|Home | About | Journals | Submit | Contact Us | Français|
In this article we present the outcomes of primary nerve reconstruction and results of secondary procedures performed to restore or enhance the function of the upper extremity. Ninety-nine patients were operated between 1978 and 2000; the results are analyzed for 84 patients with adequate follow-up. Seventy-five patients underwent 77 primary brachial plexus reconstructions and 24 patients underwent only secondary procedures. Nerve reconstruction included microneurolysis, direct end-to-end and direct end-to-side neurotizations, indirect neurotizations with interposition nerve grafting, and direct nerve-to-muscle neurotizations. Muscle (n=135 pedicled and 48 free) and tendon (n=80) transfers were used to enhance function. The results were analyzed in relation to the type of the injury (Erb's versus global paralysis) and the denervation time. The results of reconstruction showed improvement in all muscles tested at a statistically significant level (p<0.001). The results were good and excellent for 84.87% of biceps, 73.55% of supraspinatus, 71% of deltoid, and 67.8% of triceps restoration. The Mallet scores and the Gilbert-Raimondi scores improved after reconstruction in all patients at a statistically significant level. The outcomes in general were better if the number of avulsed roots was fewer. The denervation time (DT) affected primarily the outcome of the hand function. Patients with DT less than 3 months underwent less surgeries (1.3 surgeries per patient) to complete the reconstruction than patients with DT between 3 and 6 months (3.1 surgeries per patient).
Infants with obstetrical brachial plexus palsy (OBPP) face either spontaneous complete or incomplete recovery or permanent disability in the use of their paralyzed extremity. OBPP also has psychological and socioeconomic implications for both the family and the patient. It is well documented that the majority of these patients (80–90%)1 recover without surgical intervention. However, the management of the almost 20% of this population of patients who will not recover spontaneously has always been complicated. With the evolution of microsurgery and the advancement of new techniques, reconstructive surgery has been the optimal therapeutic modality for these patients.
However, the debate about the timing and strategies of surgical intervention is ongoing and it is sustained primarily by the lack of a consensus among the different centers that carry out surgical exploration and reconstruction in children with OBPP. In this article the achievable outcomes of aggressive microsurgical reconstruction performed by a single surgeon (JKT) are presented for 99 children with OBPP. The outcomes are presented according to the type of the OBPP lesion (Erb's versus global palsy), the severity of the lesion (avulsion versus rupture), and the timing of surgery (i.e., denervation time, DT). Emphasis is given to the potency of different motor donors in relation to the targets selected to be reinnervated.
From 1978 to August 2000, a total of 156 patients with obstetrical brachial plexus injuries were examined in our center. Of these patients, 99 (63.46%) had surgery to restore function to the involved extremities and 57 (36.54%) were not operated. A total of 101 extremities were reconstructed after being diagnosed with OBPP; 97 patients had unilateral paralysis and 2 had bilateral paralysis. Evaluation of the results was performed for 84 (84.84%) patients with adequate follow-up. Inclusion criteria for this study included all patients with a minimum follow-up of 12 months after primary reconstruction of the brachial plexus and 3 months after secondary reconstruction. Seventy-seven (76.24%) brachial plexuses in 75 patients underwent primary exploration and reconstruction and in 24 (23.76%) patients surgery was limited to secondary reconstruction to correct established deformities. The mean follow-up period was 56.93 months (range from 3 months to 20 years). The distribution of the patients between the two genders was almost equal, with 51 males (51.51%) and 48 females (48.48%), a pattern similar to that in other studies.2 The right side was affected slightly more often (48 patients or 48.48%) than the left side (46 cases or 46.46%). Five patients (5.05%) presented with bilateral brachial plexopathies. Of these five patients with bilateral OBPP three patients had spontaneous recovery of one side by the time of the first surgery. These numbers correlate with other published epidemiological reports on OBPP.3,4
The average age of the patients during the first office visit was 37 months or 3.1 years (range from 1 month to 22 years) and the mean DT for the whole population of patients (99 patients or 101 brachial plexuses) was 41.1 months or 3.42 years (range 1.5 month to 22 years). The DT among the patients who had exploration of the brachial plexus and primary reconstruction was 27.75 months or 2.3 years (range 1.5 month to 16 years). The DT among patients who had only secondary reconstruction was 88.17 months or 7.34 years (range 11 months to 22 years). The DT was calculated by taking into account the date of birth and the date of brachial plexus reconstruction. The same principle was also followed for the few patients who had been operated by other physicians prior to presentation to our center. Analysis of the distribution of patients according to DT showed that the majority of the cases (46 cases or 45.54%) had a DT above 2 years, with the rest of the cases distributed rather evenly between the four trimesters of the first year of life. Preterm neonates accounted for only 8.08% (eight cases) of our population of patients and post-term neonates for 2.02% (two cases). Vertex presentation accounted for 90 cases (90.9%), breech presentations for 7 cases (7.07%), transverse presentations were diagnosed in 1 case (1.01%), and shoulder presentation in 1 case (1.01%). The vast majority of the babies were delivered via the vagina (96 deliveries or 97%). Three babies were delivered via cesarean section (3%). The average birth weight was 4.163 g. From the total population studied, 32.6% of the delivered babies were considered macrosomic (>4500 g at birth), indicating that fetal macrosomia, especially in diabetic mothers, can be a strong predictor of OBPP.5 Delivery of macrosomic infants is also associated with a higher chance of instrumental vaginal delivery. In our population of patients, one out of three deliveries (31.3%) was assisted by forceps and one out of five deliveries (18.2%) was assisted by suction. Excessive manipulation by the obstetrician was documented in the majority of the cases (89%). The average 1-minute Apgar score was calculated to be 4.88 (range 1 to 9) and the mean 5-minute Apgar score was calculated to be 7.33 (range 3 to 10).
