Secondary Surgery after Cervical Disc Arthroplasty versus Fusion for Cervical Degenerative Disc Disease: A Meta‐analysis with Trial Sequential Analysis

Abstract

The purpose of this meta‐analysis was to explore whether cervical disc arthroplasty (CDA) was superior to anterior cervical discectomy and fusion (ACDF) in reducing secondary surgery. PubMed, EMBASE, and the Cochrane Library databases were systematically searched. Outcomes were reported as relative risk (RR) with the corresponding 95% confidence interval (CI). The pooled data was calculated using a random‐effect model. We also used the trial sequential analysis (TSA) to further verify our results and obtain more moderate estimates. Twenty‐one studies with 4208 patients were included in this meta‐analysis. The results indicated that compared with ACDF, CDA had fewer frequency of secondary surgery at the index level (RR, 0.47; 95%CI, 0.36–0.63; P < 0.05) and adjacent level (RR, 0.48; 95%CI, 0.36–0.65; P < 0.05), and the differences were statistically significant. In addition, in terms of the overall frequency of secondary surgery at the index and adjacent level, CDA was also significantly superior to ACDF (RR, 0.49; 95%CI, 0.41–0.60; P < 0.05). TSA demonstrated that adequate and decisive evidence had been established. Regarding the frequency of secondary surgery, CDA was significantly superior to ACDF. It was supposed that CDA may be a better surgical intervention to reduce the rate of secondary surgery for patients with cervical degenerative disc disease.

Introduction

Anterior cervical discectomy and fusion (ACDF) is viewed as the gold‐standard surgical treatment for cervical degenerative disc disease1. However, this treatment is associated with a high rate of adjacent segment degeneration, which is a major reason for patients undergoing secondary surgery2. Secondary surgery includes hardware removal, revisions, supplemental fixations, and reoperations. It is considered an important clinical index after treatment. Furthermore, ACDF may result in motion loss at the operated segment, which may cause significant neck pain, poor function recovery, and cervical instability. All of these outcomes may lead to additional operations in turn3. Tao et al. reported in their study that in the ACDF group the rate of bone fusion was 100%. Davis et al. evaluated the safety and effectiveness of 105 patients with ACDF over 4 years of follow‐up, and the results showed that 18 patients (17.14%) underwent secondary surgeries2. Thomas et al. reviewed 22 patients undergoing ACDF and found that 6 patients had an additional operation with 84 months of follow‐up4.

To avoid loss of motion and slow down the adjacent‐level disc degeneration, cervical disc arthroplasty (CDA) has been developed, and is regarded as an alternative surgical method. It is a major non‐fusion surgical method, and can maintain disc space height and motion at adjacent segments. Several studies have compared the effectiveness and safety of ACDF and CDA for the treatment of cervical degenerative disc disease, and most of the results demonstrate that CDA is similar to ACDF, or even better in regard to some clinical and radiographic outcomes. However, most of those studies were evaluated as low quality, and may overestimate the effectiveness of CDA4, 5, 6, 7. In addition, a meta‐analysis conducted by Zhang et al. reported that the secondary procedure rates at the adjacent level showed no significant difference at mid‐term follow‐up8. Bartels et al. also demonstrated that the incidence of secondary surgeries for adjacent segment degeneration did not reach statistical difference9. However, fewer secondary surgeries were found in the CDA group compared with the ACDF group in some studies10, 11, 12. Specifically, a retrospective database analysis found a higher frequency of secondary surgeries in the CDA group (8%) compared with the ACDF group (2%)13. Consequently, whether CDA is superior to ACDF in terms of the incidence of secondary surgery remains controversial.

The aim of this study was to pool the current evidence and explore whether CDA is better than ACDF at reducing the need for secondary surgery.

