Natural history and management of cerebral palsy

Cerebral palsy (CP) affects approximately 2–3/1000 people in Europe.1 Bilateral spastic CP, with predominant lower limb involvement, is the most common subtype of CP in Europe, accounting for over half of the patients.1 The severity of spastic CP is classified according to the Gross Motor Function Classification System (GMFCS).2 The Gross Motor Function Measure (GMFM) has been used along with GMFCS to describe the natural history of spastic CP in two landmark studies.2 In particular, the Adolescent Study of Quality of Life, Morbidity and Exercise found a decline in GMFM during the middle to late teenage years for patients at GMFCS III–V.2 Even though CP is considered a non-progressive condition, a decline in motor ability with age is evident. Interventions are directed at minimising or even halting this natural decline.

Spasticity and associated spasms cause muscle stiffness, pain and discomfort, which interfere with function. Muscles in children with CP are different compared with children developing normally; they are smaller, weaker and demonstrate histological architectural changes. Spasticity contributes to muscle shortening, contracture development, torsion of long bones and joint degeneration. Children with severe spasticity accumulate progressive lower limb muscular, skeletal and joint deformities before reaching skeletal maturity. Without early intervention to reduce tone, many children require multilevel orthopaedic surgery, which often includes soft tissue release in combination with femoral osteotomies and hip reconstruction. Watt et al3 prospectively studied 74 children with spastic CP and reported that 61% had already undergone orthopaedic procedures by 8 years of age. Moreover, spasticity can interfere with the ability of carers to look after children with CP.

Multidisciplinary goal-oriented early intervention promotes disease modification rather than symptom management and is preferable for positive long-term functional outcome. Occupational therapy, physiotherapy and speech therapy are first-line interventions and aim to improve activities of daily living (ADL). Adjunct pharmacological interventions for bilateral spastic CP can be either oral (baclofen, diazepam or clonazepam) or intramuscular (botulinum toxin A or phenol). Surgical treatment options include intrathecal baclofen (ITB) or selective dorsal rhizotomy (SDR) for tone management, and orthopaedic surgery to improve lower limb alignment.

Evolution of selective dorsal rhizotomy

Lumbosacral dorsal rhizotomy for spasticity was first advocated by Foerster in 1908. He observed that patients with tabes dorsalis who had hemiplegia did not develop spasticity, so he hypothesised that division of the dorsal (sensory) roots could relieve spasticity.4 Working with Tietze, his technique involved complete division of the dorsal roots of L2, L3, L5 and S1, sparing the ventral (motor) roots, leading to marked improvement in spasticity but also significant muscle weakness as well as loss of sensation and proprioception.4 Forty-five operations were conducted from 1908 using Foerster's technique; surgical complications were frequent, and eight patients died as a result of meningitis.4

The complications of deafferentation led to the disuse of SDR until the 1960s when division of only a fraction of the dorsal rootlets, maintaining sufficient afferent input to preserve sensation and proprioception, was considered.5 Partial sectioning of the dorsal nerve roots on the basis of intraoperative electrophysiological stimulation was introduced by Fasano in 1978 and is still used today. Nerve rootlets producing sustained muscle activation with abnormal widespread involvement of unrelated muscles in the trunk and upper limbs on stimulation were assumed to be contributing to spasticity and were divided. This led to good long-term results. The technique was adopted and popularised by Peacock and Arens in the 1980s.5 Surgery involved a multilevel laminectomy or laminoplasty, typically from L1 to L5. The nerve roots were identified at their individual foramina. The sensory nerve roots were dissected from the motor ones at the foramen; the afferent roots were then further dissected into rootlets, evaluated electrophysiologically and divided. However, concerns about the potential impact of extensive spinal surgery in ambulant children with CP led to the development of a single-level technique involving division of dorsal rootlets at the level of the conus.6