Mechanical factors affecting the delivery of the baby included shoulder dystocia for 72.3% of the cases (n=72) and cephalopelvic disproportion, as documented by ultrasonography prior to the delivery, for 15.6% of the cases (n=15). The umbilical cord was wrapped around the neck of the newborn on 12 occasions (12.1%). This could have caused injury by indirect compression of the underlying brachial plexus of the delivered baby, but there was also documented shoulder dystocia in all 12 cases.
Gestational diabetes during the third trimester was documented in 10 cases (11.11%). Other problems during pregnancy included hypertension in four cases (4.04%). The list of gestational complications also included anemia, hypothyroidism, cervical incompetence (necessitating cervical cerclage), ruptured membranes, threatened abortion, appendicitis (requiring an appendectomy), and fetal distress requiring emergency cesarean section.
The most common associated injury with OBPP was fracture of the ipsilateral clavicle in 13.5% of the cases (n=11). The incidence of clavicular fractures in these babies is about 2 to 3 per 1000 live births.6,7 Other associated injuries involved shoulder diastasis or dislocation in three babies (3%) and fractures of the humerus (n=2) and femur (n=2). Facial nerve palsy8 that resolved spontaneously was encountered in two cases (2%). Documented asphyxia of the baby with respiratory distress requiring intubation was documented in only four babies (4.9%), three of which were premature (<38 weeks of gestation). Cephalohematoma from the use of suction was encountered in one case (1%).
In our population of patients (99 patients, 101 brachial plexuses) 62.38% of the cases (63 brachial plexuses) were diagnosed with a clinical picture compatible with global palsy (C5-T1 involvement) and 37.62% of the cases (38 brachial plexuses) were diagnosed with Erb's palsy. However, the upper root palsy type (C5-C6 with or without C7 involvement), or else Erb's palsy, according to some authors occurs more frequently9,10,11,12 and complete lesions (C5-T1) occur second in frequency. The higher number of global palsies probably reflects the trend that infants with more devastating lesions were much more likely to be referred to our center than infants in which the hand was preserved. The isolated lower root palsy, or Klumpke's type (C8-T1), was never encountered in our series, which correlates with other reports that it is a very uncommon to nonexistent entity.12,13
The clinical examination included tests of all upper extremity musculature, always in comparison with the contralateral normal side in a proximal-to-distal fashion. A few tools have been employed (keys, toys) to drive the baby's attention to move actively the involved extremity. Special attention was paid to determine muscle contraction by palpating individual muscle groups. The British Medical Research Council Grading System, expanded with intermediate grades (e.g., M2−, M2, M2+) was used for muscle strength evaluation and it was documented in a brachial plexus chart with the rest of the clinical examination. Furthermore, the overall excursion of movement at the shoulder, elbow, and hand was documented with standardized videos.
To assess shoulder function a modified Mallet scale was used with a grading of 1 to 4 for evaluation of shoulder abduction, shoulder external rotation, hand to nape position, hand to back position, hand to pocket, and hand to mouth position. Hand function was evaluated by using a modified Gilbert-Raimondi hand scale (from 0, which equals total palsy, to 6, which is a normal functioning hand). In addition to the Gilbert-Raimondi scale, in cooperating children a dynamometer was used to evaluate grasping power of the hand and pinching power of individual fingers.
In general, if the arm was internally rotated with the elbow extended and adducted, an upper plexus injury was suspected. Pupillary changes (Horner's sign) along with a presentation of a totally flail extremity with the hand in supination gave rise to suspicion of global paralysis. Color and trophic changes of the arm were also observed. Tenderness and edema upon palpation of the supraclavicular area should raise suspicion of a possible clavicular fracture. In cases of an upper plexus lesion a neuroma is usually palpated (documented in 10 patients), which invariably signifies postganglionic rupture of one or several upper roots. Pinching different dermatomes and evaluation of the elicited responses are used to test sensibility. In older patients the static and moving two-point discrimination test along with the von Frey pressure threshold test were performed. The presence of Horner's sign (n=37, 36.63%) was also recorded because it would signify a partial or complete avulsion of C8 and/or T1 roots.
During the first visit, radiographs of the clavicle, shoulder, and the affected arm were obtained to rule out fractures of the clavicle and the humerus. Scanograms of the injured and normal extremities were obtained on a preoperative basis from the acromion to the fingertips to measure existing bone length discrepancies and to document the effect of reconstruction on bone growth. Electromyographic evaluation including nerve conduction velocities, needle electromyograms (EMGs), and the lamina test in cooperating older children was obtained at the first office visit, unless the child was less than 3 weeks old. Results from the EMGs were tabulated and correlated statistically with the intraoperative diagnosis to calculate the sensitivity and specificity of electromyography to diagnose accurately avulsed roots or ruptured roots. Follow-up visits included repeated electromyographic evaluations and repeated scanograms (every 6 months) of the involved and normal extremities to calculate the bone discrepancy. On an annual basis, computed tomography (CT) scans of bilateral upper extremities including bilateral glenohumeral joints were performed.
Intravenous sedation was usually required to complete this study and obtain adequate pictures in most of the babies. When spinal nerves are either avulsed or damaged close to the spinal cord, pseudomeningoceles formed or poor nerve root filling was demonstrated. The findings of plain myelography were verified with a CT scan of the cervical spine, which was obtained at the same time. The data collected from all the patients (number of avulsed roots, number of abnormal roots, etc.) were correlated statistically with the intraoperative diagnosis to study the sensitivity, specificity, and positive predictive value of this neuroradiological intervention.
Plain inspiratory and expiratory chest x-rays or fluoroscopy was used to assess paralysis of the ipsilateral diaphragm. The diaphragm was found to be paralyzed in 3.38% of the studies performed and this was correlated 100% with the intraoperative diagnosis.