Methods

Search Strategy

The PubMed, EMBASE, and Cochrane Library databases were systematically searched for literature up to 15 October 2017. The following search strategy was applied: (anterior cervical discectomy and fusion OR ACDF) AND (cervical disc arthroplasty OR CDA) AND (secondary surgery OR reoperation) AND (randomized controlled trial). Details of the search strategy are presented in Supplementary Appendix S1. Furthermore, the reference lists of manuscripts were also hand‐searched to make sure some studies that were not identified in our original search were also included in the present study. Moreover, there were no language restrictions.

Eligibility Criteria

Participants: Patients diagnosed with cervical degenerative disc disease underwent anterior cervical discectomy and fusion or cervical disc arthroplasty.

Interventions: Patients with cervical degenerative disc disease were treated by cervical disc arthroplasty.

Comparisons: In the control group, patients with cervical degenerative disc disease underwent anterior cervical discectomy and fusion.

Outcomes: All studies reported the outcome indicators regarding the frequency of secondary surgery, whether at index level or adjacent level.

Study design: Only randomized controlled trials were regarded as eligible in our study.

Exclusion Criteria

Abstracts, case reports, conference presentations, editorials, reviews, and expert opinions were excluded. All publications were limited to those involving human subjects and in the English language.

Data Extraction and Outcome Measures

Two reviewers independently extracted the data from the studies that were included in this meta‐analysis. Any disagreement was solved by discussion with the third reviewer. Data from those eligible studies were pooled to calculate the baseline characteristics of the patients and the frequency of secondary surgery.

For each included study, the following information was recorded: publication year, first author, study type, sample size, mean age, gender ratio, follow‐up time, and the type of the disc prosthesis. Furthermore, the secondary surgeries were defined as any hardware removal, revisions, supplemental fixations, and reoperations in this meta‐analysis, and we extracted the outcome of the frequency of secondary surgeries.

Risk of Bias Assessment

The risk of bias assessment of the included studies was performed using the Cochrane Collaboration’s “risk of bias” tool recommended in the Cochrane Handbook for Systematic Reviews of Interventions (version 5.1.0)14. This tool includes seven aspects: sequence generation (selection bias), allocation sequence concealment (selection bias), blinding of participants and personnel (performance bias), blinding of outcome assessment (detection bias), incomplete outcome data (attrition bias), selective outcome reporting (reporting bias), and other bias (baseline balance and fund). In addition, each of the aspects was ranked low risk of bias, high risk of bias, and unclear risk of bias.

Statistical Analysis

We used the Review Manager software (Version 5.3, The Nordic Cochrane Center, The Cochrane Collaboration, Copenhagen, Denmark) to analyze the data. Relative risk (RR) with 95% confidence interval (CI) was calculated. The statistical heterogeneity was assessed using the I ² statistic. When the I ² statistic was greater than 50%, heterogeneity was considered significant across studies and the meta‐analysis was performed in a random‐effects model. Otherwise, a fixed‐effects model was used. Subgroup analyses were performed based on the types of prosthesis (Bryan, Discover, Kineflex|C, Mobi‐C, PCM, Prestige LP, Prestige ST, ProDisc‐C, and SECURE‐C), duration of follow‐up (short term [1–3 years], mid‐term [4 or 5 years], or long‐term [6 years or longer]), and target level (single‐level or two‐level). To detect the possibility of publication bias, the funnel plot was applied. A P‐value below 0.05 was considered significant.

Trial Sequential Analysis

In a meta‐analysis, the risk of false positive errors (type I error) may arise. This phenomenon may result from random errors when a small amount of studies and participants are analyzed15 and with repetitive statistical testing of the accumulation of additional data. To correct for the incremental risk of type I errors, we employed trial sequential analysis (TSA) to identify whether the finding in a cumulative meta‐analysis is dependable and conclusive. TSA combines the required information size (RIS) with the trial sequential monitoring boundary, which adjusts the confidence intervals and decreases type I errors16. When the z‐curve traverses the trial sequential monitoring boundary or enters the futility area, sufficient and conclusive evidence may have been reached and further studies are not required. If any of the boundaries are not crossed by the z‐curve and the RIS has not been reached, it is inappropriate to make a conclusion.

