coefficients (SE) of these variables were 4.341 (2.190) and 120.754 (13.142), respectively.

Discussion

Recent real-world studies on nusinersen treatment in adult and older pediatric SMA patients have shown favorable efficacy and safety profiles. However, in long-term follow-up data for non-ambulatory patients, the factors contributing to discomfort from intrathecal injections remain inadequately investigated.

Our main procedural finding was the association between postprocedural pain and the planned needle angle during the procedure. This may be because using the ground or the operator’s body axis as a reference line enables quicker, more accurate procedures, reducing the need for multiple insertions or trajectory adjustments. Prophylactic pain control was considerably linked to higher pain severity, maybe because patients who experienced pain in previous procedures were more likely to receive prophylactic interventions. The only factor significantly associated with the presence of post-procedural pain was the pre-procedure HFMSE score (OR, 0.998; 95% CI, 0.988–0.998; p= 0.0101). Higher baseline motor function was related to a reduced likelihood of experiencing pain. The higher the HFMSE scores, the better muscle strength and motor function, which suggests that patients with greater physical function may have increased resilience to invasive procedures, reducing pain perception18. In addition, higher motor function may reduce anxiety and boost confidence during procedures, further lowering pain reports19.

This study highlights key factors influencing radiation dose and procedure time during CT-guided injections. In the multivariate analysis of radiation dose, several variables were significantly correlated, including injection age, CT-guided injection number, performer experience, needle length, needle angle with vertical and horizontal lines, and the number of CT scans. These factors demonstrated significant associations, with injection age and needle-related parameters playing critical roles in determining radiation exposure.

Furthermore, older patients tended to receive higher radiation doses, likely due to their larger body size. Similarly, needle length, which reflects the distance to the thecal sac, follows the same rationale. These two factors were considered more important than body mass index (BMI) in influencing radiation exposure. Additionally, performer experience was inversely correlated with radiation dose, suggesting that more experienced practitioners perform procedures more efficiently, reducing exposure and the need for additional imaging. The finding that more CT-guided injections were associated with lower radiation doses supports this, likely due to the cumulative expertise gained. Needle angle with vertical and horizontal lines was also a key predictor of radiation dose. Adjusting the needle trajectory requires additional imaging to ensure precision, thereby increasing radiation exposure. The significant correlation between the number of CT scans and radiation dose underscores the importance of precise procedural planning to minimize scans. Regarding procedure time, needle angles and the number of CT scans were associated with longer procedures, likely due to the need for more intricate adjustments. Therefore, careful planning is crucial to increase the chances of success on the first attempt, reducing procedure time and radiation exposure.

These findings emphasize the importance of modifiable procedural factors, particularly needle positioning, which impacts postprocedural pain, radiation exposure, and procedure time. Thus, to improve safety, we optimize needle placement to minimize repeated adjustments and imaging, which can considerably enhance the safety and efficiency of CT-guided injection procedures.

Our study had limitations. First, beyond clinical scores, factors, such as feeding tubes, respiratory support, and tracheostomy have a significant impact on a patient’s quality of life. Unfortunately, the limited number of patients with these factors prevented us from conducting related analyses. Second, post-lumbar puncture headache is also a significant cause of patient discomfort. It is known that the rapid removal of excessive cerebrospinal fluid (CSF) can influence the occurrence of this headache20. We analyzed the variables within our patient group to identify the factors contributing to post-lumbar puncture headache; however, no significant variables were found, likely due to the small sample size relative to the number of procedures. Third, pain was assessed in the immediate postprocedural period, and the potential for long-term pain or chronic discomfort remains unknown, warranting further investigation. Fourth, further studies are needed to determine whether these procedure-related factors apply to broader patient populations, such as those with less severe forms of the disease or other neuromuscular disorders. Finally, as this was a retrospective study, causality may not have been established, and potential confounding variables, such as patient psychological factors influencing pain perception, may not have been fully accounted for.

In conclusion, in patients with severe non-ambulatory SMA, the needle angle during the procedure was identified as a significant factor influencing post-procedural pain. Our findings may guide clinicians and spark further studies on treatment protocols to improve treatment response and post-procedural adverse events in patients with SMA requiring frequent intrathecal injections.

Methods

Ethical declarations

This retrospective single-center study was approved by the Institutional Review Board of Gangnam Severance Hospital, Yonsei University College of Medicine, Seoul, Republic of Korea, (approval number 3–2023-0339), and all the procedures adhered to the ethical standards outlined in the Declaration of Helsinki. The need for informed consent was waived because the data was fully de-identified to safeguard patient confidentiality.

Patient selection and data acquisition

Since April 2019, the Korean National Health Insurance Service (HIRA) has provided reimbursement for nusinersen therapy in patients with SMA who have a genetically confirmed diagnosis with disease onset before 3 years of age and do not require permanent respiratory support. From April 2019 to July 2023, a total of 44 patients with SMA received nusinersen therapy at our hospital. For cases where a blind lumbar puncture approach was challenging, CT-guided injections were performed. This study included all 34 patients who, regardless of age or SMA type, underwent at least one CT-guided injection. One procedure was excluded because of missing imaging data. Patient information was extracted from electronic medical records and PACS. This included birth date; injection date; onset date; diagnosis date; first injection date; spondylodesis; sex; treatment duration; number of lumbar punctures; number of CT-guided injections; total number of injections; injection level at the lumbar spine; laterality; performer experience; adverse events such as postprocedural puncture site pain, headache, radicular pain, fever, and nausea; radiation dose; procedure time; and CT scan number during the procedure.

