Introduction

Cervical spondylotic myelopathy refers to clinical changes that frequently are related to compression of the spinal cord due to degenerative spinal stenosis. With people growing older, an increase of patients with cervical spondylotic myelopathy is expected. The natural course is often poor. With surgical decompression, a stabilization of neurological deficit or even recovery may take place in the majority of the patients. Surgical decompression can be performed either anteriorly, posteriorly or both approaches combined and is variably supplemented by additional fusion. Discussion about the best anatomical approach is beyond the scope of this article.

However, when considering the posterior approach, two methods are usually being applied. The oldest posterior approach is laminectomy, which can be performed with or without fusion [1]. Recently, a modification has been introduced which is called skip laminectomy [2]. In skip laminectomy, standard laminectomies are performed in combination with partial laminectomies of selected laminae to leave the muscular attachments undisturbed.

The other posterior approach is laminoplasty. In English literature, it was first described by Tsuji in 1982 [3], although different kinds of laminoplasty were described in 1973 [4], in 1978 [5], and in 1982 [6].

Laminoplasty might prevent postoperative spinal deformities which were seen after laminectomy [3]. Laminoplasty is technically more demanding, and if implants are used it is more expensive than laminectomy. Currently, as noticed in meetings and courses, more and more surgeons seem to perform laminoplasty.

This meta-analysis was performed to investigate whether a difference exists in clinical outcome, radiological outcome and complication rate between non-instrumented laminectomy and laminoplasty for cervical spondylotic myelopathy.

Methods

A highly sensitive search strategy was performed by an experienced librarian in Pubmed, Embase, and the Cochrane library including a search with Mesh or thesaurus terms complemented with a free text search in the title and abstracts (see Table 1). Since the first report that could be retrieved in Pubmed or Embase was published in 1982 [3], the search started in that year (Fig. 1).

Table 1 Example of searchstring for Pubmed
Full size table
Fig. 1
figure 1

Results of search in EmBase, Pubmed and Cochrane library

Full size image

Only articles published in Dutch, English, German or French were included. The goal was to identify randomized controlled trials (RCTs). However, if we would identify less than three RCTs, observational studies were also included.

Inclusion criteria were adult patients suffering from cervical spondylotic myelopathy, treatment consisting of either non-instrumented laminectomy or laminoplasty, series of more than or equal to 20 cases in either group, and score according to the Japanese Orthopedic Association (JOA) or its modification (mJOA) as an outcome measurement.

Two review authors (MA and WM) independently assessed titles and abstracts for possible inclusion. If they did not agree, the opinion of a third (WP) reviewer was obtained. Subsequently, full text versions were retrieved and assessed by the same two review authors independently. The reference lists were checked for additional articles.

The following outcome measurement was considered as primary outcome: JOA or mJOA. The maximum score of both scales is 17 and the subdivision in the original and modified scale is similar. Secondary outcomes were VAS neck pain, complication rate, SF 36, EQ5d, SF 12, ROM, Ishihara’s cervical curvature index [7] and specification of complications.

The risk of bias was assessed using the criteria proposed by the Cochrane Back Review Group [8]. The level of evidence was assessed according to the guidelines of the GRADE working group [9–11].

SPSS 20 (IBM Corporation, North Castle Drive, Armonk, NY 10504-1785, USA) was used for statistical analyses. We calculated weighted means taking into account the sample size of the study and its mean with a range of the reported means.

If not provided by the article, the recovery rate according to Hirabayashi [12] was calculated: recovery rate (%) = (postoperative (m)JOA − preoperative (m)JOA)/(17 − preoperative (m)JOA) × 100.

Results

After removing duplicates, 813 references to studies were identified. Finally, the full text versions of 68 articles were retrieved. After careful review of these articles, 29 [2, 13–40] were finally selected and included in this systematic review. Only one was a RCT comparing laminoplasty with a modification of laminectomy [40]. Since the goal of the systematic review was to compare laminectomy versus laminoplasty, randomized controlled studies of two forms of laminoplasty were considered as two prospective cohorts describing the similar outcomes and also included in the review. Most articles reported a mean without any standard deviation and, therefore, a meta-analysis was not possible.

For all studies, the risk of bias was considered high. The RCTs [27, 40] did not report the process of randomization, allocation, blinding or dealing with missing data. The risk of bias of the studies is presented in Table 2. Since the observational studies were not methodologically rigorous, the quality of evidence using GRADE was not upgraded and remained very low [41].

