Incidence of C5 nerve root palsy after cervical surgery

Abstract Purpose: We aim to perform a meta-analysis on incidence of C5 nerve root palsy (C5 palsy) for patients after cervical surgery. Methods: An extensive search of the literature was performed in PubMed/MEDLINE, Embase, the Cochrane library, CNKI, and WANFANG databases on incidence of C5 palsy from January 2007 to January 2017. Prevalence of C5 palsy related to different surgery methods was calculated and data analysis was conducted with STATA 12.0. Results: A total of 61 studies containing 721 patients with C5 palsy in total 11,481 patients (6.3%) were included in our study. The incidences after anterior cervical discectomy and fusion (ACDF), anterior cervical corpectomy and fusion (ACCF), anterior corpectomy combined with discectomy (ACCDF), laminoplasty (LP) and laminectomy and fusion (LF) were 5.5%, 7.5%, 6%, 4.4%, and 12.2%, respectively. Compared with anterior approaches (5%), female patients (4%) and patients with cervical spondylotic myelopathy (CSM) (4.8%), posterior approaches (6.2%), male patients (5.7%) and patients with ossification of posterior longitudinal ligament (OPLL) (8.1%) have a higher prevalence. In ACDF and LP, patients with OPLL (5.5%, 8.1%, respectively) have a higher incidence than those in patients with CSM (4.7%, 3.1%, respectively); however, in LF, patients with CSM and OPLL have similar incidence of C5 palsy (13% vs 13.1%). In most cases, C5 palsy was unilateral (74.5%). Conclusions: Based on our meta-analysis, posterior approaches, male patients and patients with OPLL have a higher incidence of C5 palsy. In ACDF and LP, patients with OPLL have a higher incidence of C5 palsy, but in LF, patients with CSM and OPLL have similar result.


Introduction
C5 nerve root palsy (C5 palsy) is a common complication after cervical surgery including anterior and posterior approaches: anterior cervical discectomy and fusion (ACDF), anterior cervical corpectomy and fusion (ACCF), anterior corpectomy combined with discectomy (ACCDF), laminoplasty (LP), and laminectomy and fusion (LF), which was reported first by Scoville. [1] Shou et al [2] reported that the incidence of C5 palsy was 5.3% (95% CI 4.6%-6.0%). Sakaura et al [3] showed that prevalence of C5 palsy varied from 0% to 30%. Patients with C5 palsy had paresis of the deltoid muscle and/or the biceps brachii muscle after surgery without any deterioration of myelopathic symptoms. [4,5] Previous studies that reported the posterior decompression were easier to cause C5 palsy compared with the anterior decompression. [6,7,8] But the reason remained controversial. Few hypotheses reported that the spinal cord or nerve root may lead to C5 palsy. Some believed that nerve root traction caused by the cord shifting resulted in C5 nerve root lesion after posterior decompression surgery, which is called "tethering phenomenon." [9][10][11] Another hypothesis was that spinal cord lesion caused by acute decompression and expansion of the spinal cord lead to C5 plasy. [12][13][14] Hence, it is necessary to review studies related to C5 palsy for concluding the incidence of C5 palsy after all kinds of surgeries and it can give some valuable comments for spinal surgeons.
Previous reviews [2,8,15] on the prevalence and risk factors of C5 palsy have few included studies or some included studies in the 1990s and 2000s, which was far from now. So, we performed a meta-analysis on incidence of C5 palsy for the last decade.

Ethics statement
There is no need to seek informed consent from patients, since this is a meta-analysis based on the published data, without any potential harm to the patients; this is approved by Ethics Committee of The Third Hospital of HeBei Medical University.

Search strategy
An extensive search of literature was performed in PubMed, Embase, the Cochrane library, CNKI, and WANFANG databases. The following key words were used for search: "C5 never root palsy," "cervical," "anterior cervical discectomy and fusion," "anterior cervical corpectomy and fusion," "corpectomy combined with discectomy," "laminoplasty," "laminectomy and fusion," "cervical spondylotic myelopathy," and "ossification of posterior longitudinal ligament" from January 2007 to January 2017, with various combinations of the operators "AND" and "OR." Language was restricted to Chinese and English.

Inclusion criteria
Studies were included if they met the following criteria: randomized or nonrandomized controlled study; age greater than or equal to 18 years old; studies on C5 palsy after cervical surgery.

Exclusion criteria
Studies were excluded if they met the following criteria: had repeated data; did not report outcomes of interest; in vitro human cadaveric biomechanical studies; earlier trial, reviews, and casereports.