Motor function assessment

Motor function scores were collected at baseline, at month 2, and subsequently at 4-month intervals. Based on the HIRA program guidelines, reimbursement is discontinued when there is a consecutive decline in motor function scores on two occasions, each measured at 4-month intervals. To assess motor function in SMA patients under 24 months of age, the HINE-2 is mandated by HIRA. For patients with Type 2 or 3 SMA who are over 24 months of age, the HFMSE is utilized. Consequently, for patients who initiated treatment before reaching 24 months of age, both HINE-2 and HFMSE assessments are required. In addition to the mandatory assessment tools required by HIRA, the CHOP-INTEND score was additionally administered to all patients included in this study, regardless of the age at treatment initiation. This was performed to allow consistent comparison of motor function across all patients and capture subtle changes in motor function in more severely affected SMA patients.

CT-guided transforaminal injection protocol

According to our hospital protocol, if blind lumbar puncture fails or following spinal fusion surgery, intrathecal nusinersen delivery is performed through CT-guided transforaminal injection. On the day of the procedure, the participating physical therapist assessed the HFMSE, HINE, and CHOP-INTEND scores before each intervention. Then, if a patient complained of pain from a previous procedure, prophylactic pain control was administered via intravenous infusion of tramadol (50 mg) mixed with normal saline (100 mL). The procedure was performed by a randomly assigned team comprising one MSK radiologist with 4 years of experience in performing the procedure and six fellowship trainees with < 1 year of experience. Before the procedure, an entry point was designated through a pre-scan, followed by needle advancement. Subsequently, a CT scan, including only the specified region, was performed to modify the needle course. If the needle entered the thecal sac, a final scan was performed. A 22G spinal needle was used in all procedures, and local anesthesia was administered using lidocaine. All procedures were performed using the same CT equipment (SOMATOM Force; Siemens Healthcare, Erlangen, Germany). Patients were surveyed for the presence and severity of postprocedural pain at the puncture site using a numeral rating scale, and any other adverse events were documented upon admission to the ward after the intervention. Patients were asked to rate their pain on a scale from 0 to 10, categorized as follows: 1–3, mild pain; 4–6, moderate pain; 7–10, severe pain.

Metrics in CT-guided transforaminal injection

Measurements of needle length, intrathecal needle advancement, needle location in the foramen, needle angle with V line, needle angle with V and H lines, and needle angle with skin were obtained from the last images of the CT-guided injection (Fig. 1). Needle length was defined as the distance from the skin entry point to the needle tip, whereas intrathecal needle advancement was defined as the distance from the thecal sac entry point to the needle tip. The needle location in the foramen was defined by dividing the foramen into anterior, middle, and posterior thirds. Three needle angles were measured—needle angle with the V line, the angle between a line perpendicular to the ground (or CT table) and the needle; needle angle with the V-H line, the smaller angle between a line perpendicular to the ground and the needle or between the ground and the needle; and needle angle with the skin line, the angle between a line perpendicular to the patient’s skin and the needle. Therefore, smaller values of the needle angle with the V line and needle angle with the skin indicate a more vertical entry of the needle to the ground and skin, respectively, whereas a smaller needle angle with the V and H lines suggests a more vertical or horizontal needle entry to the ground.

Fig. 1
figure 1

Imaging metrics in CT-guided nusinersen injection. These images depict the 2nd (A) and 13th (B) CT-guided transforaminal injections in a 28-year-old woman. In A, angle (a) represents the needle angle with a vertical line and angle (b) represents the needle angle with the skin. In B, length (c) signifies needle length and length (d) denotes intrathecal needle advancement. The needle foramen is divided into anterior (A), middle (M), and posterior (P).

Full size image

Statistical analysis

To investigate the impact of each independent variable on the outcomes, linear mixed models were employed considering repeated measurements in each patient. The independent variables used were: Sex (reference: women), BMI (kg/m2), injection age (years), spondylodesis (reference: none), CT-guided injection number, spinal level (reference: L2/3 & L3/4), laterality (reference: right), performer experience (reference: < 1 year), needle length (mm), needle angle with vertical line (°), needle angle with vertical and horizontal line (°), needle angle with skin (°), intrathecal needle advance (mm), needle location in foramen (reference: anterior), prophylactic pain control (reference: none), CT scan number, procedure time (s), pre-procedure HFMSE, pre-procedure HINE, and pre-procedure CHOP INTEND. The dependent variables were defined as pain (presence or absence), pain severity, radiation dose, and procedure time. Univariate models were initially used, and variables showing significance were selected for multivariate analysis. A p-value of < 0.05 was considered statistically significant. Multicollinearity was assessed for the variables selected from the univariate models. In cases where multicollinearity was suspected, separate multivariate analyses were performed for each variable. Subsequently, a subgroup analysis was performed for cases in which pain was equal to or greater than 1. All statistical analyses were performed using SAS version 9.4 (SAS Institute, Cary, NC, USA).