Keywords

Introduction

Degenerative cervical myelopathy (DCM) represents a spectrum of chronic atraumatic spinal cord injury that occurs secondary to compression from disc spondylosis, hypertrophy of the ligamentum flavum, or ossification of the posterior longitudinal ligament (OPLL), etc. [1]. While surgical management has been shown to arrest the progressive deterioration and provide neurological and functional improvement, the selection of surgical procedures pertaining to specific cases is subject to much controversy [2,3,4]. While the subject of anterior versus posterior spinal decompression has been repeatedly debated amongst the experts, this chapter will focus on posterior surgical options and anterior alternatives will not be further elaborated.

The posterior cervical decompression techniques have a long track record of success in halting the progression of DCM. Prior to the introduction of cervical instrumentation strategies, posterior cervical laminectomies (LA) were a common procedure utilized to treat multilevel compressions [5,6,7]. Although initially found to be clinically effective, the development of late neurological deterioration secondary to post-operative instability and kyphosis [8,9,10,11] has led to the drive for alternative procedures. Since its introduction, laminectomy and instrumented fusion (LF) has been gaining popularity among the options in cervical decompression procedures [12,13,14,15]. In the modern era, the use of lateral mass screws and titanium rods allowed us to move away from wiring and plating of the cervical spine and to increase the safety of these procedures [16,17,18,19,20]. However, despite the advantages, the added stability with instrumentation comes at a cost of significant loss of mobility and range of motion (ROM) in the cervical spine [21,22,23].

The cervical laminoplasty (LA) procedure was designed to increase the overall spinal canal diameter by partial opening and elevating the laminae while keeping the posterior elements intact [24]. While a number of techniques have been described in the literature, they are largely the variations of two common procedures known as the “open-door” and the “French door” [25,26,27,28]. Although the posterior elements are preserved in these surgeries, post-laminoplasty kyphosis has been described in the literature [29,30,31,32,33]. Therefore, it is generally not the surgical procedure of choice when severe preoperative kyphotic deformity is present. Additionally, since the procedure itself avoids fusion, the benefit of relative preservation of motion comes at a cost of postoperative neck pain [34, 35].

Furthermore, the drive to preserve motion and prevent instability while achieving adequate decompression in the cervical spine has let to the development of a minimally-invasive, mucle-preserving posterior approach utilized in skip laminectomy (sLA), where only selective laminectomies are performed. Since this technique was introduced by Shiraishi et al. in 1998, it has shown promising results as a non-instrumented alternative for posterior decompression [36,37,38].

Nowadays, it is generally accepted that a fusion procedure should accompany any cervical decompression in the presence of kyphotic deformity [5, 39]. However, controversy exists as to the optimal procedure for DCM in a lordotic cervical spine. The goal of this chapter is to systematically review and summarize the evidence in the literature on the comparative efficacy and safety of the common posterior cervical spine procedures.

Material and Methods

Generation of Key Questions

Key questions were formulated to address important clinical inquiries: (Q1) What is the efficacy of LA or sLA compared to LF based on clinically important changes in neurological and functional status? (Q2) What is the efficacy of LP compared to LA or sLA based on clinically important changes in neurological and functional status? (Q3) What is the efficacy of LF compared with LP based on clinically important changes in neurological and functional status? (Q4) What is the safety profile of LA or sLA compared to LF? (Q5) What is the safety profile of LP compared to LA or sLA? (Q6) What is the safety profile of LF compared to LP?

Electronic Literature Search

The literature search was performed by an experienced librarian using Ovid MEDLINE, Embase, and Cochrane library databases. To ensure high sensitivities in our systematic review, appropriate search concepts were developed with relevant subject headings complimented by text word searches of titles and abstracts using relevant MeSH terms and synonyms. Appropriate truncations, adjacent operators, parentheses and Boolean operators were employed to ensure the inclusiveness as well as the precision of the search.

For the purpose of this chapter, only English language articles published after January 2000 were included. For inclusion into the review, articles must include adult human patients (age >18), diagnosed with DCM, surgically treated with either LP, LA, sLA, or LF. Studies must have a clear reporting of neurological and/or functional outcomes both preoperatively and postoperatively or describe the spectrum and incidence of complications. In order to present the highest quality of evidence in the literature, the selection was further limited to only randomized control trials (RCT) and comparative studies with ≥10 patients in each treatment group.