Table 2 Grading of quality of studies according to GRADE
Full size table

Since some articles reported the results of two cohorts of patients, these 29 articles represented 35 cohorts of a total of 1492 patients. The weighted mean age was 60.8 years (51–83 years) and mean follow-up 42 months (12–158 months) postoperatively. Six articles did not report the male to female ratio [15, 19, 22, 24, 26, 31]. In the remaining 1215 patients, the male to female ratio was 810/405. Baseline characteristics of the two treatment groups are presented in Table 3.

Table 3 Baseline characteristics of the different treatment groups
Full size table

There were no clinically important differences in pre- and postoperative (m)JOA, pre- and postoperative ROM, and Ishihara indices (Tables 4, 5, 6, 7).

Table 4 Weighted means of preoperative (m)JOA and postoperative (m)JOA with range and number of cohorts that reported this item
Full size table
Table 5 Weighted means of preoperative ROM and postoperative ROM (ranges) and number of cohorts that reported ROM
Full size table
Table 6 Weighted means of Ishihara-indices (%) with range and number of cohorts
Full size table
Table 7 Number of complications (percentage of included patients)
Full size table

All studies showed clinical improvement after surgical intervention. The overall mean Hirabayashi index was 56.9 % (35.6–73.9), for laminoplasty 57.2 % (35.6–73.9), for not skip laminectomy 54.6 % (45.8–66.7), and for skip laminectomy 55.6 % (50.7–60.5). The preoperative VAS was only reported in four cohorts and the postoperative VAS in six. All articles reported on laminoplasty. The mean pre-operative VAS for neck pain was 4.0 (1.4–6) and the mean postoperative VAS was 3.0 (1–4.6). None of the studies reported results on quality of life.

Discussion

Three posterior modalities exist for the treatment of cervical spondylotic myelopathy: laminoplasty, laminectomy, and skip laminectomy. These techniques can be supplemented by additional fusion.

From a biomechanical point of view, laminoplasty and laminectomy are similar. In both techniques, the muscles are widely dissected and ligamentous structures transected. The lamina are removed or opened. During this action damage to surrounding tissue (joints) might occur. Laminoplasty has an even higher risk of complications if the lamina is fixed with implants to the lateral mass (warranting wider dissection). The skip laminectomy is different compared with laminoplasty, since not all ligamentous attachments are sacrificed but preserved at selected spinous processes. Laminectomies are also restricted to the levels of compression and, therefore, not standard from C3 to C7 anymore.

The clinical results, the postoperative ROM, and the prevalence of a postoperative kyphotic deformity did not differ between the groups. Since a posterior approach is generally performed when the cervical curvature is lordotic, we feel confident to conclude that a difference of 6 % is clinically not important and that from this analysis it was not apparent that more kyphotic deformity occurred in the laminectomy group. None of the studies reported uniformly whether a kyphotic deformity occurred. Therefore, a more detailed analysis was not possible. This result corresponds with what you would expect from a biomechanical point of view.

Postoperative neck pain was only reported in a few studies. Some studies showed a benefit of preserving the lamina of C7 for reducing postoperative axial pain [20, 33, 35]. In a recent review, however, it was concluded that several factors may contribute to postoperative neck pain after posterior cervical surgery, but the evidence was not convincing. Definite conclusions were not possible due to the lack of uniform design of the studies and poor presentation of the results [42].

General complications seemed to be highly similar in both groups. Complications as hardware failure or malpositioning, and closing of the lamina will only occur with laminoplasty. This is a serious consideration when opting for laminoplasty.

This systematic review has some limitations. We only identified one RCT that compared laminoplasty and laminectomy, so we mainly had to rely upon data from observational studies. The quality of the studies was very low. The presentation of the results was also poor and, therefore, we were not able to perform a meta-analysis.

Considering the limitations of this systematic review and of the original studies, strong conclusions are not opportune. Due to the high risk of bias of the studies and the low quality of evidence, it is evident that at present none of the procedures has performed better than the other on clinical outcomes, postoperative axial neck pain or postoperative kyphotic deformity. However, the complication rate of laminoplasty seemed higher.

Laminoplasty is a relatively new surgical approach, which is propagated at the expense of laminectomy during courses and scientific meetings. This systematic review showed that laminoplasty has been introduced without any sound scientific support. In our opinion, laminoplasty should still be considered a new technique and cervical laminectomy as usual care. A well-designed RCT with a low risk of bias and an adequate sample size comparing laminectomy versus laminoplasty is necessary to evaluate which surgical method is performing better on clinical results (including patient reported outcomes) and complication rate. This systematic review presented some troublesome results. For a serious neurological threatening degenerative disorder with growing societal impact caused by an aging population and the global wish to stay mobile up to high age, the scientific evidence for any surgical approach is lacking. We strongly recommend performing an economic evaluation alongside this trial to evaluate which technique is most cost-effective. We are obliged to our patients to inform them properly about the safety and quality of our surgical actions.