History of Selective Posterior Rhizotomy

Sherrington in his classic studies on the neurophysiologic basis of muscle tone and spasticity, divided the brain stem from the spinal cord in cats and found that they became spastic. He then opened the spinal canal, cut the posterior rootlets and the spasticity was abolished.³

This work was applied to humans by Foerster in 1905, when he first performed the dorsal (sensory) rhizotomy in four patients. Subsequently, in 1913, he expanded his use of the dorsal rhizotomy to 159 patients.⁴ Foerster used electrical stimulation during the surgery to identify the level of the nerve roots. He described division of whole posterior nerve roots from L2 to S2, with sparing of L4 and L3 sensory roots to preserve sensation and quadriceps tone for standing. Although he was able to relieve spasticity, the dorsal rhizotomy was not widely used for the next 60 years.

In 1967, Gros⁵ modified the technique to a partial posterior rhizotomy by saving one or two rootlets in each root to safeguard proprioception. Gros demonstrated that his more conservative procedure resulted in the same beneficial reduction of spasticity. Gros’s “partial cutting” procedure, however, did not adequately predict and target which muscles would be affected by the surgery. To improve the outcomes, Gros’s associates subsequently made adjustments to the rhizotomy procedure.

In 1972, Sindou published anatomic documentation of the location of the 1A sensory fibers of nerve roots (believed to be responsible for spasticity) where they entered the spinal cord.⁶ In 1975, based on the Sindou thesis, Fraioli and Guidetti further modified the rhizotomy to sever these 1A sensory fibers and preserve other sensory fibers.⁷ In 1977, Fraioli and Guidetti⁷ modified the technique to a hemisection of nerve rootlets.

Fasano and colleagues published a method of selective posterior rhizotomy based on electromyographic responses to electrical stimulation.^10,11 Working at the level of the conus, they stimulated the dorsal spinal rootlets at increasing frequencies while recording the responses in corresponding anterior roots and muscles. This intervention, the “functional dorsal rhizotomy,” widened understanding of abnormal responses and identified which rootlets should be cut. This system of identification and selective cutting to preserve proprioceptive sensation remains the basis of current surgical rhizotomy procedures.

Operation at the level of the conus medullaris made identification of the lower sacral rootlets involved in bowel and bladder function difficult. In 1981, Peacock modified the procedure by shifting the site of surgery from the conus medullaris region used by Fasano to the cauda equina region.^12,13 This shift allowed for easier identification and preservation of both the S3 and S4 nerve roots, as well as separation of individual rootlets of the ventral and dorsal roots. In a consecutive series of 105 children with spastic cerebral palsy who underwent selective dorsal rhizotomy, wound infection, cerebrospinal fluid leak, hemorrhage, and bowel or bladder disturbance were absent.¹⁴ In a 20-year follow-up of his original series from Cape Town, 60 patients were shown to have maintained a decrease in muscle tone, increased range of motion, and improved function, including gait.¹⁵

Selective dorsal rhizotomy has proven to be a safe and effective procedure for treatment of spasticity, with long lasting results when performed in a methodical and careful fashion.^14,15

Patient Selection

Proper patient selection is critical to the desired reduction of spasticity from selective dorsal rhizotomy. First, spasticity must be confirmed and differentiated from other abnormal movement patterns or dystonia.¹⁶ Next all of the factors affecting the child’s motor function should be identified and the role of spasticity considered in the overall motor disability. A team approach to evaluation and treatment is used involving a pediatric neurosurgeon, physical therapist, orthopedic surgeon, physiatrist, pediatric neurologist, and developmental pediatrician.

Patients are first screened to rule out those with dystonic disorders. The majority of children with cerebral palsy who are purely spastic have a history of prematurity with or without intraventricular hemorrhage. Those who suffered a traumatic or anoxic injury at a later stage are more likely rigid and therefore less likely to benefit from rhizotomy. Progressive and degenerative disorders, such as a progressive encephalopathy should also be ruled out. Development milestones are documented, as are the child’s current motor function, equipment needs, therapy program, and orthopedic deformities. The goals of the patient and family are discussed and realistic expectations are set.

