Mapping lumbar efferent and afferent spinal circuitries via paddle array in a porcine model

Abstract

Background: Preclinical models are essential for identifying changes occurring after neurologic injury and assessing therapeutic interventions. Yucatan miniature pigs (minipigs) have brain and spinal cord dimensions like humans and are useful for laboratory-to-clinic studies. Yet, little work has been done to map spinal sensorimotor distributions and identify similarities and differences between the porcine and human spinal cords.

New methods: To characterize efferent and afferent signaling, we implanted a conventional 32-contact, four-column array into the dorsal epidural space over the lumbosacral spinal cord, spanning the L5-L6 vertebrae, in two Yucatan minipigs. Spinally evoked motor potentials were recorded bilaterally in four hindlimb muscles during stimulation delivered from different array locations. Then, cord dorsum potentials were recorded via the array by stimulating the left and right tibial nerves.

Results: Utilizing epidural spinal stimulation, we achieved selective left, right, proximal, and distal activation in the hindlimb muscles. We then determined the selectivity of each muscle as a function of stimulation location which relates to the distribution of the lumbar motor pools.

Comparison with existing methods: Mapping motoneuron distribution to hindlimb muscles and recording responses to peripheral nerve stimulation in the dorsal epidural space reveals insights into ascending and descending signal propagation in the lumbar spinal cord. Clinical-grade arrays have not been utilized in a porcine model.

Conclusions: These results indicate that efferent and afferent spinal sensorimotor networks are spatially distinct, provide information about the organization of motor pools in the lumbar enlargement, and demonstrate the feasibility of using clinical-grade devices in large animal research.

Keywords: Cord dorsum potentials; Epidural spinal stimulation; Neuromodulation; Spinal cord; Spinal cord injury; Spinally evoked motor potentials.

Figure 1.

Four recording sites (A) were selected from each hindlimb to record spinally evoked motor potentials (SEMP): the vastus lateralis (VL), semitendinosus (SEM), tibialis cranialis (TC), and gastrocnemius caput (GC) muscles. The epidural array (B–C) was implanted so that it spanned the L5 and L6 vertebra, with the location highlighted in red and the anatomical midline denoted using a red dashed line, left is signified with a “Left” while the caudal direction is indicated with a “Caudal”. SEMPs were elicited (D) from the left and right rostral-most and caudal-most contacts using a bipolar configuration where black denotes the cathode and red the anode. The responses shown are an average of five repetitions at a given amplitude. The evoked responses demonstrated a high left/right and proximal/distal specificity. Representative SEMP data are from P002.

Figure 2.

3D reconstruction of the lumbar spine and implanted electrode arrays in each animal was done by (A) using the 3D fluoroscopy images (1) and segmenting bone (2) shown in brown, and the epidural paddle array shown in purple using a combination of automatic threshold selection and manually adjusting the selection to ensure proper segmentation then rendered in 3D space (3). Both fluoroscopy images and the 3D model were then used to determine paddle placement (B) by taking measurements of the L5 (top left) and L6 (top right) pedicles, then rostral and caudal measurements were taken with respect to the right or left side of the array (bottom left) to determine left/right paddle placement. The 3D model was then used to determine the final 3D placement of the array (bottom right).

Figure 3.

Motor selectivity from the implanted array is determined by (A) taking the area under the curve (AUC) of the recruitment curve for a given muscle from each electrode tested and subtracting the mean of the responses for a given muscle across all tested locations except for the AUC of the muscle being calculated. The resultant value is then normalized from 0 to 1, with 0 being the location with the worst selectivity, 0.5 being the average across locations, and 1 being the best location(s) for targeting a specific muscle. The SI is then mapped to the paddle representation and interpolated for each muscle for P001 (B) and P002 (C).

Figure 4.

The approximated rostrocaudal distribution of motor pools (A) of P001 (left) and P002 (right) using the SI. The array location is denoted by the gray bar and black line separating the L5 and L6 labels, where the lengths noted are the distances between the first and last contacts of the different arrays in the rostrocaudal axis. As probability increases, the width of the plot increases. The average between the animals (B) when adjusting for paddle placement (left) includes the location of the maximum SI, which is marked by a dark line and demonstrates a similar organization to the human lumbar enlargement (right). Human spinal segment chart was adapted from [10] using data from [11].

Figure 5.

Epidurally recorded CDPs during left tibial nerve stimulation (A) for the exemplar contact circled in red during sub-motor threshold stimulation (left) and supra-motor threshold stimulation (right) for P002. Example responses shown for every other row across the paddle demonstrate the magnitude changes over the array from P001 (B) and P002 (C) at sub-motor threshold stimulation.

Figure 6.

Difference in mean peak-to-peak CDP across the array, where red, white, and blue denote a higher-than-mean response, the mean response, and a lower-than-mean response, respectively. Both P001 (left) and P002 (right) demonstrate larger responses at the rostral portions of the array during left (L) and right (R) sub-motor threshold tibial nerve stimulation.