In suspected upper obstetrical brachial plexus palsies, the supraclavicular plexus was primarily explored. In patients with suspected global paralysis the entire plexus (supraclavicular and infraclavicular) is explored as in most of these cases, the avulsed roots assume a retroclavicular position. In cases of multiple avulsions, search is instituted to identify possible intraplexus and extraplexus motor donors. Special considerations include the use of light anesthetics such as sufentanil and avoidance of any kind of muscle relaxants so that intraoperative neurophysiologic testing and diagnosis can take place. The incision used for suspected upper plexus lesions (C5/C6 and C5/C6/C7 involvement) is usually a V-shaped supraclavicular incision placed along the posterior border of the sternocleidomastoid muscle and parallel to the clavicle. In suspected global palsy the supraclavicular incision is extended to the infraclavicular region, over the border of the clavicle and along the deltopectoral groove. We do not sacrifice the clavicle to gain better exposure; instead we explore the neural structures above and below the clavicle as needed. The omohyoid muscle, cephalic vein, and transverse cervical vessels are identified and preserved. The phrenic nerve is identified in its usual anatomic position on the anterior surface of the anterior scalene muscle, traced superiorly to its C4 and C5 origin, and stimulated to assess its integrity and for consideration as a possible motor donor. All identified nerve structures are stimulated with a DC stimulator at 0.5, 1.0, and 2.0 mA to assess nerve responsiveness to different current intensities and to verify continuity with the respective muscle targets.
In obstetrical brachial plexus injuries, depending on the intensity and direction of the forces involved, neuromas are usually located in a supraclavicular or retroclavicular position. Neuromas can be found at the root, trunk, or cord level. If intraoperative stimulation of the neural tissue proximal to the neuroma yields minimal response, this indicates a neuroma, which is not in continuity; this can then be resected proximally and distally until normal tissue is encountered and healthy fascicles are visualized. Biopsies of the neuroma itself and immediate proximal and distal to the neuroma nerve cross sections are obtained at that time for carbonic anhydrase staining,14,15 which determines the connectivity of the root to the spinal cord and its sensory versus motor content. Root biopsies rich in ganglion cells are indicative of avulsion and are not used for reconstruction.
In early upper plexus lesions (3 months or less), if the shoulder and elbow are still paralyzed at 3 months, the identified neuroma is usually minimally responsive to intraoperative electrical stimulation of the more proximal roots. In these cases the neuroma should be resected and the continuity restored with interposition nerve grafts. In cases that present with DT greater than 3 months the management of the neuroma in continuity16,17 is more complicated as spontaneous regeneration may have progressed to a point that several muscle targets may respond to intraoperative electrical stimulation. In these cases the surgeon has to make a decision on the basis of his or her experience, the clinical presentation, and the risk of possibly downgrading a certain function if the neuroma is resected. If the infant has some function but this function is judged to be inadequate or useless, temporary loss of function may be preferable in anticipation of much greater function following reconstruction. Such a decision necessitates in-depth consultation with the infant's parents. If the parents are not comfortable with a temporary loss of a function, the surgeon needs to recruit alternative strategies, such as various neurotizations of the weak targets and avoidance of resection of the neuroma in continuity.
Following identification of the lesions and resection of neuromas, intraplexus and/or extraplexus motor donors are identified. Whenever possible, ipsilateral intraplexus motors are preferred for upper extremity reconstruction as they yield better results. Extraplexus donors such as intercostal nerves are used with caution in infants, especially if there is concern about the integrity of the phrenic nerve. The spinal accessory nerve is frequently used to reconstruct the suprascapular (SS) nerve, preferably with direct neurotization. At times contralateral motor donors (anterior and posterior divisions of C7) have been used successfully. In general, one should always seek ipsilateral intraplexus donors if available, as the number of fibers is much greater than the various extraplexus donors. Intraplexus donors were used in all the patients. The C5 root was the most frequently used motor donor (n=110) followed by the C6 root (n=49), following resection of an upper trunk neuroma, mostly to reconstruct the posterior cord (PC) or the lateral cord (LC). The C7 root was used also routinely (n=37) to reconstruct supraclavicular plexus elements, most commonly via interposition nerve grafting. From the extraplexus motor donors the most frequently used were the accessory nerve (n=33) and the intercostal nerves (n=35). The target nerves also varied, with the most common nerve targeted being the SS nerve (n=55) followed by the PC (n=39) and the LC (n=33). The most common neurotization performed was the reconstruction of the SS nerve from the XI (n=24) followed by the reconstruction of the same nerve (SS) from the C5 root (n=20). The entire list of the different neurotizations performed is depicted in Table Table11.
The senior surgeon (JKT) introduced an arbitrary scoring system, which she used for many years to define the severity of brachial plexus lesions. The total severity score used is the summation of the severity score of each root, according to the following scoring system (from the worst lesion to the best): avulsion=0, rupture/avulsion=1, rupture=2, rupture/traction=3, traction=4, normal=5. A normal brachial plexus has a severity score of 25 (never encountered) and a globally avulsed plexus would be given a severity score of 0 (never encountered). A total of 74 out of the 77 brachial plexuses that had primary reconstruction were explored. In three of the patients the DT was greater than 3 years and the brachial plexus was reconstructed with neurotizations, without, however, primary exploration. The mean severity score was 8.70 with a range from 2 to 22. The lower the severity score, the greater the number of avulsed roots present. In this population of patients the fact that the mean severity score was lower than one half of the normal indicates a prevalence of devastating global lesions. The distribution of the 74 explored brachial plexuses in relation to the severity score groups is depicted in Figure Figure1.1. Group A consists of all brachial plexuses with severity score between 0 to 5, group B from 6 to 10, group C from 11 to 15, and group D from 16 and greater. It is apparent that the first two groups included 46 brachial plexuses or 62.2% of all the plexuses explored, indicating a higher number of avulsions, therefore a more devastating and challenging to repair injury. Relevant to this is also the distribution of the explored brachial plexuses in relation to the number of avulsed roots present per plexus (Fig. 2). Two avulsions per plexus was the most common injury pattern (24.32%).
Of interest is the frequency of the different root injury types, that is, rupture, avulsion, and traction. Out of the 370 roots explored (74 brachial plexuses × 5 roots per plexus), a total of 122 roots were ruptured, rendering this type of injury the most common diagnosis (32.97%), followed very closely by the number of avulsions (119 total avulsed roots or 32.16%). The total number of normal roots diagnosed was only 31 or 8.37% of the total root population.