The neurologic examination begins with evaluation of muscle tone and range of motion. Velocity-dependent resistance to passive motion, brisk tendon reflexes, clonus, and clasp-knife phenomenon are all hallmarks of spasticity. Some children with spastic cerebral palsy will have relatively normal tone at rest, but their tone increases with activity highlighting the importance of observing these patients in motion, especially walking, if possible.

Spasticity can also mask underlying motor weakness or poor control of trunk muscles. The child’s unsupported sitting posture is observed. Inability to sit upright can indicate weakness of the trunk muscles. The child’s reaction to disturbances of equilibrium while sitting are also noted. Only then is antigravity control of the trunk and lower extremities best evaluated. The child’s ability to control body weight while standing upright and with unilateral standing is assessed. Evaluation of flexion and extension ability at the hip, knee, and ankle in unilateral standing is also a useful estimate of motor strength.¹⁶

Children with dystonia are relatively poor surgical candidates, unless spasticity is severe and only mild signs of dystonia are present. Because selective posterior rhizotomy does not address dystonia, it is important to differentiate it from true spasticity. More severe dystonia can be associated with rotatory movements around the axis of both upper and lower limbs, facial grimacing and dysarthria.

Orthopedic examination is performed and appropriate radiographs of the hip and spine are obtained. Range of motion is assessed and any limitations of hip abduction, the popliteal angle, and ankle dorsiflexion are documented. Hip and knee flexion contractures, increased hip internal rotation, and abnormal posture of the foot and ankle are common and need to be assessed. In instances where significant structural deformity is present, input from the orthopedic surgeon is used to determine whether corrective surgery is required before, after, or instead of rhizotomy.

In summary, the children who are appropriate candidates for selective dorsal rhizotomy fall into two groups, severely affected quadriplegic patients with little or no independent function and spastic diplegic children who are walking, with or without support. In the first group of children, the goals of surgery are improved comfort, positioning, and ease of positioning and transport by the caregiver. In spastic diplegics the goal of surgery is improving functional skills including gait and decreasing the risk of developing flexion contractures with the need for orthopedic intervention. Ambulatory patients who have poor trunk control, or poor lower extremity antigravity control, are poor surgical candidates. They are unlikely to have significant benefit from this procedure. The ideal candidates are spastic diplegic children between 4 to 6 years of age who walk independently but with an abnormal gait pattern (due to excessive hip adduction, a flexed hip and knee posture, limited range of hip and knee motion, and an equine posture of the foot and ankle). These children typically have good trunk control and antigravity muscle strength in the hip abductors which are crucial for stability in standing and walking.¹⁶

Procedure

Induction of general anesthesia may be performed with short-acting muscle relaxants. This allows easier intubation, positioning, and less need for retraction during surgical exposure. Nerve root stimulation is absolutely critical to ensuring the desired outcome with low associated morbidity. Neuromuscular paralysis must be reversed pharmacologically before nerve stimulation and recording. Adequate use of analgesics during and after the procedure are also very important to a successful outcome. Intravenous lines and an arterial line are established. A Foley catheter is placed. The child is then turned to the prone position with surgical bolsters under the hips and chest allowing the abdominal wall to move freely with respiration. Proper positioning is critical to prevent abdominal compression, secondary unwanted venous congestion and the risk of hemorrhage from epidural veins. All bony prominences are well padded with foam, and the head is turned to one side on a circular gel head rest. The patient’s back is kept in a neutral position, and the hips are flexed. The legs are supported by pillows. The arms are bent at the shoulder and the elbow. The plane between the highest point on the iliac crests is used as the level of the interspace between the spines of the fourth and fifth lumbar vertebrae. The spinous processes are carefully counted up to the spine of L1 and down to that of S2 allowing a midline incision to be marked out.