Two authors independently screened titles and abstracts for relevant articles pertaining to the scope of this chapter. Full manuscripts of the selected articles subsequently underwent extensive review by the same authors. When conflicting opinion on the inclusion and exclusion of articles arose, the issues were either resolved by discussion, or when necessary, the advice from a third author was sought. For each article selected, the references list was carefully reviewed for additional relevant articles to include.

Data Extraction

An exhaustive assessment of outcome scores was attempted to address Q1–3 in this review, but the overall paucity of generalized reporting algorithms hindered this effort. Therefore, owing to their relatively consistent appearance across studies, the Japanese Orthopaedic Association (JOA) and the modified JOA (mJOA) Score for the Assessment of Cervical Myelopathy along with the Neck Disability Index (NDI) were selected as the primary measures of neurological and functional outcomes. The Visual Analogue Scale (VAS), Ishihara’s cervical curvature index [40], and neck range of motion (ROM) were included as secondary outcomes. For Q4–5, the spectrum and incidence of key intraoperative and postoperative complications of each procedure were evaluated.

Evaluation of Level of Evidence and Strength of Literature

Gradings of the level of evidence were independently performed by two authors for each published article based on the criteria suggested by the Journal of Bone & Joint Surgery and Agency for Healthcare Research and Quality (AHRQ) [41, 42]. The overall strength of evidence for each outcome of interest was determined based on the recommendations by Grading of Recommendation Assessment, Development and Evaluation Working Group [43,44,45,46] and the AHRQ [42].

Data Analysis

For Q1–3, in order to compare the clinical effectiveness of the procedures, the difference in means, standardized mean difference (SMD), and 95% confidence interval (CI) between the treatment groups were calculated based on the preoperative and postoperative data reported in the manuscript. Missing data were input using methods proposed by the Cochrane Handbook for Systematic Reviews of Intervention [47]. To address the effectiveness of an intervention, the suggested measure by Cohen [48] was used whereby an effect size of 0.2 is considered as “small”, 0.5 as “medium” and ≥0.8 as “large”. For Q4–5, the spectrum and incidence of complications were extracted from the studies. For each complication, relative risk or risk ratio (RR) and 95% CI were calculated. Statistical calculations were performed using the Comprehensive Meta Analysis Version 3.3.070 (Biostat, Inc. Englewood, NJ, USA) [49].

Results

The literature search resulted in 6344 articles after removal of duplications. Through exclusion based on title, abstract, and year of publication, 178 articles remained for full manuscript review. Electronic manuscripts were obtained, and careful review of the articles was performed. By applying our inclusion and exclusion criteria, 16 articles were finally selected for the systematic review plus three additional articles that were found through screening of the reference lists (Fig. 11.1). A total of two articles were found that addressed Q1 [50, 51], six articles addressing Q2 [36, 37, 50,51,52,53], 12 articles for Q3 [21,22,23, 50, 51, 54,55,56,57,58,59,60], two articles for Q4 [51, 61], four articles for Q5 [36, 37, 51, 62] and 11 articles for Q6 [21,22,23, 51, 54,55,56, 58,59,60, 63]. The list of all included articles as well as their level of evidence are presented in Table 11.1.

Fig. 11.1
figure 1

Flowchart of literature search

Full size image
Table 11.1 Summary of the level of evidence for included studies
Full size table

For Q1 and 2, due to the limited number of studies identified through the literature that specifically addressed these questions, a qualitative review is presented in this chapter. The summary of reported outcome scores, imputed values, calculated SMD and 95% CI for each comparison are presented in Table 11.2. For Q3, although an adequate number of articles were identified, due to high heterogeneity of the studies, the pooling of the data for meta-analysis was not performed. The calculated SMD and 95% CI for each outcome are presented in the form of forest plots (Fig. 11.2). For Q4 and 5, qualitative review of the literature is presented due to the limited number of studies. Results of calculated RR and 95% CI are summarized in Table 11.3. Finally, for Q6, due to the larger number of studies, a forest plot was used to present the summarized results (Fig. 11.3).