We also studied the distribution of the different types of injury in relation to the different brachial plexus nerve root levels (Fig. 3). From the analyzed data it is noted that ruptures were more common in the upper plexus roots; C5/C6 ruptures equaled 100 ruptured roots or 81.9% of all ruptures, whereas the number of ruptures in the lower roots was significantly lower (p<0.001); C8/T1 ruptures equaled only 7 ruptured roots or 5.73% of all ruptures. On the contrary, avulsed roots were more prevalent at the lower plexus; C8/T1 avulsions equaled 61 avulsed roots or 51.26% of all avulsed roots, whereas C5/C6 avulsions equaled 25 avulsed roots or 21% of all avulsed roots (p<0.001). The roots where avulsion was most commonly diagnosed were the C7 and C8 roots (33 each or 27.73% per root). The root where rupture was most commonly diagnosed was the C5 root (62 or 50.8% of all ruptures). Of great interest is the fact that normal roots were never encountered at the upper plexus level (C5/C6). This pattern of injury distribution—upper roots are most likely ruptured and lower roots are most likely avulsed—could be explained by the stronger connective tissue sheath surrounding the upper roots. The five different types of injury that one root could sustain (rupture, avulsion, traction, rupture/traction, and rupture/avulsion) resulted in a variety of different combinations of brachial plexus injury. Forty-eight different injury patterns were surgically diagnosed. The most common pattern was that of C5 rupture and C6-to-T1 avulsion (n=7), followed by C5C6 rupture and C7-to-T1 avulsion (n=5).
The reconstruction of the brachial plexus included a combination of different techniques (Table 2) including microneurolysis, direct end-to-end repairs, direct end-to-side repairs, indirect neurotizations, and in a few occasions direct nerve-to-muscle neurotizations. A total of 483 nerve repairs were performed in 75 patients with 77 brachial plexus injuries. Microneurolysis was performed in 38 patients, especially when a root or cord was diagnosed as tractioned. Totally, 132 nerves were microneurolyzed. The palpated elements were usually hard and following longitudinal epineurial release the bulging of the compressed fascicles was an indication of the necessity of this type of reconstruction. Direct nerve coaptation by an end-to-end repair (n=81) was performed more frequently than by end-to-side coaptation (n=14) and involved primarily nerve elements distal to the root level. Direct coaptation of nerve roots following excision of neuromas is rarely possible and not indicated. In selected cases when the C7 root escaped injury, we have used the anterior and posterior divisions of the C7 for direct coaptation with neighboring more important plexus elements, such as the proximal posterior and lateral cord or C8 and T1 roots in global palsy cases. However, in the majority of cases, following scar or neuroma excision, interposition nerve grafting was indicated. Indirect neurotization with interposition nerve grafting was used in all but one of the cases in which primary reconstruction took place (n=76 brachial plexus reconstructions with interposition nerve grafting or 97.4% of patients undergoing primary reconstruction of the brachial plexus). Totally, 252 neurotizations or nerve repairs with interposition nerve grafting were performed. The ratio of indirect to direct neurotizations was 2.65 indirect neurotizations to 1 direct neurotization. When performing interposition nerve grafting, care was taken to position the grafts on an underlying vascularized bed.
The nerve grafts harvested included the sural nerves (n=150 nerve grafts in 75 patients), the saphenous nerves (n=37 nerve grafts in 19 patients), and the MAC nerve (n=7 grafts in 7 patients). The ulnar nerve was used once as a pedicled vascularized nerve graft, in a patient with global palsy and DT greater than 2 years. Sural nerves were usually harvested through a longitudinal incision on the posterolateral aspect of the calf, even though harvesting has been reported endoscopically in older patients.18 The saphenous nerves were harvested through a medial thigh incision and were routinely used as cross-chest nerve grafts, especially when the contralateral C7 root was used as a motor donor. All dissections and explorations were performed under loop magnification and all coaptations and neurotizations were performed under the operating microscope. Vascularized sural or saphenous nerve grafts were not used because the gaps to be bridged with the interposition nerve grafts were relatively short, as compared with the adult population.
From the analyzed surgical data it is noted that a total of 583 interposition nerve grafts (cables) have been used. The average number of grafts per patient was 7.77 and the average length of each nerve graft was 4.8 cm. In addition, 71 nerve grafts were coapted with proximal motor donors and “banked” in the subcutaneous tissue of the arm, near the elbow, to innervate future free muscles, mainly for reanimation of the hand. An average of 2.02 banked nerves per patient were placed. Furthermore, a total of 56 cross-chest nerve grafts were used in 28.57% of the patients with primary reconstruction (3.5 cross-chest nerve grafts per patient).
We consider stability of the shoulder to be the most important goal.19 Therefore we always attempt reconstruction of the supraspinatus and deltoid muscles through reinnervation of the axillary nerve through elements of the proximal upper trunk or direct neurotization of the SS nerve from the ipsilateral accessory. Unless elbow flexion is restored, the normally functioning hands in Erb's palsy are useless. Therefore biceps function is of great importance and during surgery we try to provide the musculocutaneous nerve with the best proximal motor donors, usually those that sustained the least damage. If the C6 root has escaped injury and is judged to be adequate for neurotization purposes, it is used to obtain motor fibers for the neurotization of the median cord (MC) in the majority of the cases via interposition nerve grafts. If the C6 is avulsed but the neighboring C7 is judged to be in continuity and usable, the anterior division of C7 is transferred to the LC directly by end-to-end repair. Occasionally the MC was neurotized from the contralateral anterior division of the C7 root via saphenous or sural cross-chest interposition nerve grafts, especially in cases in which the ipsilateral C5, C6, and C7 roots could not be used. A common extraplexus motor donor to the MC also included intercostal nerves, especially if the quality of the upper plexus roots was questionable. Depending on the availability of motor donors, reinnervation of the triceps was the next priority and in most of the cases it was accomplished with reinnervation of the PC with interposition nerve grafts from intraplexus donors or from the ipsilateral intercostals.
