Introduction

Cervical degenerative disc disease (CDDD) and associated changes to surrounding structures can impinge on the spinal cord and spinal nerves presenting as myelopathy, radiculopathy, or both. For decades, anterior cervical discectomy and fusion (ACDF) has been considered the optimal surgical management of CDDD [3]. However, ACDF is associated with adjacent segment disease (ASD) [4, 6, 14, 26] due to a reduced range of motion (ROM) at the fused segments and a compensatory increase in mobility in adjacent vertebral levels [24]. Symptomatic ASD may require revision surgery [3, 4, 14, 29]; an additional burden on patients and healthcare providers. Treating multilevel CDDD with ACDF is associated with greater morbidity and poorer outcomes with each additional vertebral level involved [7, 33].

Cervical disc arthroplasty (CDA) is as an alternative procedure. Clinical and biomechanical studies report superior preservation of the kinematics of the cervical spine [9, 18], a reduction of abnormal motion at adjacent levels [9], and a lower incidence of ASD after CDA than ACDF [8], although ASD may still occur long-term [19]. Complications of CDA include implant subsidence and heterotopic ossification (HO) of surrounding soft tissue [19, 22, 25]. Multilevel CDA is a safe and effective alternative to ACDF [11], with preservation of cervical ROM and lower incidence of ASD [31, 39] and fewer revision surgeries [38]. Functional and clinical outcomes are equivalent to single-level CDA [37]; however, extensive use of CDA for multilevel CDDD is restricted by numerous contraindications [1, 19, 22].

Hybrid surgery (HS) involving a combination of ACDF and CDA at adjacent vertebral levels is being utilised in the treatment of multilevel CDDD. HS is appropriate in a select group of patients with different types of disease and different degrees of degeneration at contiguous levels, with the most appropriate intervention used at each. [35] The definition of HS varies throughout the literature, frequently including anterior cervical corpectomy and fusion (ACCF) as well as ACDF or CDA. Previous systematic reviews and meta-analyses have examined the outcomes of HS versus ACDF [10, 20, 28, 34,35,36], and included HS involving ACCF [10, 20, 36]. The objective of this review is to assess the safety and efficacy of HS comprising ACDF and CDA only, compared with both ACDF and CDA alone in the management of multilevel CDDD.

Materials and methods

Study selection

This review was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines [23]. Review protocol: PROSPERO ID CRD42017065980. A systematic literature review was performed using PubMed/Medline, Embase (via OVID), and Web of Science. Databases were searched from their date of inception to August 2017. A later search filtered for papers published between August 2017 and December 2018. Searches were generated using MeSH terms and keywords (Appendix 1). Two reviewers (M.A.H. and E.C.G.) screened studies by title and abstract. The full text of potentially eligible studies was assessed. Reference lists of included studies and previous systematic reviews were searched. References were managed using Mendeley and Microsoft Excel.

Inclusion and exclusion criteria

Studies meeting the following criteria were eligible for inclusion: (1) adult patient population with multilevel CDDD; (2) randomised controlled trials (RCTs) and non-randomised studies comparing HS with either ACDF or CDA or both; (3) appropriate clinical and/or radiographical outcomes reported. For studies with overlapping populations, the most recent study was included. Study types excluded were the following: non-comparative, biomechanical/in vitro/cadaveric, systematic reviews and meta-analyses, letters, or abstracts. Other exclusions were the following: full text not in English, HS involving corpectomy, revision surgeries, non-contiguous operative levels, serious comorbidities, or a non-hospital setting.

Risk of bias assessment

We used the Methodological Index for Non-Randomised Studies (MINORS) to perform the risk of bias assessment of non-randomised studies [27], and the Cochrane collaboration’s tool for assessing the risk of bias in the RCT using Review Manager (RevMan) software (version 5.3). This was performed independently by two reviewers (M.A.H and E.C.G) and disagreement resolved by discussion. A third reviewer (A.K.D) was available for mediation, throughout the review process.

Data extraction

Data was extracted into an Excel spreadsheet by two reviewers (M.A.H. and E.C.G.). General characteristics of the included studies were extracted as well as the following outcomes: Neck Disability Index (NDI), Visual Analogue Scale (VAS), C2–C7 range of motion (ROM), superior adjacent segment (SAS) ROM, inferior adjacent segment (IAS) ROM, operation time, intra-operative blood loss, return to work, and complications.