Selection of studies
All subjects, abstracts, and the full text of articles were reviewed independently by 2 reviewers. According to the inclusion criteria, we selected the eligible trials. If disagreement occurred, we consulted the third reviewer to resolve the disagreement.

Data extraction and management
Two reviewers extracted independently data. The data extracted including the following categories: study ID, study design, study location, total patients, follow-up, mean age, gender, incidence of C5 palsy after anterior or posterior approaches including ACDF, ACCF, ACCDF, LP, and LF, sex of patients with C5 palsy.

Statistical analysis
We used STATA 12.0 to analyze data (Stata Corporation, College Station, TX). Both were reported with 95% confidence intervals (CI) and a P value of.05 was applied as the level of statistical significance. We used I 2 tests to assess statistical heterogeneity, which was from 0% to 100% in meta-analysis assessments. When I 2 >50% among the included studies, we chose random effects model; if not, we chose fixed-effects model. [17,18]

Test for risk of publication bias
We used a visual inspection of the funnel plot to assess publication bias. If there is publication bias, the funnel plot should be asymmetric, if not, the funnel plot is symmetric. We also performed the Egger and Begg tests to measure the funnel plot asymmetry by using a significance level of P<.05.
Additionally, we applied trim and fill computation to estimate the effect of publication bias.

Search results
We had searched 321 English studies in MEDLINE, EMBASE, 63 Chinese studies in WANFANG and CNKI. Of these, 136 English articles and 30 Chinese after duplicates removed, 80 English articles and 13 Chinese articles were excluded due to unrelated studies. Fifty-four English articles and 10 Chinese article were excluded due to eligibility criteria. As a result, a total of 61 studies were identified for this meta-analysis. The literature search procedure was shown in Fig. 1.

Baseline characteristics and quality assessment
A total of 61 studies including 721 patients with C5 palsy in total 11,481 patients (6.3%) were included in our study. Baseline characteristics of included articles are shown in Table 1.
All included studies were retrospective studies, we used the Newcastle Ottawa Quality Assessment Scale to assess the quality of each study. This scale for nonrandomized case controlled studies and cohort studies were used to allocate a maximum of 9 points for the quality of selection, comparability, exposure, and outcomes for study participants. Of these studies, 53 studies scored 8 points and 8 studies scored 7 points. Hence, the quality of each study was relatively high ( Table 2).

Surgical approaches-related C5 palsy
The results revealed that anterior approaches (5%) have a lower incidence of C5 palsy than these in posterior approaches (6.2%).

Diseases type-related C5 palsy
We only computed the incidence of C5 palsy for CSM and OPLL, because other cervical diseases lack enough data. The Table 2 The quality assessment according to the Newcastle Ottawa Quality Assessment Scale (NOQAS) of each study.

Publication bias
According to the shape of the funnel plot (Fig. 22) and P value (P = .557,.117) of the Begg and Egger regression tests, no visually asymmetrical and statistical evidence of publication bias of included studies is revealed. Likewise, for subgroup analysis, publication bias was also not found in included studies.

Discussion
A number of studies focused on the occurrence of C5 palsy after cervical surgery. Even though some mechanisms trying to explain this common complication have been proposed, it remains a     controversial issue. C5 palsy after cervical surgery is considered to be a result of nerve root injury or segmental spinal cord disorder. [64] As some conditions may contribute to C5 palsy, we reviewed 5 pathologic mechanisms as follows: inadvertent injury to the nerve root during surgery; [79] shifting of the cord caused nerve root traction after surgery; [80] spinal cord ischemia caused by decreased blood supply; [81] segmental spinal cord disorder; [82] and reperfusion injury of the spinal cord. [83] Several meta-analyses reported on the incidence of C5 palsy. Shou et al [2] performed a meta-analysis focused on the epidemiological prevalence estimates of C5 palsy following cervical surgery and it was based on 13,621 patients from 79 articles. He concluded that cervical surgery is associated with C5 palsy, particularly in patients who received LF and male patients are risk factors of C5 palsy. Gu [16] explored the incidence and risk factors of C5 palsy after posterior cervical surgery by a metaanalysis and found that patients with excessive spinal cord drift, preexisting intervertebral foramenal stenosis, OPLL, laminectomy, and male patients were risk factors for C5 palsy. Gu just assessed the C5 palsy after posterior approaches. We conducted a meta-analysis on C5 palsy for the last decade.