We estimated a diversity‐adjusted required information size in accordance with the diversity of the intervention effect estimates among the included studies. The TSA was conducted to maintain a type I error of 1% with a power of 90%. In the present meta‐analysis, we calculated the required information size applying the estimates of the intervention effects of trials with adequate allocation concealment16, 17. Trial sequential analysis software version 0.9.5.10 beta (www.ctu.dk/tsa) was used to perform these analyses.

Results

Study Search

A total of 260 records were identified. After the duplicated literature was removed, 193 records were left. Then the titles and abstracts were screened, and 51 records were evaluated for eligibility. Finally, 21 studies 4, 5, 6, 7, 11, 12, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32 were included in this meta‐analysis. Information about the search process is provided in Fig. 1. A total of 4208 patients with cervical degenerative disc disease were randomized to the ACDF group (n = 1948) and the CDA group (n = 2260), respectively. The characteristics of those studies are shown in Table 1.

Figure 1.

Flow diagram of study selection.

Table 1.

Baseline characteristics of studies included in the meta‐analysis

Sources	Study characteristics	Prostheses	Number of patients		Mean age (years), female (%)	Number of cervical levels	Follow‐up (months)
			CDA	ACDF
Burkus 2014	USA, October 2002 to August 2004	Prestige ST Disc Prosthesis	276	265	43.59, 53.79	Single‐level	84
Coric 2011	USA, NA	Kineflex|C Disc Prosthesis	136	133	43.80, 59.11	Single‐level	24
Donk 2017	Netherlands,October 2003 to June 2010	Bryan disk prosthesis	50	47	43.62,46.39	Single‐level	108
Gornrt 2017	USA, NA	Prestige LP Disc Prosthesis	209	188	47.19, 54.16	Two‐level	24
Hisey 2016	USA, NA	Mobi‐C Cervical Disc Prosthesis	164	81	43.53, 53.47	Single‐level	48
Hou 2016	China, January 2008 to July 2009	Mobi‐C Cervical Disc Prosthesis	51	48	47.37, 41.41	Single‐level	61
Janssen 2015	USA, August 2003 to October 2004	ProDisc‐C Disc Prosthesis	103	106	42.81, 54.07	Single‐level	84
Loumean 2016	USA, NA	ProDisc‐C Disc Prosthesis	22	22	NA	Single‐level	84
Luo 2015	China, January 2009 to October 2011	Discover Disc Prosthesis	34	37	46.73, 46.48	Single‐level	48
Nabhan 2007	Germany, April 2004 and May 2005	ProDisc‐C Disc Prosthesis	25	24	44.00, 36.73	Single‐level	36
Pandey 2017	India, July 2012 to April 2014	Prestige LP Disc Prosthesis	17	17	39.7, 20.59	Single‐level	18
Phillips 2015	USA, January 2005 to December 2007	PCM Disc Prosthesis	218	185	44.57, 48.14	Single‐level	60
Radcliff 2016	USA, April 2006 to March 2008	Mobi‐C Disc Prosthesis	225	105	45.59, 52.15	Two‐level	60
Rozankovic 2017	Croatia, October 2008 to June 2010	Discover Disc Prosthesis	51	50	41.63, 50.50	Single‐level	24
Sasso 2011	USA, May 2002 to October 2004	Bryan Disc Prosthesis	242	221	44.54, 51.84	Single‐level	48
Sasso 2017	USA, NA	Bryan Disc Prosthesis	22	25	NA	Single‐level	120
Skeppholm 2015	Sweden, April 2007 to May 2010	Discover Disc Prosthesis	81	70	46.84, 51.66	Mixed‐level	24
Sundseth 2017	Norway, November 2008 to January 2013	Discover Disc Prosthesis	68	68	44.05, 53.68	Single‐level	24
Vaccaro 2013	USA, NA	SECURE‐C Disc Prosthesis	151	140	43.88, 48.80	Single‐level	24
Zhang 2012	China, May 2004 to May 2006	Bryan Disc Prosthesis	60	60	45.17, 44.17	Single‐level	48
Zhang 2014	China, February 2008 to November 2009	Mobi‐C Disc Prosthesis	55	56	45.76, 54.05	Single‐level	24

ACDF, anterior cervical discectomy and fusion; CDA, cervical disc arthroplasty; NA, not available.