Table 11.2 Summary outcomes of laminectomy vs. laminectomy and fusion, laminoplasty vs. laminectomy, laminoplasty vs. selective/skip laminectomy
Full size table
Fig. 11.2
figure 2

(a) Summary of standardized mean difference comparing the improvements of neurological and functional outcomes as measured by JOA/mJOA and NDI achieved through laminoplasty and laminectomy and fusion as reported by individual studies. (b) Compiled summary of standardized mean differences of secondary outcomes achieved by laminoplasty and laminectomy and fusion as reported by individual studies

Full size image
Table 11.3 Summary of studies reporting complications of laminectomy vs. laminectomy and fusion, laminoplasty vs. laminectomy, laminoplasty vs. selective/skip laminectomy, and laminoplasty vs. laminectomy and fusion
Full size table
Fig. 11.3
figure 3

Summary of risk ratios comparing rates of complications between laminoplasty and laminectomy and fusion

Full size image

What Is the Efficacy of LA or sLA Compared to LF Based on Clinically Important Changes in Neurological and Functional Status?

Comparison of LA versus LF was found in two retrospective comparative studies and no studies were found comparing sLA to LF (Table 11.2). Only one identified study compared the neurological recovery between the two procedures. Du et al. [50] reported significant JOA improvement in both the LA (n = 30) and LF (n = 32) treatment groups, with a statistically significant difference in recovery in favor of LF. In terms of functional recovery, one of the studies compared NDI recovery in patients with cervical myelopathy secondary to OPLL treated with either LA (n = 15) or LF (n = 21) [51]. In this study, Lee et al. [51] noted that patients in both treatment groups demonstrated substantial improvement in the NDI score, and the degree of recovery was not significantly different between the groups.

In terms of the secondary outcomes, a single study by Lee et al. [51] found no significant difference in the postoperative improvement in the VAS-neck score between LA and LF. The cervical curvature was assessed by both studies using the Ishihara Index [50, 51]. Their conclusions, however, were inconsistent with Du et al. [50] reporting significantly more loss of cervical lordosis after LA, while Lee et al. [51] reported an equivocal decrease over time with both procedures. Interestingly in the latter study, a lower mean preoperative Ishihara Index was noted in the LF group, demonstrating the authors’ predilection for fusion in patients with less cervical lordosis [51].

None of the studies reported on postoperative ROM changes.

What Is the Efficacy of LP Compared to LA or sLA Based on Clinically Important Changes in Neurological and Functional Status?

LP Versus LA

Three retrospective studies compared clinical outcomes between patients treated with LP and LA (Table 11.2). Du et al. [50] reported significant improvements in postoperative JOA score with both LP (n = 36) and LA (n = 30), but revealed significantly higher recovery rates in patients treated with LP. In terms of NDI improvement, one article by Lee et al. [51] did not find a statistically significant difference between LP (n = 21) and LA (n = 15).

For the secondary outcomes, two studies evaluated the change in VAS-neck score. The study by Lee et al. [51] showed equivocal postoperative improvements in both procedures. A relatively larger study by Nurboja et al. [53] (n = 48 for LP, n = 33 for LA) reported no difference in VAS-Neck scores with either surgical technique. Interestingly, authors of the latter study found that the pain relief in the LA group only became significant when surgery was performed on ≥4 vertebral levels.

The changes in Ishihara Index were also compared between the two procedures and reported by all three studies. In the retrospective study by Du et al. [50], a loss of cervical lordosis was described in both groups with more pronounced changes noted after LA, but their results were not reproduced by the other studies. Lee et al. [51], on the other hand, found the loss of lordosis to be of similar magnitude amongst techniques, and Nurboja et al. [53] in their evaluation of LP (n = 75) and LA (n = 34) reported minimal change in sagittal alignment over time in either group.

None of the studies in this review compared the postoperative changes in ROM.

LP Versus sLA

A total of three articles were identified in the literature search addressing the outcomes of LP compared to sLA (Table 11.2). In a single prospective randomized control trial, Yukawa et al. [37] concluded that the long-term JOA score improvements or other functional outcomes were not significantly different between DCM patients treated with LP (n = 21) and sLA (n = 20). Chang et al. [52], in a retrospective review of patients treated with LP (n = 35) and sLA (n = 32), showed improvement in both NDI and VAS-Neck with no difference between the two treatment groups.