The skin is prepped in the usual sterile fashion. Two screens are placed to allow the head and the lower extremities to be visible to the anesthesiologist and the physiologist, respectively, during the procedure. The lower portion of the drapes is placed over a Mayo table so that the physiologist will be able to see and feel for muscle contractions and clonus in the lower extremities during the procedure. Needle electrodes for electromyography are placed in five lower extremity muscle groups bilaterally: adductor, quadriceps, tibialis anterior, hamstrings, and gastrocnemius. Electrodes are also placed in the external anal sphincter so that reflex arcs involved in sphincter control can be detected.

The incision is infiltrated with 0.25% Marcaine with 1:400.000 epinephrine at a total of 1 ml/kg. The incision is made with a #15 blade and self-retaining retractors are placed at both ends. Hemostasis is maintained with bipolar cautery. The lumbodorsal fascia is cut on either side of the supraspinous ligament, which is kept intact, and subperiosteal dissection is carried out to retract the paraspinal muscles laterally from L1 to S1. This can all be done with minimal blood loss, or use of the cautery.

Using a #15 skin knife the ligamentum flavum is opened at the inferior margin of the L5 lamina, and entrance in the epidural space is confirmed by visualization of fat or dura. A Penfield #4 can be used to confirm entry into the epidural space as well, but care must be taken not to cause any bleeding from epidural veins. The footplate of a small craniotome is inserted and brought up against the inferior margin of the lamina of L5, halfway between the spinous process and the facet. Care is taken not to injure the facets. A multilevel laminotomy is performed from L5 up to L2 in one step (Fig. 63-1). The same process is replicated on the opposite side. The supraspinous and interspinous ligaments are divided between L5 and S1 and the inferior portion of the laminar plate is then gently lifted up towards the cephalad extent of the surgical incision (Fig. 63-2). First, drill holes are placed at the base of the spinous process on either side at each level. Then, drill holes are placed in the lateral portions of the lamina at each level and threaded with non-absorbable sutures, each of which is held out laterally with a small hemostat. Then the plate is secured with suture and mosquito hemostats to the drapes superiorly. The lamina of S1 is cut on either side of the spinous process using a thin angled bone rongeur, with the supraspinous ligament left intact (Fig. 63-3). A suture is placed through the spine, which is then reflected caudally to expose the upper portion of the sacral dura. Bone bleeding is controlled with a slurry made from fibrin and Avatin. Bone wax is used to stop any persistent bone bleeding. Use of bone wax is minimized, as it can interfere with bone healing. Bleeding from epidural veins is controlled with bipolar coagulation.

FIGURE 63-1 A multilevel laminotomy is performed from L5 up to L2 in one step.

FIGURE 63-2 A, The inferior portion of the laminar plate is then gently lifted up towards the cephalad extent of the surgical incision. B, The lamina of S1 is cut on either side of the spinous process with a thin, angled bone rongeur, with the supraspinous ligament remaining intact.

FIGURE 63-3 A, Using the rhizotomy electrodes, the groove between the posterior and the anterior roots is identified and the two are gently separated. B, The rootlet is elevated with rhizotomy electrodes and then sectioned with microscissors.

At this point the child is put in mild reverse Trendelenburg position to decrease the amount of cerebrospinal fluid loss during the procedure. After the dura has been cleared of epidural fat, it is opened in the midline with a sharp #15 skin knife and held by traction sutures to the muscular layer. The arachnoid layer is left intact initially. A patty is placed over the arachnoid before a small opening is made. Following gentle cerebrospinal fluid aspiration, the arachnoid is opened widely to expose the cauda equina (Fig. 63-4).

FIGURE 63-4 Intraoperative micrograph illustrating arachnoid dissection with microscissors to expose cauda equina.