Risk of Bias in the Included Studies

All the included studies were assessed as having “unclear risk of bias.” The main shortcomings in all studies were performance bias and detection bias. Because of the nature of these studies, blindness of these studies may not be achieved. Out of 21 studies, 15 documented detailed information about allocation concealment. Information on the methodological quality of all the included studies are exhibited in Fig. 2.

Figure 2.

Risk of bias assessment of each included study: (A) risk of bias summary and (B) risk of bias graph.

Frequency of Secondary Surgery at the Index Level

In this meta‐analysis, 15 studies, with a total of 3379 patients, reported the frequency of secondary surgery at the index level. In the CDA group, there were 71 patients (3.88%) undergoing secondary surgery at the index level; and in the ACDF group, the incidence rate of secondary surgery at the index level was 8.19% (127 out of 1550). By pooling the data, the result showed that the frequency of secondary surgery at the index level in the CDA group was significantly lower than that in the ACDF group (RR, 0.47; 95%CI, 0.36–0.63; P < 0.05; Fig. 3). Furthermore, the I ² was 40% and no significant heterogeneity was observed.

Figure 3.

Comparison of frequency of secondary surgery at the index level in participants undergoing CDA versus ACDF. ACDF, anterior cervical discectomy and fusion; CDA, cervical disc arthroplasty.

Frequency of Secondary Surgery at the Adjacent Level

Adjacent‐level secondary surgery was provided in 18 studies, with a total of 3568 patients (1923 patients in CDA groups, and 1645 patients in ACDF groups). The frequency of secondary surgery at the adjacent level in CDA groups was 62 out of 1923 (3.22%), and 114 out of 1645 (6.93%) in ACDF groups. The difference was significant (RR, 0.48; 95%CI, 0.36–0.65; P < 0.05; Fig. 4). In addition, there was no significant heterogeneity and the I ² was 0%.

Figure 4.

Comparison of frequency of secondary surgery at the adjacent level in participants undergoing CDA versus ACDF. ACDF, anterior cervical discectomy and fusion; CDA, cervical disc arthroplasty.

Frequency of Secondary Surgery at the Index and Adjacent Level

The pooled results showed that the CDA group had significant fewer frequency of secondary surgery at the index and adjacent level than the ACDF group (RR, 0.49; 95%CI, 0.41–0.60; P < 0.05; Fig. 5). There were 146 patients (6.46%) undergoing secondary surgery at the index and adjacent level in the CDA group, and in the ACDF group, the frequency of secondary surgery at the index and adjacent level was 254 (13.04%). In addition, there was no significant heterogeneity.

Figure 5.

Comparison of frequency of secondary surgery at the index and adjacent level in participants undergoing CDA versus ACDF. ACDF, anterior cervical discectomy and fusion; CDA, cervical disc arthroplasty.

Subgroup Analyses, Trial Sequential Analysis and Publication Bias

The results of subgroup analyses are presented in Supplementary Appendix S2. For the outcome of index‐level secondary surgery, the results were consistent in all except for the following subgroup analyses: the Bryan, Discover, and Kineflexl/C prosthesis, and the short‐term and long‐term follow‐up groups. In terms of frequency of secondary surgery at the adjacent level, the subgroup of Bryan, Discover, Kineflexl/C, Prestige LP, and SECURE‐C prosthesis, and the short‐term follow‐up and two‐level target level showed that there were no significant differences between the two groups. Furthermore, for the frequency of secondary surgery at the index and adjacent level, the subgroup analyses demonstrated that there were no significant differences between the two groups in the Bryan, Discover, Kineflex/C, and PCM prosthesis, and the short‐term follow‐up subgroups.