Shiraishi et al. [36] reported the postoperative changes in Ishihara index on their cohorts undergoing sLA (n = 43) in comparison to LP (n = 51), and noted that while the curvature index was maintained after sLA, a significant decrease was seen in the LP group.

Finally, the pre- and postoperative ROM was assessed by two of the studies with inconsistent findings. While Yukawa et al. [37] reported slightly more preserved ROM after sLA, the article by Chang et al. [52] reported significantly less postoperative ROM.

What Is the Efficacy of LF Compared with LP Based on Clinically Important Changes in Neurological and Functional Status?

For eight studies that reported the preoperative and postoperative JOA/mJOA scores, the calculated SMD and the 95% CI are demonstrated in Fig. 11.2a. The majority (five studies) reported no or “small” effect without statistical significance. Only in one study was a significantly “large” effect detected in favor of LP (SMD 1.231). In this retrospective analysis by Stephens et al. [58], the comparison was made between LP (n = 85) and LF (n = 52). Although significant results were noted, the authors reported baseline differences between groups and possible selection bias in treatment. Two studies reported “medium” effect with conflicting findings. One single retrospective study by Miyamoto et al. [57] favored LF (n = 30) over LP (n = 30) in JOA improvement, while a large multicentered, prospective, observational study by Fehlings et al. [54] (n = 100 for LP, n = 166 for LF) showed more improvement in the LP group [54]. The assessment of functional improvement using preoperative and postoperative NDI was reported by five studies. Most of the support for LP derives from two studies with one “large” (SMD 1.165) and one “medium” (SMD 0.655) effect by Stephens et al. [58] and Yang et al. [23], respectively. The study by Stephens et al. [58] suffers from the possibility of bias as mentioned above, but the retrospective study by Yang et al. [23] showed significant NDI improvement in LP (n = 75) compared to the LF (n = 66) group. Two other studies were in favor of LP with “small” and non-significant effect [21, 54], while one study showed no difference between procedures [51].

The results of the assessment of secondary outcomes were summarized in Fig. 11.2b. A total of six studies reported postoperative improvement of VAS-neck pain. Of the studies that were found to favor LF, the retrospective study by Highsmith et al. [55] (n = 30 for LP, n = 26 for LF) showed the largest effect (SMD 1.217), with the LP group experiencing an increase in VAS-neck pain while the LF group reported a significant improvement. To a lesser extent, the retrospective study by Stephens et al. [58] also supported LF with statistical significance (SMD 0.732). Three other studies showed “small” effects in favor of LF. On the other hand, the study by Yang et al. [23] was found to have a “large” (SMD −1.050) effect favoring LP.

In terms of the Ishihara Index, two of the three studies favored LF in maintaining the curvature index with a “small” effect, while one study showed a SMD of 0.083 in favor of LP. None of the studies showed a strong effect. Finally, all three studies comparing ROM revealed a “large” effect favoring LP in preserving more neck motion.

What Is the Safety Profile of LA or sLA Compared to LF?

The comparison of complication rates between LA and LF were found in two studies (Table 11.3). The rate of C5 palsy was reported by a single retrospective review by Lee et al. [51]. The authors found no incidence of the event (0/75) in the LA group and 9.52% (2/21) in the LF group [51]. In a prospective randomized comparative study, postoperative infection was documented in 6.67% (2/30) and 9.38% (3/32) after LA and LF, respectively [61]. In the same study, the authors reported the observed incidence of dural tear/CSF leak after LP at 3.33% (1/30) while it was 6.25% (2/32) for LF. None of the studies reported or compared the rates of postoperative kyphosis, and no studies comparing sLA to LF were found.

What Is the Safety Profile of LP Compared to LA or sLA?

Two studies compared complication rates between LP and LA (Table 11.3). Della Pepa et al. [62], in their retrospective review, found no infection (0/33 for LP, 0/24 for LA) or dural tear/CSF leak (0/33, 0/24) in both groups. However, the LA group demonstrated a 12.5% (3/24) rate of postoperative kyphosis in comparison to none (0/33) reported in the LP group [62]. Regarding the rate of C5 palsy, Lee et al. [51], reported 0% in both LP (0/21) and LA (0/15) groups in their retrospective study.