Two insulated rhizotomy hook electrodes are connected by sterile cords to the electrical stimulator. Electric stimulation and recording are then used to identify root levels, and differentiate between motor and sensory bundles. The nerve roots are inspected at the exit foramina. At the dural exit of the nerve root, the motor root generally lies just slightly cephalad and anterior in the neural foramen. The motor nerve roots are generally slightly larger and have increased tortuosity of vasculature compared to the posterior roots (Fig. 63-5).

FIGURE 63-5 Intraoperative micrograph demonstrating the larger caliber and vascular tortuosity of a representative ventral nerve root.

The largest nerve root is that of S1. Using the rhizotomy electrodes, the groove between the posterior and the anterior roots is identified and the two are gently separated (Fig. 63-6). The anterior root is stimulated. This should produce flexion at the knee and plantar flexion at the ankle. Stimulation of the anterior root of S2 should produce plantar flexion at the ankle and flexion of the toes. By counting upward, the L2 root is identified and separated into its anterior and posterior components. The posterior root of L2 is initially stimulated with a single 0.1-millisecond stimulus until the threshold for muscle contraction is reached. After a threshold to a single-pulse stimulus has been established, a subthreshold tetanic stimulus at 50 Hz is applied, and the response is monitored. Abnormal responses are used to identify which rootlets should be divided. The selection of rootlets for division is discussed in more detail in the section on intraoperative monitoring. The stimulation and recording are done at each level, starting at L2 and continuing to S2 on the side opposite to the surgeon. There are normally two to four rootlets at the L2 level and an increasing number at each subsequent level down to S1, where there are approximately 8 to 11 rootlets. The S2 root is the first root that is smaller than the preceding root and usually contains two to six rootlets. The surgeon then moves to the opposite side of the operating room table and the process is repeated on the contralateral side of the patient. This allows for superb visualization of the nerve roots on each side while limiting the lateral extent of the laminotomy.

FIGURE 63-6 The laminar plate is secured back into position using the previously threaded sutures. The spinous process of L5 is also sutured to the spinous process of S1.

After those dorsal rootlets associated with an abnormal response have been sectioned (Fig. 63-7), the subarachnoid space is irrigated. Persistent bleeding is controlled with bipolar cautery at low settings. The dura and arachnoid layers are then closed in a continuous locking watertight fashion. The child is taken out of reverse Trendelenburg and a tuberculin syringe and needle are used to inject intrathecal preservative free morphine. The dura is covered with fibrin glue. A brief Valsalva maneuver is applied to confirm watertight dural closure. The laminar plate is secured back into position using the previous partially threaded sutures. The spinous process of L5 is also sutured to the spinous process of S1 (Fig. 63-8). The paraspinal muscles are reattached to the interspinous ligament by suturing through the muscle on one side, then through the interspinous ligament into the muscle on the opposite side, and then back again through the interspinous ligament into the muscle on the original side before the suture is tied. This brings the muscles into tight apposition with the bone and ligaments, eliminating any surgical dead space. The supraspinous ligament is then sutured to the lumbosacral fascia covering the muscle before the skin closure. The skin is closed in two layers. The inner layer is closed with vicryl suture while the outer layer is closed with a continuous locking nylon suture. The wound is dressed, and the child is taken to the recovery room.

FIGURE 63-7 Intraoperative micrograph depicting multiple dorsal nerve roots after selective rhizotomy and sectioning.

FIGURE 63-8 The paraspinal muscles are sutured to the interspinous ligament by suturing through the muscle on one side, then through the interspinous ligament into the muscle on the opposite side, and then back again through the interspinous ligament into the muscle on the original side before the suture is tied.

Intraoperative Monitoring

Responses to rootlet stimulation are evaluated by a neurologist with expertise in intraoperative neuromonitoring, and if possible a physical therapist or physiatrist who is familiar with the child’s clinical picture. Rootlet selection for division is based on electromyographic response to dorsal rootlet stimulation, visual and palpable muscle contraction, and the child’s clinical picture. Sacral-level rootlets associated with anal sphincter activity are spared.