This single-level technique, as it is often practised today, involves identification of the conus on intraoperative ultrasonography through a small T12–L1 midline fenestration (figure 1). The conus is then exposed by removing the spinous process and central component of the appropriate lamina. At this level, removal of a single lamina allows exposure of the dorsal (sensory) and ventral (motor) nerve roots, from L1 to the sacral roots, as they enter and leave the conus. The dorsal (sensory) nerve roots are located superficially and the ventral (motor) ones lie deeper ; there is a clear identifiable plane between the two. The motor roots are protected throughout the procedure. The L2 to S1 dorsal nerve roots are identified, divided into rootlets and systematically stimulated to determine their threshold amplitude (figure 2). Electromyographic (EMG) responses to 50 Hz tetanic stimulation at threshold amplitude are graded for each nerve rootlet (figure 3). Rootlets with diffuse responses beyond their segment (grade 3 and 4) are divided; in total, up to 60–70% of the sensory roots are divided. Fifty per cent of the L1 sensory nerve root is then divided as it exits its foramen. Recovery following the single-level technique is rapid, and children typically resume physiotherapy after 2–3 days of bed rest.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Intraoperative ultrasound is essential to locate the conus intraoperatively in the single-level procedure. These ultrasound images demonstrate (A) sagittal view defining the triangular shape of the conus (white arrow); thin parallel lines around the conus represent the nerve roots that constitute the cauda equina within the hypoechoic cerebrospinal fluid. (B) An axial view showing the circular hypoechoic conus; the white arrow defines the boundary between the dorsal sensory nerve roots (single asterisk) and the ventral motor roots (double asterisks).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Intraoperative photomicrographs demonstrating intradural exposure at the level of the conus, typically through an L1 laminotomy in the single-level procedure. (A) Following a midline longitudinal durotomy, the afferent nerve roots on the right (white arrow) and the conus (asterisk) are evident. (B) The afferent nerve roots, from L2 to S1, are mobilised medially to expose the ventral motor roots (asterisk). (C) A small protective cottonoid is placed over the motor roots (asterisk). (D) A sling is passed deep to the sensory nerve roots, and Peacock probes are used to stimulate the roots systematically (white arrow) from lateral, corresponding to L2, to medial, corresponding to S1.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Intraoperative electromyography showing responses to tetanic stimulation of individual nerve roots to 50 Hz stimulated at their own threshold. In (A), the response is restricted to a single muscle group on stimulation of S1 (hamstrings) (grade I). In (B), a response from an adjacent muscle group on stimulation of L3 is recorded (vastus lateralis; adductor longus stimulated also) (grade II). In (C), there is diffuse activation of several muscle groups on the same side on stimulation of S1 (hamstrings) (grade III). In (D), there is a diffuse and bilateral muscle group response on stimulation of L5 on the left (peroneus longus) (grade IV). The objective of the procedure is to divide 60–70% of the sensory nerve roots between L2 and S1, including as many grade III and grade IV roots as possible. Also, 50% of the L1 sensory root is subsequently divided at its intervertebral foramen.

Patient selection for selective dorsal rhizotomy

The decision as to whether SDR is the optimal procedure for a particular child at that stage of motor development is not easy and needs to be made within a multidisciplinary context. Agreement by all involved, including carers, on the goals of treatment for the individual child is crucial. The principal goals of SDR depend on a child's motor abilities and include improved motor function, increased mobility and independence, improvement in ease of care and reduction in pain.

The general selection criteria defined by Peacock in 1987 still apply.5 Our current criteria are shown in table 1. Currently, most children selected for SDR are between 3 and 14 years of age.

As illustrated in a recent review, selection criteria for SDR vary between centres and have not been generally validated.7 They are primarily based on clinical rationale rather than clinical evidence. In many centres, several inclusion and exclusion criteria are not based on standardised or reproducible measurements. Most centres, however, use a multidisciplinary approach that evaluates some or all of the International Classification of Function domains, particularly the body structure and function, activity and personal and environmental factors. Very strict selection criteria may influence outcome. In the Oswestry series, only 35% of referred children satisfied the selection criteria and underwent SDR.8 In this series, children consistently showed improvement in the GMFCS level after SDR; this was not reproduced in other large series, where children tended to improve only within the same GMFCS level. It is important to remember that as children with CP get older their mobility difficulties include more orthopaedic-type musculoskeletal difficulties rather than just spasticity, which can explain differences in outcome post-SDR. Although a trial of oral baclofen is not necessary prior to making a decision on SDR, targeted botulinum toxin injections to reduce spasticity in some muscle groups are useful in evaluating the potential functional impact of SDR.

Outcomes of selective dorsal rhizotomy

SDR is a permanent and effective treatment for spasticity in children with bilateral spastic CP.9 Several studies show there is no loss of gross motor control in patients with bilateral spastic CP, with most patients showing sustained improvement in gait and spasticity.5 ,8 ,10–14 There is high-level evidence that SDR combined with physiotherapy has better results than physiotherapy alone, although patients are unlikely to decrease GMFCS grade.9 ,14–17 This conclusion is largely based on the results of three well-designed randomised controlled trials (RCTs) conducted in North America (Toronto, Vancouver and Seattle) in 1997 and 1998.15–17 In the Toronto study,15 evaluation at 12 months showed significant improvements in GMFM scores, knee and ankle tone, passive ankle range of motion, soleus EMG reflex activity on forced dorsiflexion and foot-floor contact pattern. In the Vancouver trial,16 significant improvements were observed at 1 year in GMFM, spasticity and range of movement in the group undergoing SDR combined with physiotherapy. At 24 months in the Seattle study,17 the combined SDR/physiotherapy group showed a significant reduction in spasticity compared with the physiotherapy-only group. However, improvements in GMFM were not significant. This result could be attributed to the fact that only a mean of 25% of dorsal roots were divided in the Seattle study, which is in contrast to the currently accepted SDR technique, in which between 50% and 70% of the sensory nerve roots are divided.