What is unique in global palsies is the involvement of the hand in the injury.19 In neonates there is greater potential for hand reinnervation, as compared with adults, because the distances from the roots to the target muscles of the hand are shorter. The usual scenario in global paralysis is rupture of the upper roots with avulsion of the lower plexus. In these cases, the best root is the one farthest away from the injury, which is the C5 root. The reconstructive strategy here is to neurotize the two lower avulsed roots from the C5 root and to use the C6 root for reconstruction of the lateral and posterior cords. If C7 has escaped the injury, which indeed is a rare occurrence, then C7 motor fibers should be used to neurotize the C8 and T1 roots directly by end-to-end repair because of the proximity of the involved structures. This strategy allows multiple powerful axons to invade the lower roots and enhances the chances for a better recovery.
In cases of multiple avulsions, reconstruction of the median, ulnar, and radial nerve has also been achieved in 10 cases from the contralateral C7, anterior and posterior division, through saphenous or sural interposition nerve grafts. In cases with longer DTs, during the first-stage reconstruction, nerve grafts are banked at the elbow level so that they can be used later in a second-stage reconstruction for the innervation of free muscles transferred for hand reanimation. In global palsies the median nerve also needs to be restored both for sensory protection and for finger and wrist flexion. Supraclavicular sensory nerves and ipsilateral sensory intercostals have been used for this purpose. A total of 13 nerve transfers for direct sensory reinnervation have been performed.
Again, stability of the shoulder remains of great importance; therefore the reinnervation of both the supraspinatus and deltoid muscles needs to be addressed. Definitely, elbow flexion is of great importance, so if the injured hand has restored function, the child will be able to bring it to the mouth level. Thus, in global paralysis reinnervation of the lateral and posterior cords is pursued preferably from intraplexus remaining donors, such as the C6 root, after reanimation of the hand has been addressed. The SS nerve is also addressed during the primary reconstruction of the plexus, by direct neurotization from the distal ipsilateral accessory nerve.
These procedures were reserved for patients who already had brachial plexus reconstruction and aimed to improve overall function. They are also of great importance in late nonoperated obstetric brachial plexus palsy patients with long DTs and with established deformities, who were not good candidates for brachial plexus reconstruction. In our population of patients, 24 patients were categorized as late cases. All patients had adequate follow-up (at least 3 months) and thus they were included in this outcome study. These patients received a combination of pedicled muscle transfers, free muscle transfers, tendon transfers, rotational osteotomies, and other secondary procedures. During muscle or tendon transfers the decision is made to “weaken” one function to replace or enhance another critically needed function. The muscles or tendons to be transferred need to be adequately innervated and were selected by clinical examination and needle EMGs.
The most frequently performed pedicled muscle transfer (Table 3) was the rerouting of the proximal latissimus dorsi (n=49) and the teres major (n=47) around the humerus to restore dynamic external rotation. Common muscle transfers also included the transfer of the distal trapezius with or without fascia lata to the lateral humerus to enhance abduction (n=9) and transfer of the pectoralis minor to enhance elbow flexion (n=7). At the elbow level, the transfer of the ipsilateral latissimus dorsi muscle flap to substitute for the biceps as a myocutaneous pedicled flap was used to restore elbow flexion (n=4). The transferred latissimus was either free of injury or was reinnervated during the first stage with ipsilateral intercostals. A total of 135 pedicle muscle transfers were performed.
Free muscle flaps were also used as second-line transfers (Table 4). A common procedure in this category was the transfer of the gracilis muscles to the forearm for finger flexion (n=13) and finger extension (n=14). In four cases the gracilis and the rectus femoris muscle were used to substitute for intrinsic function combined with the “lasso” procedure at the level of the MP joints. The motor donors used to innervate banked nerves for future innervation of free muscle transfers included the ipsilateral intercostal nerves (n=26), which were the most common motor donors, with the motor fibers of the contralateral C7 root being second (n=14). A total of 48 free muscle transfers were performed.
Tendon transfers in the hand region included a variety of transfers (Table 5) of which the most common was the transfer of the flexor carpi ulnaris (FCU) to extensor digitorum communis (EDC) for restoration of finger extension (n=11) and the transfer of the brachioradialis for thumb extension (n=11). The most common tendon transferred regardless of the targeted function was the FCU (n=22) and the most common function restored or enhanced regardless of the tendon donor was thumb extension (n=25) followed by finger extension (n=19). A total of 80 tendon transfers were performed in this series.
If there were no available muscles to restore external rotation of the shoulder, a rotational osteotomy of the humerus was employed to improve external rotation by 30 to 40 degrees and was usually done at the midhumerus level (n=3). Rotational or wedged osteotomies of the radius were also performed in 10 cases to correct supination or pronation deformities.
To correct discrepancies of the length of the humerus, we have performed a transverse cortical osteotomy and applied an external fixator to initiate distraction osteogenesis20,21,22 in two cases. In these two patients (both were late cases with severe limb discrepancy), who have undergone this procedure, the achieved limb elongation was 5 cm and 10 cm, respectively. Radiographic evidence in these cases showed that the reinnervated humeral bone healed very well. Of importance also is to mention that distraction, which was performed several years after plexus reconstruction, did not affect the reinnervated muscle or nerve function. Additional secondary reconstruction procedures included correction of winging of the scapula with transfer of the contralateral trapezius and/or rhomboids to the medial border and inferior angle of the affected scapula in 26 patients.23
The most common complication encountered was wound infection. Free flap necrosis occurred on one occasion (1 out of 48 free muscle transfers or 2.1%). Vascular anastomosis compromise that necessitated reexploration of the anastomotic site occurred on three occasions (3 out of 96 anastomosis or 2.97%). Two of the reexplorations revealed patency of the anastomotic site and the compromise was attributed to arterial spasm that was present at the time of reexploration. One reexploration revealed thrombosis of the venous anastomosis and required thrombectomy and irrigation of the free flap with heparin solution to salvage the flap.
