This observational study (Ref: [2022]413) was approved by the Institutional Research Ethics Committee. From January 2022 to July 2023, all pediatric and adult SMA patients with scoliosis who were referred to the Department of Anesthesiology for intrathecal nusinersen administration were included. Patients were referred there because of the anticipated difficulty in obtaining intrathecal access or previously failed LP. Exclusion criteria included patients with contraindications for LP and Cobb angle less than 11°. Written informed consent was obtained from all subjects and/or their parents.

Preprocedural planning and measurements via CT

When available, previous spinal CT scans were reviewed to assess the feasibility of the LP at various levels and to choose the best level with the largest interlaminar space. The L5/S1 space was considered as a last resort due to the decreased volume of the thecal sac. At the preplanned level, the degree of vertebral rotation [22] and the interlaminar space size were measured using the axial CT scan (Supplemental Digital Content 1, Fig. S1). The Cobb angles were measured using both the thoracic and lumbar centered curves for each patient, with the larger angle used to quantify the severity of spinal curvature, classified as mild (11° to 25°), moderate (25° to 50°), and severe (> 50°) [23]. The lumbar lordosis angle was measured employing Cobb’s method [24].

Positioning the patients with ultrasound guidance

As most of the patients were unable to sit independently, they were placed in a standard lateral decubitus position with the preplanned puncture site in the dependent position (Fig. 1a, b). The shoulders and hips were kept in line with one another, and the torso as perpendicular to the bed as possible although the coronal plane of the spine varied due to the rotation of the vertebral column. The knees were flexed to overcome the lumbar lordosis that narrowed the interspace between adjacent laminae.

Fig. 1

Posture correction guided by the orientation of the ultrasound probe. a Sagittal CT view demonstrating the posterior aspect of the L4 vertebra is oblique under standard lateral decubitus position, with the shoulder and the hip being placed at the edge of the bed. b-c Note the orientation of the probe relative to the edge of the bed (cephalad direction) to obtain the clear anterior complex image in the paramedian sagittal oblique view. d Sagittal CT image demonstrating the posterior aspect of the L4 vertebra being adjusted to align with the edge of the bed following posture correction. e–f Modified position with the shoulder moved forward to eliminate the cranial angulation of the probe is required for the clear anterior complex image, ensuring that the needle enters the spinal canal perpendicularly to the edge of the bed. g Axial CT view demonstrating the potential horizontal needle trajectory via the standard paramedian approach. h-i Note the probe being placed perpendicularly to the edge of the bed and horizontally, with the spinal canal centered. The blue dashed lines indicate the needle trajectory. The white dashed lines indicate the edge of the bed. The white arrow indicates moving the shoulders forward, away from the edge of the bed and the yellow arrow indicates movement of the hips backward, towards the edge of the bed

After palpating and marking the position of the spinous processes (SP) of the lumbar spine, a line connecting the SP which we referred to as “SP line” was drawn. Preprocedural ultrasound scanning was then performed using a 5–2-MHz frequency curved probe (Fujifilm Sonosite, Inc., Bothell, WA, USA) by an investigator (YX or CH) who was experienced in neuraxial ultrasound, following a protocol described previously by Dr. Karmakar [25]. While keeping the probe lateral and parallel to the “SP line”, the interlaminar spaces were identified and marked on the skin by counting upwards from the sacrum in the paramedian sagittal oblique (PSO) view. The preplanned intervertebral level was noted and marked and the anterior complex (AC), involving the anterior dura, posterior longitudinal ligament and posterior aspect of vertebral body (Fig. 1b, c) was identified. To obtain optimal visualization of the AC, the direction of the ultrasound beam should be kept as close to perpendicular to the posterior aspect of the vertebral body as possible. The orientation of the posterior aspect of vertebrae segments varied in scoliotic patients, due to the patient’s increased lumbar lordosis or kyphosis.

The patient was positioned on the bed to maintain visualization of the AC and at the same time to ensure the probe was perpendicular to the bed edge. This was done through adjusting the relative position of the shoulders and hips to align the posterior aspect of target vertebra with the edge of bed. The degree of adjustment was dictated by the cephalad or caudad angulation of the probe (Fig. 1a-c). Cephalad angulation of the probe indicated an anterior oblique direction of the posterior aspect of target vertebra. In response the shoulders should were moved forward away from the edge of the bed while the hips were moved backward towards the edge of the bed to eliminate the need of cephalad angulation (Fig. 1d-f). Caudal angulation of the probe indicated a posterior oblique direction of posterior aspect of target vertebra, and, thus, the shoulders were moved backward towards the edge of the bed while the hips moved forward away from the edge of the bed.