Other studies have shown SDR to be superior to other interventions in the management of spastic CP. In their RCT comparing SDR with botulinum toxin A injections, Wong et al18 found that the effectiveness of SDR was of a longer duration than botulinum toxin A. Moreover, patients with SDR have reduced requirements for orthopaedic interventions and intramuscular botulinum toxin A injections.10 ,18–21 In their non-randomised patient series, Kan et al compared two groups of age and GMFCS score-matched children; an SDR group underwent surgery prior to 1997, and an ITB group underwent surgery after 1997. The SDR group showed significantly better improvements in Ashworth scale, lower extremity passive range of motion and GMFM scores.22 The SDR group had a significantly reduced need for orthopaedic surgery compared with the ITB group. A study evaluating the rate of orthopaedic surgery after SDR showed that in all age groups 25% of independent walkers and 44% of assisted walkers required orthopaedic surgery over a 9-year follow-up.19 Those undergoing SDR at a young age demonstrated the lowest requirement for orthopaedic surgery after SDR.21

Evidence on the long-term effects of SDR is emerging and is generally positive.5 ,10–12 Nordmark et al11 reported on a group of 35 children with spastic diplegia over a 5-year postoperative period. SDR resulted in immediate reduction of tone in adductors, hamstrings and dorsiflexors, with no recurrence of spasticity over 5 years. Similarly, there was significant improvement in passive range of movement in hip, knee and ankle joints, as well as significant improvements in GMFM. In a 10-year follow-up study of 24 children, undertaken by the same group, additional improvement was evident in the functional skills, mobility and caregiver assistance (both self-care and mobility) domains of the Paediatric Evaluation of Disability Inventory (PEDI) scores between 5 and 10 years after SDR.12 This was particularly evident in children in the GMFCS I to III subgroups. Children in the GMFCS IV and V subgroups demonstrated only small changes between 5 and 10 years after SDR.

Dudley et al10 reviewed long-term follow-up data of children who were evaluated by a multidisciplinary team preoperatively and at 1, 5, 10 and 15 years after SDR. After SDR, through adolescence and early adulthood, statistically significant and durable improvements in lower limb muscle tone, gross motor function and performance of ADL were identified. Only 28% of children required further lower extremity orthopaedic surgery after SDR during the follow-up period, a significant improvement compared with the expected 61% intervention rate for similar children by the age of 8 years.3 The multidimensional benefits of SDR, reflected in the PEDI score, were also evident in this study, where significant gains in self-care and mobility persisted to early adulthood. However, a recent systematic review of interventions for CP found no clear evidence that SDR improves general activities and participation, which arguably should be the primary goal of the procedure.9 Although there are enduring improvements in GMFM after SDR, most children will remain within their GMFCS grade following surgery, with the exception of the Oswestry experience. Nevertheless, the majority of former patients with SDR report improvements in their ADL and that they would recommend SDR, with very few reporting negative impressions of the procedure.21 Indeed, the Cape Town experience found that none of Peacock's original cohort of patients required help with ADL.5

Some studies have reported less satisfactory long-term outcomes, thus contributing to decisions on patient selection. Children with spastic quadriplegia had poorer outcomes compared with those with diplegia.23 Children over 10 years of age were demonstrated to have better long-term outcomes with multilevel orthopaedic surgery than with SDR.24 This underlines the particular challenges related to patient selection in this age group, where the primary difficulties with mobility often arise from weakness and structural lower limb deformities rather than pure spasticity. It is now generally agreed that good long-term results are achieved in young children, who are diplegic rather than quadriplegic, and those whose GMFCS grade is II–III. A recent study of 54 children followed up for 2 years confirmed that children between 4 and 7 years old with preoperative GMFM scores between 65% and 85% benefit most from SDR.13

Permanent complications are now rare after SDR. In their review of long-term adverse effects of SDR, Grunt et al25 reported that back pain and spinal abnormalities were common, including kyphosis, scoliosis, lumbar lordosis, spondylosis and spondylolisthesis. But they found insufficient evidence to conclude such abnormalities are the direct result of SDR rather than related to the natural history of spastic CP. A large patient series has shown that limited laminectomies at the level of the conus are not associated with long-term spinal deformity.6 Transient dysaesthesiae are common, but permanent hypoesthesia is rare. Transient urinary retention was frequent during the earlier days of the resurgence of SDR, while permanent urinary incontinence was rare. Most centres now advocate pudendal monitoring and limitation of the division of the S2 nerve root in order to limit adverse effects related to detrusor function. As a result, the current risk of incontinence is very low.

In practice, the importance of setting appropriate specific objectives with families, within the context of the surgical risks and the need for intensive postoperative rehabilitation, cannot be overemphasised. In our experience, in ambulant children, SDR leads to improved mobility, increased stamina, better balance and fewer falls. Children who walk with assistance become more independent. Sitting and standing posture improves. In addition, the pain associated with spasticity responds well to SDR.