To evaluate better the outcome of brachial plexus reconstruction in our population of patients we divided our patients into primary reconstruction patients and secondary reconstruction patients. We also divided them into Erb's palsy patients and global palsy patients. Then the patients in each group were divided into four groups according to the DT. Group1 included patients with DT between 0 and 3 months, group 2 included patients with DT between 4 and 6 months, group 3 between 7 months and 1 year, and group 4 included patients with DT higher than 1 year. The evaluation was performed by three independent reviewers who did not participate in the surgeries to obtained unbiased results. The evaluators studied charts, preoperative and postoperative standardized videos, colored slides, and black-and-white photographs.
Statistical analysis of the results was performed by a senior full-time professor of statistics at the Old Dominion University. Statistical methods used included Student's t-test, paired t-test, analysis of variance (ANOVA), regression analysis, and other parametric and nonparametric methods.
Comparison was done between preoperative and postoperative EMGs to assess the improvement in muscle function in the primary reconstruction population. The same neurophysiologist, who has been affiliated with our center for many years, executed all EMGs. In the primary reconstruction population we elected to compare EMG results in 12 representative muscles that have been routinely reinnervated (supraspinatus, infraspinatus, deltoid, pectoralis major, latissimus dorsi, biceps, triceps, FCU, flexor carpi radialis [FCR], extensor carpi radialis longus [ECRL], flexor digitorum sublimis [FDS], EDC).
To be able to analyze statistically the EMG data, we used an arbitrary system for muscle grading based on the electromyographic performance of the muscles tested. The grading system grades muscle from 0 to 3. A value of 0 was given for muscles that did not generate any action potentials at all (no electrical activity), a grade of 1 was given to muscles that generated poor action potentials, a grade of 2 was given to muscles generating moderate electrical activity, and a grade of 3 was given to muscles with full electrogenesis, that is, normal electrical activity. The data were analyzed by paired t-test and are presented as average preoperative and postoperative EMG values for each muscle, accompanied by the respective p value.
In the primary reconstruction population (Table 6) statistical significant improvement by EMG testing was noticed for all major muscles tested. The p values for all muscles were less than 0.05. The highest statistically significant improvement was noticed for the SS, infraspinatus, deltoid, biceps, and ECRL (p<0.0001) and the lowest statistically significant improvement was achieved for the pectoralis major (p=0.0388).
The postoperative functional assessment for six major targets (latissimus dorsi, supraspinatus, infraspinatus, deltoid, biceps, and triceps muscles) was graded according to the following scale to express the outcome: a grade of M0 to M2 constituted a poor result (group A), M2+ to M3 was fair (group B), M3+ to M4− was good (group C), and M4 or M4+ was excellent (group D). Comparison of the data from preoperative muscle functional assessment with postoperative data shows improvement in all muscles tested at a statistically significant level (p<0.001) except for the latissimus dorsi muscle (p>0.05). The results were excellent in 56.97% of biceps restoration, 44.82% of supraspinatus restoration, 40.96% of deltoid restoration, 34.14% of infraspinatus reconstruction, and 30.9% of triceps restoration. Results were less rewarding for latissimus dorsi restoration, as the improvement at the excellent level was only 32.75% from 28.9%. However, if one combines the “excellent” group (group D) and the “good” group (group C), which combined reflect muscle power from M3+ and above, the results are more impressive for all muscles tested, including the latissimus dorsi muscle (Fig. 4).
Statistical analysis of the results of neural microsurgery in this population of patients was performed using data from the preoperative and postoperative muscle grading per muscle target, without collapsing the data into the functional scale mentioned earlier. The preoperative and postoperative muscle grading data were treated as independent groups for each muscle target (11 muscle targets were tested). These variables were analyzed with paired t-tests. The postoperative values for all variables tested were significantly higher than the preoperative values (p<0.001) (Fig. 5).
With the same statistical method (paired t-tests) we analyzed the effects of primary reconstruction on improving functional outcome as assessed by the Mallet scale grading system (Fig. 6) and the improvement of the degrees of excursion in the shoulder and elbow joints. All functions of the Mallet scale have improved after primary reconstruction at a very significant level (p<0.0001). The mean grade for four out of six functions was postoperatively above 3. Preoperatively, the mean grade for all functions was below 2.
To assess the effects of primary reconstruction on hand function we divided the patients who underwent primary reconstruction into four distinct groups: group A included patients with only C5C6 lesions, group B included patients with C5C6C7 lesions, group C included the patients of groups A and B combined, and group D included patients with global palsy (C5 to T1 lesions). Comparison between preoperative and postoperative Gilbert-Raimondi scores was also done for the entire primary reconstruction population (group E). In all groups there was improvement in hand function after reconstruction in a statistically significant manner. The biggest improvement was achieved for patients with global palsy (from a preoperative mean of 1.38 to a postoperative mean of 3.38, p<0.0001). The analysis for this population's data was done by paired t-test. Data were analyzed by repeated measures ANOVA comparing C567 and C56 lesions pre- to postoperatively. There was a significant increase in Raimondi scores from pre- to postoperatively for all patients in groups A and B combined, that is, group C (4.69±0.95 to 5.75±0.45, p=0.0005), but there was no difference in change in scores from pre- to postoperatively between group B (C567 lesions) and group A (C56 lesions) (p=0.8591).