Quality of evidence assessment

The quality of the evidence was assessed according to the Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) working group which categorises evidence as being of high, moderate, low, or very low quality (Appendix 2) [2].

Statistical analysis

Data analysis and forest plots were produced using RevMan v5.3. Heterogeneity between studies was estimated using chi2 (p > 0.05) and the value of I2. A random effects model was applied to data due to substantial heterogeneity. For continuous data, the mean difference (MD) and 95% confidence intervals (CI) were calculated (inverse variance), and for dichotomous data, the odds ratio (OR) and 95% CI (Mantel-Haenszel). p value < 0.05 was considered statistically significant.

Results

Search results

A flow diagram of the selection process is shown in Fig. 1. Searching three electronic databases identified eight eligible studies (seven observational studies and one RCT) with a total of 424 patients [12, 13, 15,16,17, 21, 30, 32]: Grasso [12] n = 60 (14.2%); Hey [13] n = 21 (4.95%); Jang [15] n = 49 (11.6%); Ji [16] n = 40 (9.43%); Kang [17] n = 24 (5.66%); Mende [21] n = 111 (26.2%); Wang [30] n = 77 (18.2%); Xiong [32] n = 42 (9.91%). Of 424 patients, 175 underwent HS, 208 ACDF, and 41 CDA. Characteristics of included studies are in Table 1, reported outcomes in Table 2, and complications in Table 3.

Fig. 1
figure 1

Flow diagram of study selection process

Full size image
Table 1 Characteristics of included studies
Full size table
Table 2 Reported outcomes
Full size table
Table 3 Details of complications
Full size table

Risk of bias

MINORS scores for the seven non-randomised studies are shown in Table 4 (range 11–18, mean 16.4). Six studies were considered to have a moderate risk of bias and one study a high risk. The risk of bias assessment of the RCT is seen in Fig. 2: participants were randomised using the odd or even hospital number, giving a high risk of selection bias.

Table 4 MINORS scores of non-randomised studies
Full size table
Fig. 2
figure 2

Risk of bias of RCT

Full size image

NDI

Measurements of post-operative NDI were extracted from seven studies [12, 13, 15,16,17, 30, 32]. All seven compared HS and ACDF and three HS with CDA [12, 13, 30]. There was no significant difference in NDI after HS versus ACDF (MD = − 1.69; 95% CI − 4.07 to 0.69; p = 0.16; I2 = 83%). Values from Ji et al. [16] and Kang et al. [17] were higher following ACDF than following HS but no standard deviations were provided so they were not included in pooled analysis. NDI scores were not significantly different following CDA compared with HS (MD = 0.57; 95% CI − 3.01 to 4.16; p = 0.75; I2 = 71%).

VAS

Post-operative VAS scores for HS versus ACDF were obtained from all eight studies [12, 13, 15,16,17, 21, 30, 32] and for HS versus CDA from three [12, 13, 30] (Fig. 3). Mende et al. did not provide standard deviations so their data was not included [21]. The pooled estimate did not show favour to any of the procedures: ACDF (MD = − 0.08; 95% CI − 0.43 to 0.26; p = 0.64; I2 = 40%), CDA (MD = − 0.41; 95% CI − 1.17 to 0.34; p = 0.28; I2 = 76%). Values from the last follow-up appointment were used. Four studies reported a combined VAS score for neck and arm pain [12, 13, 17, 30], three provided separate VAS scores for neck and arm pain [16, 21, 32], and one reported VAS arm pain alone [15]. Where two scores were provided, the VAS for arm pain was used as it was common to all.

Fig. 3
figure 3

Forest plots comparing Visual Analogue Scale (VAS) scores for neck and arm pain after hybrid surgery (HS) versus a anterior cervical discectomy and fusion (ACDF) and b cervical disc arthroplasty (CDA)

Full size image

C2–C7 ROM

The post-operative C2–C7 ROM was reported in seven studies for HS versus ACDF [12, 13, 15,16,17, 30, 32] and three for CDA [12, 13, 30] (Fig. 4). ROM was significantly greater after HS compared with ACDF (MD = 6.14; 95% CI 1.94 to 10.35; p = 0.004; I2 = 74%). There was no significant difference in C2–C7 ROM between HS and CDA (MD = 6.42; 95% CI = − 12.97 to 25.80; p = 0.52; I2 = 98%).