Trial sequential analysis showed that the cumulative z‐curve crossed both the conventional boundary and the trial sequential monitoring boundary for benefit, indicating sufficient and decisive evidence had been achieved for frequency of secondary surgery at the index level, the adjacent level, and both index and adjacent levels. Thus, further studies were not needed and were unlikely to change the inferences (Figs 6, 7, 8).

Figure 6.

Trial sequential analysis of 11 trials comparing CDA with ACDF for frequency of secondary surgery at the index level. Trial sequential analysis of 11 trials (black square fill icons) illustrating that the cumulative z‐curve crossed both the conventional boundary and the trial sequential monitoring boundary for benefit and the required information size had also been reached, indicating sufficient and conclusive evidence had been ascertained. A diversity‐adjusted required information size of 1609 patients was calculated using α = 0.05 (two‐sided), β = 0.10 (power 90%), a relative risk reduction of 56.63% based on trials with adequate allocation concealment, and an event proportion of 8.7% in the control arm. ACDF, anterior cervical discectomy and fusion; CDA, cervical disc arthroplasty.

Figure 7.

Trial sequential analysis of 14 trials comparing CDA with ACDF for frequency of secondary surgery at the adjacent level. Trial sequential analysis of 14 trials (black square fill icons) illustrating that the cumulative z‐curve crossed both the conventional boundary and the trial sequential monitoring boundary for benefit and the required information size had also been reached, indicating sufficient and conclusive evidence had been ascertained. A diversity‐adjusted required information size of 2192 patients was calculated using α = 0.05 (two‐sided), β = 0.10 (power 90%), a relative risk reduction of 45.49% based on trials with adequate allocation concealment, and an event proportion of 6.8% in the control arm. ACDF, anterior cervical discectomy and fusion; CDA, cervical disc arthroplasty.

Figure 8.

Trial sequential analysis of 15 trials comparing CDA with ACDF for frequency of secondary surgery at the index and adjacent level. Trial sequential analysis of 15 trials (black square fill icons) illustrating that the cumulative z‐curve crossed both the conventional boundary and the trial sequential monitoring boundary for benefit and the required information size had also been reached, indicating sufficient and conclusive evidence had been ascertained. A diversity‐adjusted required information size of 1703 patients was calculated using α = 0.05 (two‐sided), β = 0.10 (power 90%), a relative risk reduction of 50.5% based on trials with adequate allocation concealment, and an event proportion of 14.16% in the control arm. ACDF, anterior cervical discectomy and fusion; CDA, cervical disc arthroplasty.

The funnel plot was performed for all the outcomes. The funnel plot was visually reviewed and did reveal some asymmetry (Supplementary Appendix S2 [Fig. B.1‐3]).

Discussion

Anterior cervical discectomy and fusion has been regarded as the gold standard to alleviate the symptoms of symptomatic cervical degenerative disc disease. It can relieve patients’ pain significantly. However, more and more studies have found that there is a high rate of secondary surgery at the index and adjacent levels4, 33, 34. Given the shortcomings of ACDF, CDA was developed to retain as much intervertebral disc height and segmental activity as possible, and then reduce the incidence of secondary surgery. Zigler et al. reported that the reoperation rate was 2.9% in the CDA group compared with 14.5% in the ACDF group. There was a significant difference between them35. Another study demonstrated that the rate of secondary surgery was 27.3% after 7‐year follow‐up in the ACDF group, while in the CDA group, no secondary surgeries occurred4. However, the CDA also has its disadvantages, such as heterotopic ossification, which is a common occurrence after CDA and may result in adjacent segment degeneration and then require secondary surgery. Therefore, it remains controversial whether the rate of secondary surgery was lower in the CDA group than that in the ACDF group. To explore this question, we performed a meta‐analysis based on randomized controlled trials. The results of our study demonstrated that compared with ACDF, CDA had a significantly lower rate of secondary surgery.