With the same statistical method as in primary reconstruction (paired t-tests) we analyzed the effects of secondary reconstruction on improving functional outcome as assessed by the Mallet scale grading system and the improvement of the degrees of excursion in the shoulder and elbow joints. In all functions of the Mallet scale grading improved significantly after secondary reconstruction. With the exception of the “hand to back” function, the average postoperative grading was well above 3 (p<0.001 for five functions). The hand to back function also improved but not as impressively (p=0.0147). The positive effect of secondary reconstruction was also proved by the improvement in the excursion in both the shoulder and elbow joints. The average postoperative shoulder abduction increased from 51.5 to 99.8 degrees (p<0.0001), the average external rotation increased from 4.1 to 47.2 degrees (p<0.0001), and the elbow flexion from an average of 131.3 to 158.2 degrees. Good results were also noted in anterior flexion restoration (p=0.0165). The only function that did not improve statistically was that of elbow extension (p=0.0716). That may be due to the small amount of secondary procedures performed for this reason (n=9) and the fact that at the time of surgery many patients had already established elbow flexion contractures.
To assess the effects of secondary reconstruction on hand function we divided the patients who underwent secondary reconstruction (n=24) into three distinct groups: group A included patients with Erb's palsy and spared hand function, group B included patients with global palsy, and group C included patients with Erb's palsy and some hand dysfunction. Comparison between preoperative and postoperative Gilbert-Raimondi scores was also done for the whole secondary reconstruction population (group D). Data were analyzed by repeated measures ANOVA comparing global and Erb's palsy groups pre- and postoperatively. The main finding was that there was a statistically significant increase in Gilbert-Raimondi scores from pre- to postoperatively for global palsy patients (2.14±1.77 to 3.71±1.38, p<0.05), but the change in scores from pre- to postoperatively for Erb's palsy was not significant (5.61±0.70 to 5.89±0.32, p>0.05). Statistical improvement was also achieved for all patients in group C (Erb's with some hand dysfunction), in which there was a significant increase in Raimondi scores from pre- to postoperatively (2.80±1.81 to 4.22±1.56, p=0.0417). Overall, in the entire population that underwent only secondary reconstruction (group D, n=24), the improvement in hand function was not statistically important (p>0.05), despite the fact that the average Gilbert-Raimondi score improved from 4.625 preoperatively to 5.083 postoperatively. These results can be explained when comparison is made between groups. Overall Erb's scores (group A) were significantly higher (p<0.0001) than global scores (group B), both preoperatively and postoperatively, as expected. ANOVA was also used to compare global paralysis patients (group B) and Erb's palsy patients with some hand dysfunction (group C) pre- and postoperatively and group C was found to have significantly greater Gilbert-Raimondi scores both pre- and postoperatively than global patients (p=0.0270). In conclusion, the improvement in hand function achieved in group B and group C proves that local tendon transfers can enhance or restore hand function in late cases.
DT was also an important factor in determining final outcome. All patients who underwent neurotizations (i.e., primary reconstruction patients) were divided into four groups: group A had a DT of 0 to 3 months, group B had a DT of 4 to 6 months, group C had a DT of 6 to 12 months, and group D had a DT above 1 year. Pre-and postoperative comparisons were made with paired t-tests for each group. Differences between denervation groups postoperatively were compared with one-way analysis of variance. Protective sensation data were analyzed by chi-square.
In a general statement, the average mean grades for most of the variables tested were higher for group A (0–3 months) than group B (4–6 months) and between those two groups and the longer DT groups C and D. However, it has to be noticed that between groups A and B many postoperative values were not statistically significantly different (p>0.05) because of lack of power between the two groups (group A included 8 patients, group B included 18 patients, group C included 7 patients, and group D included 29 patients). Function of the following muscles was statistically better in group A than group B: FCU (p<0.001), FCR (p<0.001), FDP and FDS (p<0.001). Even if the average postoperative muscle grading for the deltoid and biceps muscles was higher in group A (3.8±0.7 for both muscles) than in group B (3.4±0.5 and 3.6±0.5), the difference was not statistically significant. These results show that the improvement in the more proximal muscles of the shoulder and elbow region was not much different in the early group A than in groups B or C. However, greater improvement was observed for the more distal hand muscles in group A than in group B.
The effects of DT on the Gilbert-Raimondi score were also studied. In all four DT groups, statistically significant improvement was noticed after surgical reconstruction. Comparison between groups A through D in relation to the Gilbert-Raimondi score showed slightly better values for the earlier DT groups than the later DT groups. The average postoperative score for group A was 4.6 (±2.1), group B 4.4 (±1.3), group C 4.4 (±1.1), and group D 3.8 (±1.9). However, this difference among the four groups was not statistically significant (p>0.05)
Postoperative shoulder abduction as assessed with the Mallet scale was significantly less for group D than for groups A and B postoperatively (but not group C) (p<0.001), and the average postoperative degrees of shoulder abduction were also significantly less for group D than for A and B (but not C) (p<0.001). This possibly reflected the trend for greater grading achieved for the deltoid muscle for the earlier DT time groups. For the rest of the Mallet scale functions no statistical difference was found among the four denervation groups. Comparison of sensation restoration among the four DT groups showed no difference among the groups.
The mean number of surgeries per denervation group was calculated and compared between groups (Fig. 7). The mean number of surgeries ±1 standard deviation for group A was 1.3 (±0.5), group B 3.1 (±1.7), group C 2.1 (±1.6), and group D 2.9 (±1.2). The mean number of surgeries was significantly less than for groups B and D. No other statistical differences among groups were observed. This is extremely important because it indicates that in patients with DT of 3 months or less, a minimum of 1.3 surgeries per patient is needed to complete reconstruction and achieve acceptable outcomes. This number increases threefold to approximately three surgeries per patient for longer denervation groups.
The debate whether and when to correct OBPP lesions surgically is still active. Also active is the discussion concerning evaluation of patients prior to surgery with CT-myelograms and EMGs and which grading scales should be used to assess better the function of the upper extremity. In addition to preoperative radiological evaluation, EMG and nerve conduction studies have been for years great tools in assisting in diagnosis and follow-up. In our study the effects of primary reconstruction on the electromyographic recovery of the reinnervated muscles have been positive. Despite claims24 that EMG and other neurophysiologic assessment methods have a high false-positive rate in the pediatric population compared with adults with brachial plexus injuries, which can suggest a falsely optimistic prognosis, we believe that, especially in the initial diagnosis, EMG is an invaluable asset.