In terms of LP and sLA, the data on complications were acquired from two studies. In the RCT by Yukawa et al. [37], the rate of infection was found to be 0% in both groups (0/21 for LP, 0/20 for sLA). In the retrospective study by Shiraishi et al. [36], no incidence of CSF leak was reported in the LP (0/51) group compared to 4.65% (2/43) in the sLA group. Both studies reported on postoperative C5 palsy, with rates of 5.88% (3/51) versus 0% (0/43) [36], and 0% (0/21) versus 0% (0/20) [37], respectively, compared between LP and sLA.

What Is the Safety Profile of LF Compared to LP?

A total of 11 articles presented a comparison of complication rates between LP and LF (Table 11.3). The calculated RR and 95% CI are presented in Fig. 11.3. Of the studies that reported on rates of C5 palsy, the majority (seven of the eight articles) showed reduced risk of this complication with LP. However, only three studies [21, 23, 63] reached statistical significance with RR of 0.227, 0.240, and 0.138. Only one article showed a slight favoring of LF (RR 1.245), which failed to reach statistical significance [54].

Three studies reported rates of dural tear/CSF leak, with one study showing no difference in the rate between the two procedures [54], while the other two studies were inconsistent with one favoring LF (RR 2.613) [55] and the other favoring LP (RR 0.293) [23]. However, none of the studies reached statistical significance.

There were eight articles reporting on postoperative rates of infection. Analysis of RR revealed that all studies trended toward reduced risk of infection with LP, however, none of the studies reached statistical significance.

The rate of postoperative kyphosis was also described in four studies. Although none showed statistically significant differences, two studies reported a lower risk of postoperative kyphosis in LP (RR 0.332 and 0.692) [22, 54], while one study was in favor of LF (RR 1.320) [23].

Level of Evidence

The strengths of evidence are presented in Table 11.4.

Table 11.4 Strength of evidence summary
Full size table

Discussion

With the rapid advancement in the field of DCM, evidence-based clinical management guidelines have been developed to assist clinicians and surgeons in formulating treatment decisions [64,65,66,67,68,69]. While the clinical importance and efficacy of surgical decompression is not called into question, controversy still exists concerning the best surgical approach in the treatment of DCM. This chapter focuses on the common types of posterior cervical procedures and provides a synopsis of current evidence.

The clinical efficacy of posterior cervical decompressive procedures has a long track record of proven success in the treatment of DCM by providing clinically important improvements in neurological and functional outcomes [21, 23, 36, 37, 50,51,52, 54, 55, 57, 58, 60, 70]. Due to the unique characteristics of each technique, the selection of one over the other is still based mostly on surgeon preference and remains a heated debate in the spine community. However, it is generally accepted that due to the risk of instability and delayed deformity development, pre-existing cervical kyphosis is a contraindication for LA and LP [5, 8,9,10,11, 29,30,31,32,33, 39].

To best present the evidence, a systematic review of the literature was conducted focusing on the contemporary literature (after year 2000) and on studies with comparative data between the procedures. Due to the heterogeneity intrinsic to the available literature, the inconsistent methods of reporting outcomes and complications, as well as the lack of well-designed high-quality studies, it was not possible to recommend one of the surgical techniques as being superior. However, from the review of the current evidence, several important conclusions can be drawn to assist the surgeons and patients in making evidence-based decisions concerning surgical approach.

Regarding clinical outcomes, the systematic literature search resulted in only two retrospective studies addressing the efficacy of LA versus LF, and three retrospective studies evaluating LA versus LP. Similarly, the number of articles identified regarding the safety profile of the procedures was low (two studies comparing LA and LF and two studies for LP and LA). The paucity of literature on these comparisons likely reflects the dramatic decline in the utilization of LA in the contemporary era due to the increased awareness of the high-risk for postoperative kyphosis, instability, as well as the added possibility of neurological deterioration with their development [8,9,10,11, 50]. The extracted data from the limited number of studies provides “insufficient” evidence to recommend LA as a primary procedure for DCM (Table 11.4). In fact, there is “low” evidence in favor of LF regarding postoperative neurological improvement, and a trend toward potentially added improvement in functional outcomes with LF and LP [50].