In our meta‐analysis, 21 studies were included, and 15 studies with a total of 3379 patients reported the rate of secondary surgery at the index level. The results demonstrated that the frequency of secondary surgery at the index level was significantly lower in the CDA group than that in the ACDF group. A previous study demonstrated that symptomatic pseudarthrosis was a frequent reason for secondary surgery at the index level in the ACDF group, and the risk of pseudarthrosis was eliminated in the CDA group5. However, CDA has its shortcomings, such as device removal, which is regarded as the most common reason for secondary surgery at the index level24. It was conjectured that the removal of the index‐level device corrected adjacent‐level pathology, but did not consequentially rectify pathology at the index level. Furthermore, in our study, the results of subgroup analysis demonstrated that the difference in secondary surgery at the index level was not statistically significant between the two treatments in Bryan, Discover, and Kineflexl/C prosthesis groups. However, the number of studies included was not large enough, and more studies are needed in future analyses18, 36, 37.

Compared with ACDF, CDA had fewer secondary surgeries at the adjacent level in our meta‐analysis, which concurred with the findings of Xu et al. 38. Although the definite risk factor for secondary surgery is still controversial, some studies indicate a relationship between the occurrence of adjacent segment degeneration and secondary surgery39, 40. CDA was developed to maintain the index segment motion and to prevent abnormal stress on adjacent segments, which may be theoretically beneficial to delay the occurrence of adjacent segment degeneration in the long term, and decrease the rate of secondary surgery at the adjacent level22, 41. In vitro studies also support the above points42, 43. Studies have also indicated that ACDF may increase the stress on adjacent segments and lead to kinematic strain, disc degeneration, and mechanical instability. Michael et al. report that the incidence of adjacent segment degeneration was 37.1% in the CDA group at 5‐year follow‐up, while the incidence was 54.7% in the ACDF group5, which demonstrated that patients in the ACDF group may have a higher risk of undergoing secondary surgery at the adjacent level.

In this meta‐analysis, we provided evidence of a lower overall frequency of secondary surgery at the index and adjacent level in the CDA group compared with the ACDF group. It was indicated that CDA was superior to ACDF in regard to frequency of secondary surgery at the index and adjacent level. The findings of our meta‐analysis were consistent with previous meta‐analyses. However, our present meta‐analysis included 21 trials including 4208 patients. The present meta‐analysis is the latest and the most complete one, which generally coincides with and further enhances the findings of earlier meta‐analyses. Furthermore, TSA was used to provide more conservative estimates. TSA suggested that the present meta‐analysis established adequate and decisive evidence. The present findings may be beneficial for guiding the choice of surgical procedures for clinicians and patients with cervical degenerative disc disease. Nevertheless, because of the inherent limitations in the original studies, this conclusion should be interpreted cautiously.

Although the methodology recommend by the Cochrane Collaboration was used in this meta‐analysis to make our results more credible, there are also some limitations that cannot be ignored. First, although subgroup analyses of different types of prostheses were performed, the number of studies with certain types of prostheses was relatively small, which may affect the reliability of subgroup analyses. Second, according to the results of the funnel plots, there was positive‐outcome bias in the studies. This means that more studies with negative outcomes may be conducted but not published.

Conclusions

In this meta‐analysis, the results indicate that CDA is superior to ACDF in regard to the frequency of secondary surgeries at the index level, the adjacent level, and both index and adjacent levels. However, subgroup analyses demonstrate that no significant differences were found between the two interventions in some subgroups. Hence, more studies of high quality are needed to confirm these conclusions.

Supporting information

Appendix S1 Search strategies

Appendix S2 Subgroup analyses of cervical disc arthroplasty compared with anterior cervical discectomy and fusion