Fig. 4
figure 4

Forest plots comparing C2–C7 ROM after hybrid surgery (HS) versus a anterior cervical discectomy and fusion (ACDF) and b cervical disc arthroplasty (CDA)

Full size image

SAS ROM

The post-operative SAS ROM for HS and ACDF was reported by four studies [16, 17, 30, 32] (Fig. 5). One study measured SAS ROM after CDA [30]. Ji et al. reported the relative difference in ROM to the pre-operative ROM; therefore, the data is too heterogeneous for inclusion in the pooled analysis [16]. SAS ROM was significantly lower after HS than after ACDF (MD = − 2.87; 95% CI = − 4.06 to − 1.68; p < 0.00001; I2 = 0%). SAS ROM was significantly greater after HS than after CDA (11.3 ± 2.7 and 10.7 ± 1.5, respectively; p < 0.05) [30].

Fig. 5
figure 5

Forest plot comparing superior adjacent segment (SAS) ROM after hybrid surgery (HS) versus anterior cervical discectomy and fusion (ACDF)

Full size image

IAS ROM

The post-operative IAS ROM for HS and ACDF was reported by four studies [16, 17, 30, 32] (Fig. 6). One study measured IAS after CDA [30]. Ji et al. reported the relative difference in ROM compared with the pre-operative ROM; therefore, the data is too heterogeneous for inclusion in the pooled analysis [16]. IAS ROM was significantly lower after HS than after ACDF (MD = − 3.11; 95% CI = − 4.87 to − 1.35; p = 0.0005; I2 = 35%), and after HS compared with CDA (10.0 ± 5.0 and 11.1 ± 1.8, respectively; p < 0.05) [30].

Fig. 6
figure 6

Forest plot comparing inferior adjacent segment (IAS) ROM after hybrid surgery (HS) versus anterior cervical discectomy and fusion (ACDF)

Full size image

Intra-operative blood loss

Four studies reported intra-operative blood loss during HS and ACDF [16, 17, 30, 32] and one during CDA [30]. There was no significant blood loss for HS versus ACDF (MD = − 14.58; 95% CI − 36.18 to 7.02; p = 0.19; I2 = 93%). Blood loss during CDA was 38.9 ± 9.6 mL compared with 38.8 ± 15.2 mL during HS (p > 0.005) [30]. Hey et al. used the post-operative drop in haemoglobin as a surrogate for blood loss (not included in analysis), which showed no significant difference following HS compared with ACDF or CDA (p > 0.05) [13].

Operation time

Operation time (minutes) was reported by seven studies for HS and ACDF [12, 13, 15,16,17, 30, 32]: there was no significant difference (MD = 9.96; 95% CI − 16.99 to 36.92; p = 0.47; I2 = 99%). Three studies reported the operation time for HS and CDA and there was no significant difference (MD = − 10.67; 95% CI − 26.85 to 5.52; p = 0.20; I2 = 93%) [12, 13, 30].

Return to work

Patients’ return to work following HS and ACDF was reported in three studies [12, 13, 21], and following HS and CDA in two [12, 13]; Mende et al. did not provide standard deviations so data could not be used in the pooled analysis, but reported a shorter return to work for HS than for ACDF by approximately 63 days [21]. Patients who underwent HS returned to work significantly sooner than those after ACDF (MD − 32.01; 95% CI − 33.13 to − 30.90; p < 0.00001; I2 = 0%) and after CDA (MD = − 32.92; 95% CI − 43.58 to − 22.06; p < 0.00001; I2 = 27%).

Complications

All eight studies reported post-operative complications (Table 3; Fig. 7) [12, 13, 15,16,17, 21, 30, 32]. A total of 37 events were reported after HS and 36 after ACDF (OR = 1.45; 95% CI = 0.57 to 3.66; p = 0.44; I2 = 53%). Three studies also reported a total of 13 complications after HS and 6 after CDA [12, 13, 30] (OR = 1.24; 95% CI 0.40 to 3.79; p = 0.71; I2 = 0%). Neither was statistically significant.

Fig. 7
figure 7

Forest plots comparing complications after hybrid surgery (HS) versus a anterior cervical discectomy and fusion (ACDF) and b cervical disc arthroplasty (CDA)

Full size image

Discussion

Hybrid surgery comprised of ACDF and CDA for multilevel CDDD aims to overcome the poor outcomes and contraindications associated with its two component procedures used in isolation [1, 3, 4, 6, 7, 14, 19, 22, 26, 29, 33]. The main findings of this systematic review and meta-analysis are that HS may be associated with greater post-operative C2–C7 ROM, reduced ROM in the adjacent segments, and a quicker return to work than ACDF. CDA patients also returned to work quicker than HS patients. High-quality evidence is lacking, and large robust comparative studies are not available.