From the included studies, it is recognizable that a trend exists suggesting loss of cervical lordosis post LA [50, 62]. Interestingly, it was noted that many of the authors, being aware of the issue of postoperative kyphosis, either limited their study by excluding patients with severe preoperative cervical deformity or tailored the treatment to offer fusion in the less-lordotic cases [50, 51]. Therefore, although the evidence presented was graded as “insufficient” in confirming an increased risk of kyphosis in LA, these studies indirectly provided insight that increased awareness surrounding this issue has led to changes in the practices of the spine community.

Given its relatively recent introduction, the evidence on the comparative effectiveness and safety of sLA compared to other posterior cervical procedures was limited to three studies found by literature search. Due to the overall lack of evidence and imprecision with reporting, no conclusive recommendation can be generated on the utility of sLA for DCM (Table 11.4), but studies appear to suggest that the clinical outcome and safety profile of sLA is comparable to that of LP [36, 37, 52]. The sLA procedure was designed to address issues with post decompression instability and kyphosis while preserving ROM [36, 38], and the limited reports on the potential of this procedure to maintain cervical curvature are promising [36,37,38], but this analysis fails to confirm these due to the limited and inconsistent evidence.

With the advancement of knowledge and the evolution of instrumentation technology, the field of spine surgery has seen rapid development over recent years. As LA has been gradually phased out in favor of LP and LF, a shift in the research focus nowadays is geared towards quality control and improvement. As a reflection of this change, 12 studies addressing the clinical outcomes and 11 studies comparing the complications of LP and LF were found by literature search. However, given the resources, there is still “insufficient” overall evidence to favor either procedure regarding neurological improvement, functional outcome or safety profile (Table 11.4). Given the inconsistencies seen in the literature and the difficulty in proving the superiority of one procedure over the other, it appears that LP and LF have overall clinical equipoise in treating DCM. While it is generally agreed upon that LP is associated with more postoperative neck pain [34, 35], insufficient evidence was found in this review in terms of VAS-neck pain score changes. However, it appears that the overall trend tends to favor more improvement with LF (Fig. 11.2). Unsurprisingly, a large difference was found with “moderate” evidence confirming greater loss of ROM with LF [21,22,23]. Of interest, it appears that “low” evidence exists supporting a lower rate of C5 palsy in LP, which is an observation that has been previously reported [71, 72]. However, given the state of the literature, it is difficult to provide absolute evidence favoring either procedure, and further research is necessary.

Conclusion

In conclusion, this chapter summarizes the current evidence concerning the common posterior spinal surgical approaches for the treatment of DCM. The evidence presented here is not a strict guideline but should serve as a suggestion to assist clinicians, surgeons, and patients when deciding a treatment plan. Ultimately, the approach of choice should be based on clinical judgement and tailored toward each individual patient depending on their clinical presentation, imaging findings, and the surgeon’s own experience. Overall this systematic review strengthens the evidence that LP and LF have overall clinical equipoise in terms of clinical outcome, while finding insufficient evidence to recommend LA or sLA as primary treatment options for DCM.

Patient Preference

See Table 11.5.

Table 11.5 Patient preference chart
Full size table

Box

  1. 1.

    What is known:

    1. (a)

      Cervical laminectomy has a risk of postoperative instability and kyphosis.

    2. (b)

      Cervical laminoplasty is associated with postoperative neck pain.

    3. (c)

      Cervical laminectomy and fusion are associated with decreased postoperative ROM.

  2. 2.

    What is new:

    1. (a)

      There is a trend showing reduced cervical lordosis with laminectomy and fusion procedures, although not reaching statistical significance.

    2. (b)

      Cervical laminoplasty and laminectomy and fusion have shown similar effects regarding overall clinical outcomes and safety profile, with the exception of low level-evidence suggesting decreased rates of C5 palsy in the laminoplasty group.

  3. 3.

    Consequences for clinical practice:

    1. (a)

      We do not recommend cervical laminectomy as the primary treatment for degenerative cervical myelopathy.

    2. (b)

      We cannot make recommendations for cervical skip laminectomy due to insufficient evidence showing its superiority over other common posterior cervical procedures.

    3. (c)

      We do not recommend the use of cervical laminoplasty when there is present or suspected instability.

    4. (d)

      We do not recommend the use of cervical laminoplasty in the presence of cervical kyphosis.

    5. (e)

      We recommend both cervical laminoplasty and laminectomy and fusion in degenerative cervical myelopathy but recommend laminectomy and fusion when instability and/or cervical deformity are present.