Anterior approaches versus posterior approaches
Bydon et al [84] reported on comparison between anterior and posterior approaches, the incidence of C5 palsy was significantly higher in the posterior method than the anterior method (8.6% vs 1.6%) (P<.001). Chang et al [40] performed another study to compare anterior and posterior approaches. In their study, the incidence of C5 palsy has been reported as 0.7% and 8.8% for anterior and posterior, respectively. Chen et al [24] reported the incidence of C5 palsy following posterior surgical routes as 24.3% and 0% following anterior surgical routes. In the same study, the highest incidence of C5 palsy of all published studies has been reported (25% for posterior laminectomy and fusion). These studies revealed that anterior surgical routes for patients had lower risk of developing postoperative C5 palsy. Shou et al [2] suggested that patients who received posterior cervical surgery (5.8%) had a slightly higher prevalence than patients who underwent anterior surgery (5.2%). Our results showed that posterior approaches (6.2%) had a higher incidence of C5 palsy, compared with anterior approaches (5%). We had a similar trend with Shou's, but we had difference of the incidence of C5 palsy  Our included studies were derived from last decade, but Shou's were from 1989 to 2014, which might lead to the difference. As we know, posterior shift of the spinal cord at C4-5 in posterior group has been significantly greater than that in anterior group. Nakashima et al [54] reported that C5 palsy was caused by posterior shift of the spinal cord and additional iatrogenic foraminal stenosis due to cervical alignment correction after posterior instrumentation with fusion.

Different anterior procedures
Liu et al [50] accessed complications of different techniques anterior decompression and found that patients who underwent  multilevel corpectomy group had the highest incidence of C5 palsy (11.9%). Lin et al [49] explored the same topic on comparison of ACDF and ACCF in patients with multilevel CSM. Their conclusions revealed that the incidence of C5 palsy was 3.5% in ACDF and 4.8% in ACCF. ACCF had higher incidence of C5 palsy. In our meta-analysis, patients who received ACDF had the lowest incidence of C5 palsy (5.5%) and ACCF was reported as the highest prevalence (7.5%) in anterior group. ACDF could preserve more vertebral body and provide more points of distraction and fixation except for the graft and interbody space shaping than these of ACCF. ACCF may lead to significant drift of spinal cord away from ventral side.

Different posterior procedures
Posterior cervical decompression, LP and LF, is a well-recognized surgical approach for multilevel CSM or OPLL. Yang et al [19] believed that the C5 palsy rate in the LP group is significantly higher than that in the LF group. However, Xia et al [85] had diverse point that there was no significant difference between LP  and LF. Although many studies reported on the occurrence of C5 palsy after posterior cervical decompression, its detailed mechanism remained poorly understood. Our results showed that the incidence of C5 palsy was significant higher in LF, which was similar to results of previous meta-analysis, and suggested LF as a significant risk factor. That may be because the LF removes the intact posterior arch of the vertebra, thus providing an excessive space for the spinal cord to shift posteriorly and showing greater change in dural sac area. For this reason, we considered LP as a more viable posterior option for patients with CSM or OPLL.

CSM and OPLL
We evaluated incidence of C5 palsy for patients with CSM and OPLL. We found that patients with CSM (4.8%) had a lower incidence of C5 palsy than patients with OPLL (8.1%). We also assessed the incidence of C5 palsy for CSM or OPLL in ACDF, LP, or LF. The consequences showed that incidences of C5 palsy for CSM in ACDF, LP, and LF were 4.7%, 3.1%, 13%, respectively and for OPLL were 5.5%, 8.1%, 13.1%, respectively. The incidence of C5 palsy for CSM group by ACDF and LP was lower than these in the OPLL group; nevertheless, the prevalence for both CSM group and OPLL group in LF was similar. Above all suggested that patients with OPLL, compared with patients with CSM, were more susceptible to this complication.

Sex and sides for C5 palsy
Our results revealed that male patients (5.9%), compared with female patients (4.1%), were more likely to have C5 palsy, which were similar to Shou's. In most cases, C5 palsy was unilateral.

Limitations
There were several limitations of this study. First, there was no RCT on C5 palsy, we need RCT to further study; Second, the statistical power could be improved in the future by including more studies. Some parameters, like 1-level CSM for C5 palsy, 2-level CSM for C5 palsy, or multilevel CSM for C5 palsy, due to the lack of data could not be analyzed by subgroups to avoid a high heterogeneity which may exert instability on the consistency of the outcomes; Third, the searching strategy was restricted to articles published in the English and Chinese languages. Articles with potentially high-quality data that were published in other languages were not included because of anticipated difficulties in obtaining accurate medical translations. In summary, in posterior approaches, male patients and patients with OPLL have a higher incidence of C5 palsy. In ACDF and LP, patients with OPLL had a higher incidence of C5 palsy, but in LF, both patients with CSM and OPLL had similar incidence of C5 palsy. Future more studies with high