The debate is also productive in relation to the best clinical evaluation scoring system that should be used. As with most reconstructive microsurgeons, we prefer to use the classic Mallet scale and the modified Gilbert-Raimondi scale so that we can compare our results with previously published outcome studies. However, one should be aware of other scoring systems, designed to measure active limb movements and compare them with the normal side to obtain a ratio, which is then converted to a score17,25 that may be more effective in assessing functional outcome.
Since the revival of microsurgical reconstruction of the brachial plexus in the early 1970s, a general rule is that these babies without normal deltoid muscle or biceps by 3 months should not expect a good outcome. In a thesis presented by Tassin, the author followed the spontaneous recovery of shoulder function of 44 babies over a period of 5 years. By using the Mallet score he determined that a Mallet grade IV shoulder (good shoulder) was not obtained unless contractions of the deltoid and biceps started by 3 months of age. Complete recovery could be expected only if these muscles started contracting by the first month. If contractions started after 3 months, the spontaneous recovery of the shoulder function was less than average (Mallet grade III).
Gilbert proposed three indications for surgery. The first indication is complete palsy with flail arm and Horner's. He claimed that the parents should be informed early about the likely incomplete spontaneous outcome. He stated that the infant should be operated by 3 months. The second indication is C5 and C6 palsy with no contraction of biceps or deltoid by 3 months. The third is C5 and C6 palsy with no recovery of biceps to a 3+ level by 3 months.26
In our institution, indications for surgery include (1) global paralysis, for which we advocate surgery at 3 months, and (2) M0 function of the biceps or deltoid by 3 months. In this outcome study it is concluded that, on the basis of the preceding criteria, in both the Erb's palsy group and the global palsy group, near-normal function of deltoid (M3+ and above) was achieved when the infants were operated before the third month of age. The chances for functional return in the hand were also greater if the 3-month frame was followed. Furthermore, it was noted that in cases with early brachial plexus surgery, the need for secondary reconstruction and muscle transfers in general was minimized (1.3 surgeries per patient for group A versus 3.1 surgeries for group B).
Zancolli, on the contrary, delays the decision for surgery until 6 to 8 months. He believes that the absence of clinical or electrical signs of recovery in the biceps and deltoid for upper plexus lesions, or the triceps for middle plexus lesions, or no improvement of the function of these muscles beyond the M2 value indicates poor recovery and is an indication for reconstruction.27 In general, most authors accept that upper plexus palsy with phrenic nerve involvement should urge the surgeon to perform early reconstruction. Other criteria indicating an unfavorable prognosis and supporting the indication for early surgical treatment are persistent sensory disturbances with skin ulcers and persistent denervation with absence of action potentials in serial EMGs.
Laurent et al28 have reported useful arm function with attempted repair as late as 24 months. They believe that the very well documented late consequences of obstetrical brachial plexus paralysis, such as internal rotation contractures, hypoplasia of the arm, elbow flexion contractures, and dislocations of the humeral head, can be prevented even if the surgical intervention takes place between 12 and 24 months of age. In most centers surgeons prefer to gather information from myelography and electrophysiological studies before proceeding to a major reconstructive effort. Some spontaneous improvement in the first 3 months is also considered seriously when formulating the appropriate reconstructive plan. Narakas29,30 would operate between 6 weeks and 2 months if signs of root avulsion are suspected, such as Horner's syndrome; electrophysiologic loss of serratus, rhomboid, and pectoral muscles; elevated diaphragm by fluoroscopy; negative sensory evoked potentials; and fractures of transverse processes. Narakas also believed that in high lesions early repair is of great importance for an optimal result because of the distance required for nerve regeneration. Milessi,31 on the basis of clinical examination, believes that if an advancing Tinel sign stops moving distally, distal function returns before proximal, and there is no return of function by 6 months, the surgeon should explore the lesion and attempt to reconstruct the plexus. According to the same author, early exploration within the same time limit of 6 months should also be performed for infraclavicular lesions. Supporters of early intervention also include Xu et al,32 who advocate the 3-month rule for biceps recovery as their major criterion for primary reconstruction. On the contrary, Gilbert33 makes his decision when to operate based on spontaneous recovery by 3 months rather than trusting electromyographic findings, which may give a better picture than the clinical examination.
In our center we believe that several infants, especially with early use of electrical stimulation and behavioral manipulation, will improve substantially beyond the age of 3 months. However, we cannot ignore the fact that less favorable results are encountered if the surgical correction is delayed. In this outcome study among 99 patients who had been operated for obstetrical brachial plexus injuries, those with DTs longer than 3 months needed a secondary procedure, such as muscle transfers or rotational osteotomies or even34,35,36 wrist fusions. Moreover, in global paralysis cases, regardless of the pathology of the lesion, adequate hand function, as evaluated with the modified Gilbert-Raimondi scores,37,38 was restored in babies operated before 3 months. The average scores in these patients were higher than in longer DT groups, even if this is was a trend and not statistically significant. A statistically significant difference may have been achieved if the population in the early DT group was greater (group A included 8 patients and group B 28).
In general, regardless of the type of the injury and the repair, the postoperative muscle grading was significantly higher in each patient. That is why we advocate strongly that microsurgery is the mainstay of treatment for the majority of patients with OBPP. In general, the longer the DT the more secondary procedures were needed to complete the reanimation of the arm. Especially patients who were operated within 3 months (group A) did not need any secondary procedures to improve the function of their hand. All patients in the fourth group (DT >1 year) needed some sort of secondary reconstruction, minimizing the socioeconomic cost of these reconstructive procedures. That is why we strongly disagree with advocates39 of the “wait and decide” philosophy, who claim that functional outcome is not influenced if the DT is greater than 6 months or by the number of avulsions present. In addition to this, the early reconstruction assists in reversing developmental apraxia,40 another negative effect in patients with neonatal brachial plexus palsy that has been left untreated. Therefore, we firmly believe that, early before 3 months, surgical intervention should be attempted, especially in suspected global palsies.