Recovery patterns from C5 palsy after anterior cervical decompression and fusion, posterior cervical decompression and fusion, and laminoplasty for degenerative cervical myelopathy: systematic review and meta-analysis of 748 C5 palsy cases

Literature search

Five different databases (Google Scholar, Embase, PubMed, Web of Science and Cochrane Library) were thoroughly searched for relevant literature on October 15, 2024. Studies reporting on C5P published until 2024 were scrutinized. The search was performed using the following keywords in combination with Boolean operators: (((ACDF) OR (cervical fusion) OR (laminoplasty) OR (global fusion) OR (posterior decompression)) AND (C5 palsy)).

Inclusion or exclusion criteria

Studies that reported on the occurrence, management, or recovery of C5P following surgery for the management of degenerative cervical myelopathy were considered for inclusion. Studies reporting on C5P associated with procedures for cervical radiculopathies were excluded to make meaningful comparisons across the procedures performed for a given condition (i.e., myelopathy). Narrative or systematic reviews, opinions, letters to the editor, and manuscripts published in non-English languages were also excluded. The selection criteria are detailed in Table 1. To avoid heterogeneity, we only included cases explicitly diagnosed with C5P in the manuscripts following cervical spine surgeries based on a reduction in muscle power and only those who experienced complete recoveries during follow-up.

Table 1

Inclusion and exclusion criteria of selection of articles to be included in the review

Manuscript selection and data extraction

Search outputs were downloaded from the specific databases. They were then extracted to EndNote, de-duplicated, and then selected manually. Title screening was independently performed by two authors, after which individual manuscripts were separately screened as described in the aforementioned criteria. Thereafter, the next round of screening was performed, which involved extracting the complete manuscripts and detailed analysis in duplicate. The final selection of articles was then completed. Any discrepancy among the authors was clarified based on mutual discussions with the senior author.

Quality assessment

The included studies were assessed for quality based on the Newcastle Ottawa Scale for non-randomized studies, with the results being presented in Table 2 [5,6,10,13–39]. The studies demonstrated sufficient quality for inclusion.

Table 2

Quality assessment of the included studies

Statistical analysis

This meta-analysis was performed using Stata ver. 16.0 software (Stata Corp., College Station, TX, USA). Reported incidences in the included studies were pooled and calculated along with their 95% confidence intervals (CIs). The random-effects meta-analysis model was used for data synthesis when the studies had high heterogeneity (I²>50%; p<0.10); otherwise, the fixed-effects model was implemented. Sensitivity and subgroup analyses were performed if heterogeneity was noted among the reported results. Recovery according to reported follow-up time points were analyzed using Kaplan-Meyer analysis.

Time to C5P

A total of 30 studies including 21,231 patients with 748 C5P cases provided information on the time to C5P following surgery. To analyze this, we used the random-effects model considering the heterogeneity in the reported incidence across the included studies. The overall time to C5P reported in the included studies was 3 days (95% CI, 2.56–3.60). As shown in Fig. 2 [5,6,10,13–39], C5P occurred earlier following anterior cervical discectomy/decompression and fusion (ACDF) at 2 days (95% CI, 0.35–4.54), followed by laminoplasty (LP) at 3.2 days (95% CI, 2.02–4.34) and posterior cervical decompression and fusion (PCDF) at 3.6 days (95% CI, 2.81–4.37). However, no significant difference in the time to C5P was noted among the procedures as shown in Fig. 2 (p=0.520). Significant heterogeneity was noted among the included studies despite categorizing them based on procedure. We considered that the heterogeneity in the subgroups according to individual procedures could possibly be due to variations in patient population, surgical technique, and rehabilitation protocols followed in the individual studies.

Fig. 2

Forest plot showing the pooled time to C5 palsy following surgery for cervical myelopathy. CI, confidence interval; ACDF, anterior cervical decompression and fusion; PCDF, posterior cervical decompression and fusion.

C5P recovery

Recovery from C5P improved with time. The included studies reported recovery data at serial follow-up points, such as 3 months (four studies), 6 months (six studies), 12 months (10 studies), 18 months (three studies), 24 months (seven studies), and 36 months (nine studies). Kaplan-Meyer analysis of recovery at the reported follow-up timepoints is summarized in Fig. 3. Notably, recovery success rates were significantly higher with ACDF than with PCDF based on the reported recovery pattern at various timepoints. At the 1-year follow-up, the reported recovery rates were 100%, 50%, and 52.9% for ACDF, PCDF, and LP, respectively (p<0.001). We noted that recovery rates with PCDF improved from 27.9% at 3 months to 39.5% at 6 months and 50% at 12 months, further improving to 90.7% at the 2-year follow-up. Similarly, improved C5P recovery rates were observed with LP, with rates of 40%, 66%, 52.9%, and 100% having been observed at 3, 6, 12, and 18 months, respectively. C5P recovery rates for all the included procedures at all reported timepoints are presented in Fig. 4.