Identification of the needle entry point for a horizontal and perpendicular needle trajectory

The transverse median (TM) scan was then conducted by positioning the probe perpendicular to the long axis of the spine (Fig. 1g-i). If a rocking maneuver of the probe was required to see the laminae or articular processes on both sides at the same level with a clear AC, it indicated a clockwise axial rotation of the target vertebra from caudal-to-cephalic view (Fig. 2a-c). Conventionally, the degree of rocking required to produce the interspinous views was approximately the needle insertion angle via the median approach. To convert this angle onto a horizontal paramedian interlaminar approach, we introduced the concept of a lumbar puncture’s triangle. The right triangle’s hypotenuse was the angled median approach, its base was the horizontal paramedian approach, and its height was the lateral offset from the midline (Fig. 2b & e). To locate the needle insertion point, the probe was held perpendicular to the edge of the bed at the marked interspace and kept horizontal as verified by a spirit level attached on the probe, with the dural sac centered on the ultrasound screen (Fig. 2d-f); note that the thecal sac was at the lateral side of the interspinous ligament. Finally, the needle insertion point was noted and marked by the intersection of the midpoint of the probe’s long edge and short edge – all while keeping the needle horizontal and perpendicular to the bed (Fig. 3).

Fig. 2

Determination of the needle entry point for a horizontal and perpendicular needle trajectory. a-c The probe is rocked to produce the normalized interspinous view with the articular processes or laminae at the same horizontal level, indicating an angled interspinous median approach for lumbar puncture. d-f The probe is kept parallel, as verified by attaching a spirit level to obtain the image of the spinal canal with the dural sac centering in the ultrasound screen. Skin marks are made at the midpoint of the probe's long and short edges. The intersection of these two marks is identified as the needle entry point, through which we could direct the needle in the horizontal plane. Point A indicates the center of the spinal canal. Point B indicates the needle entry point for an angled interspinous median approach. Point C indicates the needle entry point for a horizontal needle trajectory via the paramedian interlaminar approach

Fig. 3

A horizontal needle path demonstrated by CT volume rendering reconstruction images. a The determined needle entry point (white arrow) of a SMA patient with scoliosis, 3.5 cm from the spinous process (SP) line. b Three-dimensional CT reconstruction in a volume rendering technique of this patient demonstrating the needle directed horizontally to the center of the dural sac on the convex side through the determined needle entry point at the L3 – L4 interspace. SMA, spinal muscular atrophy

Intrathecal injections

After sterile preparation of the skin, 2–5 ml of 2% lidocaine was infiltrated locally and a skin puncture was performed at the marked insertion point. A 25-gauge or 22-gauge needle was inserted horizontally and perpendicular to the edge of the bed to access the subarachnoid space. The needle position was confirmed by successfully aspirating 5 ml of CSF. Nusinersen (5 ml, 12 mg; Spinranza ®; Biogen; Cambridge, MA, USA) was intrathecally injected over 1–3 min.

A video that shows the whole process including the preprocedural ultrasound scan, needle entry site identification, and intrathecal nusinersen administration is available (Supplemental Digital Content 2).

Outcome measures

Patient demographics and clinical data were obtained from the electronic medical records. A video camera was used to record the process of all LP. Two investigators independently reviewed the video and recorded the following parameters. Discrepancies were resolved by a senior team member.

The primary outcome was the first pass success rate, defined as the needle achieving successful dural puncture through a single skin puncture and no needle redirection [12, 26]. Secondary outcome measures included the following: 1) overall success, with successful puncture defined as a confirmation of the CSF flow through the spinal needle and subsequent nusinersen administration; 2) first-attempt success, defined as the needle achieving successful dural puncture through a single skin puncture; 3) number of needle passes: defined as the number of needle direction adjustments in one puncture, including the first pass; 4) number of needle attempts, defined as the number of any separate skin punctures by a needle; 5) needling time, defined as the time from first needle contact with the skin (beginning with local anesthesia with lidocaine) to visualization of the outflow of CSF; 6) locating time: the time from when the probe was placed on the skin until the skin marking was complete; 7) total procedure time, defined as locating time plus needling time; 8) number of interspace levels tried for successful puncture; 9) adverse events during the procedure and within three post-dural puncture days.

Prior to the study, we devised a classification schema of LP procedural difficulty. A difficult case was defined as a patient who experienced > 2 attempts for at least two LP procedures or > 3 passes for at least two LP procedures.

Statistical analysis

The normality of data distribution was assessed using the Shapiro–Wilk Test. Continuous data were reported as mean ± standard deviation (SD) with range or median [inter-quartile range (IQR)] with range and analysed with the two-sample t-test or Mann–Whitney U-test, respectively. Categorical variables were reported as number and percentages and analysed with the Pearson Chi-Square test or Fisher’s exact test where appropriate.

All statistical analyses were performed using JMP Pro 16.2.0 software (SAS Institute Inc., Cary, NC, USA). A two-sided p-value of less than 0.05 was considered statistically significant.