HS versus ACDF

Hybrid surgery was associated with a significantly greater post-operative C2–C7 ROM (p = 0.004), which echoes similar reviews [20, 28, 34, 35]. However, some of these included corpectomy procedures in some hybrid constructs [20, 35], which limits the usefulness of comparing their results with our own. Only three included studies reported adjacent segment ROM [17, 30, 32], with one case of ASD reported after HS and three following ACDF (Table 3). HS aims to reduce the incidence of ASD seen after ACDF while preserving cervical spine kinematics [4, 6, 14, 26], and we found significantly lower SAS ROM (p < 0.00001) and IAS ROM (p = 0.0005) after HS than after ACDF, as reported elsewhere in the literature [20, 34, 35].

Patients who had HS returned to work 32 days sooner than ACDF patients (p < 0.00001). There was no significant difference in patients’ functional and pain scores (NDI, VAS) at final follow-up but information in the months immediately following surgery is lacking. Return to work may be also be affected by post-operative care, with different types of collar worn for varying times between groups and between studies [12, 13]. Sociodemographic factors may influence this variable such as the type of work or the government support available. Unfortunately, patient professions are not reported by any paper, but the studies were conducted in different countries (Table 1), making differences in available support likely.

Unlike similar reviews, we found there was no significant difference in blood loss between included studies [20, 28, 34]. Although the ACDF element of HS has been associated with less blood loss than that of CDA [10], the results from meta-analyses of multilevel CDA versus ACDF have been equivocal [31, 39]. The greatest amount of blood loss was recorded by Kang et al. [17] perhaps because their 24 patients required 3-level intervention, whereas the majority in the other studies were 2-level. Jang et al. compared 3-level HS with 3-level ACDF but they did not measure intra-operative blood loss [15].

HS versus CDA

HS patients returned to work approximately 33 days sooner than the CDA group, with no significant difference in functional outcomes. As discussed in relation to HS versus ACDF, post-operative care and sociodemographic factors may influence this, although it is not clear why this strong effect is seen between HS and both its component procedures.

Heterogeneity between the studies may influence the ROM. The majority were 2-level procedures except 3/7 of the HS and 3/7 of the CDA procedures performed by Hey et al. were 3-level operations [13]; however, this represents a small percentage of the total number of HS and CDA operations. All three studies that compared HS with CDA used the ProDisc-C or MOBI-C discs (Table 1) but do not clearly state which were used for each procedure or the reasons behind choosing one over another [12, 13, 30]. A meta-analysis comparing the durability of CDA using different prosthetic devices found that MOBI-C and ProDisc-C were the best and second best, respectively [5]. Durability of the device has implications for long-term complications, although we identified no significant difference in complications. It is difficult to say how a different prosthesis may affect the ROM particularly within a hybrid construct.

Strengths and limitations

We excluded studies in which HS involved vertebral corpectomy, thus reducing potential confounding which such heterogeneity in the procedure may introduce. This is also the only review we could find which attempted to compare HS with both of its constituent procedures. Despite there being less evidence than multilevel ACDF or CDA, some surgeons choose to provide hybrid surgery to a select group of suitable patients who they feel would benefit. This review provides evidence in support of hybrid surgery, which may encourage more surgeons to use it which in turn would provide a greater pool of HS patients that are eligible for inclusion in future research.

A small number of studies met the inclusion criteria which, combined with the low n value, limited the statistical power. The studies also come to different conclusions as to which intervention is favoured. The quality of evidence from the included studies is very low (Appendix 1): most studies are observational with only one RCT, all of which were affected by bias to some degree. The small number of relevant studies means it was not possible to perform subgroup analyses for 3-level or 2-level surgery.

Future systematic reviews on the topic could utilise a network meta-analysis, to robustly ascertain the comparative effectiveness of the three interventions.

Conclusion

Hybrid surgery appears to be an effective treatment with improved post-operative C2–C7 ROM and less ROM in the adjacent segments compared with ACDF alone in the treatment of multilevel CDDD. The paucity of studies, high heterogeneity between studies, and low quality of evidence preclude strong recommendations in favour of HS over other interventions. Large randomised controlled trials under standardised settings and standardised surgical procedures are required to provide high-quality evidence that is currently lacking.