Fig. 3

Kaplan-Meyer recovery analysis of C5 palsy following various procedures for cervical myelopathy. CI, confidence interval; ACDF, anterior cervical decompression and fusion; LP, laminoplasty; PCDF, posterior cervical decompression and fusion.

Fig. 4

Reported C5 palsy recovery rates at various time points for various procedures for cervical myelopathy. ACDF, anterior cervical decompression and fusion; LP, laminoplasty; PCDF, posterior cervical decompression and fusion.

C5P prognosis

A study by Pennington et al. [5,6] found that at a mean final follow-up of 11.85 months, 40.7%, 56%, and 3.3% of patients experienced complete, partial, and no (or minimal) recovery of symptoms, with the mean time to recovery ranging from 5 to 45 weeks (mean of 35 weeks). Saadeh et al. [9], in their series of 38 patients, demonstrated that 52.8%, 34.7%, and 12.5% exhibited full, partial, and no recovery of useful strength in the antigravity musculature (shoulder abduction and elbow flexion), respectively, at the end of the 12th postoperative month.

Palsy resolution has also been correlated with the severity of the deficit [51]. Sakaura et al. [43] demonstrated that although 47.8% of mild deficits recovered within 3 months, 52% of severe C5P tended to persist for at least 6 months after surgery [52]. A study by Nassr et al. [47] reported a mean time to recovery of 20.9 weeks. Overall, studies have reported that 19.1% to 33% of patients are usually left with some degree of residual weakness [53,54]. However, studies have also shown that recovery, in specific situations, may even extend until 5 years following surgery [8,49]. No study has shown any significant association between intraoperative neuromonitoring signal changes and the onset, severity, and duration of recovery of C5P [38].

Time to C5P and recovery pattern

Overall, we observed that the mean time to C5P onset after surgery was 3 days. The pattern of onset varied between the different surgical procedures. While C5P developed the earliest following ACDF (after a mean duration of 2 days following surgery), it developed more gradually in patients undergoing LP and PDCF (at 3.2 and 3.6 days, respectively). In general, studies have reported relatively higher rates of C5P following posterior approaches, with relatively more delayed presentation. For instance, in the meta-analysis by Wang et al. [55,56], C5P was reported in 6.2% and 5% of patients undergoing posterior and anterior cervical spinal approaches, respectively. Moreover, a study by Lim et al. [49] showed no significant relationship between time to palsy onset (early versus delayed) and the overall duration of recovery. They observed that early- and delayed-onset palsy could indicate different underlying mechanisms. For instance, early-onset palsy could be attributed to a peripheral nerve injury (e.g., brachial plexus injury), whereas delayed palsy could indicate an underlying reperfusion injury.

In contrast, our findings showed that that improvement in neurological outcomes occurred with the passage of time. A majority of the reviewed studies reported recovery rates after 1 year following surgery. We observed a significant improvement in recovery rates among patients undergoing ACDF as the index surgery (recovery rate of 100%). For those undergoing posterior surgical approaches (LP and PDCF), however, the recovery rate approached only around 50% at the end of 1 year. However, considering the retrospective nature of the included studies, selection bias in reporting the recovered C5P cases could not be ruled out among the procedures analyzed. Hence, these results must be interpreted with caution before making any generalizable conclusions.

A previous study by Lim et al. [49] also demonstrated better outcomes following anterior cervical approaches, which they attributed to lower degrees of spinal cord shift, mitigated reperfusion-associated cord damage, and reduced tethering effect on the nerve root (compared to posterior surgeries) [57,58]. In addition, Blizzard et al. [59] observed that excessive restoration of cervical lordosis may be detrimental to C5P given that it could potentially cause greater posterior spinal cord shift and root tethering. Nevertheless, other studies have contradicted this observation and identified inadequate restoration of cervical lordosis or alignment as an inciting factor for C5P [13,39,58,60].

Predicting outcomes following the C5P

A study by Lubelski et al. [8] found that deltoid strength improvement was a key predictor of recovery from C5P. They observed that patients who experienced complete (60% of cases) or partial recovery (29% of cases) demonstrated improvement in motor power by at least one Medical Research Council (MRC) grade at around 6 weeks following the deficit. They emphasized the significance of examination at the 6-week timepoint to predict any meaningful recovery given that grade 4/5 or greater motor power at 6 weeks was predictive of complete recovery. Conversely, patients who demonstrated a motor power of 3/5 or below in their antigravity muscles alone were bound to only experience partial recovery. In short, the cohort with little or no recovery within 6 weeks following the onset of C5P was unlikely to experience good long-term outcomes. In addition, studies have reported that electrophysiological testing (electromyography) could be used as a means to identify patients unlikely to experience any meaningful recovery following postoperative C5P.