Ten lower extremity muscle groups and the anal sphincter are evaluated with electromyography. The rootlets are stimulated using the rhizotomy electrode hooks with a constant-voltage electrical stimulator that allows delivery of single pulses and trains of stimuli. Recordings are made with pairs of noninsulated 2.5-cm needle electrodes and displayed on an 11-channel electroencephalography chart recorder.

Motor roots can be identified from sensory roots based on their lower threshold for inducing muscle contraction. The motor and sensory roots are identified and easily separated using the rhizotomy hook electrodes (Fig. 63-9). Because the threshold for muscle contraction is much lower for anterior roots (usually 100 mV to 1 V) than for whole posterior roots (usually 1 to 10 V), electrical stimulation can also be used to confirm separation of the posterior and anterior roots.

FIGURE 63-9 Intraoperative micrograph of nerve root dissection performed with gentle sweeping movements of insulated rhizotomy hook electrodes in line with nerve fibers.

Stimulation of the anterior roots provides confirmation of spinal levels. Identifying root levels helps prevent injury to the lower sacral roots, thereby reducing the risk of developing bowel or bladder incontinence. First the S1 level is identified as the largest nerve root. Stimulation should produce flexion at the knee and plantar flexion at the ankle. Stimulation of the S2 nerve root should produce plantar flexion at the ankle and flexion at the toes. Stimulation of posterior rootlets is then carried at every level beginning at the L2 dorsal root. Rootlets producing contraction of the anal sphincter (S3–S4) can be identified using EMG.

Using gradual increases in the voltage of single square-wave pulses of 1-millisecond duration, threshold levels are established for simulation of the posterior root. This serves as a guideline to the stimulation threshold for the individual rootlets in that root. The root is then subdivided into its rootlets, which are stimulated in turn. The operator applies trains of stimulation, starting at about 50% of the single-pulse threshold and gradually increases the voltage until a small movement of the lower extremities is detected or a distinct electromyography recording is obtained.

The electromyographic response is analyzed in three ways: by duration, location, and pattern. The responses associated with rootlets that are to be spared have a duration no longer than that of the stimulus (1 second), are located in muscles innervated by that nerve root level, and have a decremental, squared-off wave pattern. Responses associated with rootlets that are to be divided may be associated with one or more of the following abnormalities: sustained duration (longer than 1 second), spread to muscles not innervated by that level, and an incremental, clonic, multiphasic or irregular waveform.

Suprathreshold stimulation should be avoided, as it may produce diffuse spread throughout the lower extremity, which may be misinterpreted as an abnormal response. Ongoing background activity may also be present and confuse the operator. If excessive background firing is present, or on the other hand responses are obtained only with higher voltages, small changes in the depth of anesthesia may alleviate the problem.

If there were not an appropriate number of rootlets at S1 associated with an abnormal response as would be expected in that child, S2 rootlets are evaluated for abnormal responses and sectioned only if stimulation did not result in contraction of the anal sphincter. If the patient does not have much spasticity in the quadriceps, but is in severe equinus, only the S1, L5, and L4 dorsal roots are sectioned.

Postoperative Care

Postoperatively, the patient is kept comfortable with appropriate analgesics. The Foley catheter is kept in place for 72 hours, given the risk of urinary retention with the use of intrathecal morphine. For the first three postoperative days, patients are kept flat in bed to help prevent cerebrospinal fluid leaks. On the third postoperative day, physical therapy is begun for positioning, range of motion, bed mobility, and family instruction. Sitting and maximally-assisted transfers are started on the fourth postoperative day. Flexor spasms, hypersensitivity of the lower extremities, and weakness are common in the patients in the early days after surgery and may last for several weeks. Continued physical therapy must be provided on an outpatient basis following discharge to help patients regain previous functional status and to improve skills and abilities beyond the preoperative levels.