A study by Saadeh et al. [9] found that among the patients with severe C5P (defined based on antigravity strength of MRC grade ≤2) 3 months after surgery, 50% recovered useful strength by 12 months. Moreover, among those with persistently severe C5P at 6 months, only 25% recovered sufficient strength by 12 months. however, among those with motor strength of MRC grades 0 or 1 at 6 months, none had a useful strength at the end of 1 year. The mentioned study showed that while the female gender was associated with good recovery of useful strength, the presence of diabetes mellitus significantly impaired the final outcome.

A study by Lim et al. [49] involving 36 patients who developed C5P following cervical decompression surgery found that 50% of patients (among whom 91.7% recovered between 6 months and 2 years, whereas 8.3% did not recover until 2 years) required longer the 6 months to experience useful neurological recovery. The factors associated with prolonged recovery (>6 months) included motor grade ≤2 (p<0.001), multi-segment paresis involving segments apart from the C5 root (p=0.002), extent of posterior spinal cord shift (p=0.04), and the absence of somatic sensation with pain (p=0.008).

Pennington et al. [5,6] concluded that patients who underwent C4–5 foraminotomy had the greatest likelihood of developing a permanent C5 deficit (p=0.004). Among the other radiological parameters, mean cord-lamina angle (p=0.06) and length of laminectomy (p=0.08) showed a tendency toward significance but ultimately failed to attain the threshold for statistical significance (based on analysis of variance).

Factors underlying delayed or compromised neurological recovery

Hashimoto et al. [53] suggested that underlying asymptomatic damage to the anterior horn cells of the gray matter could promote severe postoperative C5P. It is well acknowledged that severe palsies have been associated with slow and poor neurological recovery. Multilevel, associated paresis has also been purported as a factor for poor C5P outcomes, which may be explained by focal reperfusion injury to the spinal cord following spinal decompression [13]. Similarly, significant sensory involvement (>50% involvement) and the presence of intractable pain can be indicators of serious spinal cord injury or substantial cord ischemia or reperfusion injuries, which in turn have been reported as poor prognostic indicators in patients with C5P.

Interventions for patients with poor spontaneous neurological recovery

Traditionally, patients with poor recovery after C5P have only been treated with supportive care. Alternatively, the use of steroids, C4–5 foraminotomy on the side of the palsy, and prophylactic foraminotomy at the index surgery have been utilized as treatment options [41,47,50,61]. Although some studies have concurred on the beneficial effects of corticosteroids in effectuating quicker recovery from the palsy, certain other studies found no significant relationship between steroid administration and recovery duration. Thus, the role of steroids still remains controversial [62,63].

Nerve transfers have been successfully applied to a growing number of indications, such as pre-ganglionic brachial plexus palsy, post-ganglionic brachial plexus palsy, nerve avulsion injuries, and spinal cord injury [8,50,64]. This procedure involves the use of an intact nerve with a duplicated function to restore the function of the injured nerve [65]. Nerve transfers have also been successfully performed in patients with C5P. However, the nerve transfer must be completed and the muscles reinnervated by the axons (i.e., neurotized) before the phenomenon of muscle fibrosis sets in, which roughly corresponds to 12 to 18 months following denervation [66,67]. Indeed, a study by Saadeh et al. [9] recommended such an intervention in patients who lacked useful motor strength (MRC grade ≤3) at the end of 3 postoperative months. Alternatively, studies have recommended a time window of 3 to 9 months following injury, considering that a period of several weeks to months is required for axons to grow from the site of nerve coaptation to the motor end plate at a typical growth rate of 1 mm/day [68,69].

Thompson et al. [50] has recommended surgical decompression (especially ipsilateral C4–5 foraminotomy) for patients with C5P not responding to conservative measures. However, another study by Lim et al. [49] demonstrated that surgical management was helpful only when definitive indications like fluid or seroma collection or compressive hematoma were identified. Therefore, at this point, no strong recommendation can be made in favor of surgical decompression for C5P unless a compelling indication suggestive of severe, residual spinal cord or nerve root compression is identified.

Limitations

Our study has some limitations that need to be acknowledged. Although this study has been the most comprehensive among all meta-analyses available on the topic to date, we did exclude non-English publications, which might have limited the inclusion of potentially relevant articles published in other languages. Furthermore, most of the studies included were retrospective in nature, thereby reducing the level of evidence for the conclusions derived from them. However, when analyzing complications arising from surgical interventions, one cannot simply rely on studies of higher quality, such as randomized controlled trials, relevant data are mostly reported and analyzed via retrospective case-control studies. We would like to acknowledge our failure to analyze all risk factors for C5P due to the heterogeneity in the reported risk factor data among the included studies. Furthermore, given the paucity and heterogeneity in reported data among the included studies, we failed to explore further causes of heterogeneity in time to palsy among the included studies apart from procedural categorization. Furthermore, we have excluded studies that reported C5P following procedures for cervical radiculopathies. Hence, caution must be experienced when applying the results of this study to conditions other than myelopathy.