Incidence of Heterotopic Ossification at 10 years After Cervical Disk Replacement

Study Design.

Systematic review and meta-analysis.

Objective.

This study aimed to assess the incidence of heterotopic ossification (HO) 10 years after cervical disk replacement (CDR).

Summary of Background Data.

HO is a common complication after CDR and may limit the range of motion of the artificial disk. As HO usually progresses slowly, a long-term follow-up is required to better understand its incidence. In recent years, the increasing number of original articles reporting 10-year outcomes gives us the opportunity to better understand the long-term incidence of HO.

Materials and Methods.

We searched PubMed, Medline, Embase, and Cochrane Library databases to identify eligible studies. The incidence of HO was pooled, and subgroup analysis was performed. Meta-regression analyses were conducted to identify factors contributing to heterogeneity.

Results.

Eleven studies with at least 10 years of follow-up comprising 1140 patients who underwent CDR were included. The pooled incidence of overall HO was 70% (95% CI, 60%–81%) at 10 years postoperatively, 60% (95% CI, 44%–75%) at five or six years postoperatively, and 50% (95% CI, 27%–72%) at one or two years postoperatively. The pooled incidence of severe HO (grade 3 or 4) was 37% (95% CI, 29%–45%), and mild HO (grade 1 to 2) was 30% (95% CI, 17%–44%) at 10 years of follow-up. Pooled range of motion decreased from 8.59° before surgery to 7.40° 10 years after surgery. Subgroup analysis showed that HO incidence differed according to the prosthesis type. The earlier publication was associated with a higher pooled incidence of severe HO in the meta-regression analysis.

Conclusions.

This is the first meta-analysis providing detailed information on the pooled 10-year incidence of HO after CDR. The incidence of HO seems to increase with the length of follow-up.

Level of Evidence.

3.

Anterior cervical discectomy and fusion (ACDF) is a classic surgical treatment for cervical spondylosis. However, it alters the biomechanics of adjacent segments, which may be one of the reasons for adjacent segment degeneration (ASD).1,2 Cervical disk replacement (CDR), as an alternative to ACDF, preserves surgical segment range of motion (ROM) and reduces the risk of ASD.3–6 However, heterotopic ossification (HO), a complication of CDR, is a cause for concern. Some studies7–9 suggest that HO limits the ROM of the prosthesis, causing dysfunction and even becoming an “expensive fusion device.” Therefore, HO may decrease the longevity of prostheses.10 In addition, a meta-analysis11 has shown severe HO impact on the visual analog scale for pain, but other studies12,13 have debated this.

HO incidence is time dependent and varies greatly in different studies, ranging from 100%13 in at least 24 months of follow-up to 0%14 in at least 60 months of follow-up. However, there is no systematic review focusing on the 10-year incidence of HO at present. Moreover, different prostheses have different biomechanical effects, which may also affect the incidence of HO. In addition, the Food and Drug Administration-approved Investigational Device Exemption (IDE) clinical trials reported lower HO rates than several independent nonconflicted studies.15

There are three previously published systematic reviews,15–17 which do not have data sufficient for the assessment of the long-term incidence of HO. There are currently 15 studies with 10-year results published since the last performed meta-analysis in 2020. Seven of these were simply matured to 10 years, and eight were newly published. This study aimed to systematically review studies reporting HO incidence with at least 10 years of follow-up after CDR.

MATERIALS AND METHODS

This systematic review and meta-analysis was registered on PROSPERO (CRD42021257276) and was carried out in accordance with the Preferred Reporting Items for Systematic Review and Meta-Analysis Protocols updated in 2020.18 The guidelines for academic spine surgeons19,20 were used in this study.

Search Strategy

The PubMed, Medline, Embase, and Cochrane Library databases were searched in December 2021. The following words related to CDR and HO were used: “cervical,” “disc replacement,” “disk replacement,” “arthroplasty,” “heterotopic ossification,” “heterotopic bone,” and “paravertebral ossification.” There was no language restriction in the process of article retrieval. The reference lists from all publications retrieved were reviewed to identify any potentially related studies.

Eligibility Criteria

Two independent reviewers searched the literature, removed duplicates, and eliminated irrelevant articles by browsing the titles and abstracts. Full-text articles were then evaluated again based on the inclusion and exclusion criteria. If more than one article included the same series of patients, the articles with the latest and most comprehensive data were included in the analysis. When the opinions of the two authors contradicted each other, a consensus was reached through a discussion. When there was no consensus, senior authors were consulted.

Studies that met the following criteria were included: a prospective or retrospective study reporting the incidence of HO after CDR: participants were aged above 18; patients underwent surgery for a cervical disk degeneration disease or disk herniation with myelopathy, radiculopathy, or both; with no more than three operative levels; were followed up for at least 10 years; and the study used McAfee,21 Mehren,22 or an improvement23 of McAfee classification to grade HO. Studies that met the following criteria were excluded: hybrid surgery with fusion, participants with trauma, infection, malignancy, and previous surgical procedures at the involved level. The following exclusion criteria were also applied: review studies, case reports, editorials, laboratory research, or cadaver studies; duplicated records; the incidence of HO was not reported, or the data were incomplete; and research focusing on the lumbar spine.

Data Extraction

The following data were extracted independently by two reviewers: author’s name, year of publication, funds, follow-up time, sample size, surgical indication, age, prosthesis type, surgical levels, ROM, HO rate, and ASD rate. HO rate was recorded according to the number of cervical levels. Grades 1 and 2 were defined as mild HO, and grades 3 and 4 were defined as severe HO. Only ASD was included and recorded according to the number of patients. We attempted to contact the corresponding authors if the data in the articles were not expressed in the format required for extraction.

Quality Assessment

For single-arm and nonrandomized trials, methodological quality was assessed using the Methodological Index for Non-Randomized Studies.24 The quality assessment was completed by two authors independently, and any disagreement not solved after discussion was decided by the senior author.

Statistical Analysis

Statistical analysis was conducted with STATA 15.0 (Stata Corp LP, College Station, TX). A P-value of < 0.05 was considered statistically significant. The heterogeneity of the studies was assessed using the I ² test. If I ² was >50%, the random-effects model was used. Otherwise, the fixed-effects model was used. Subgroup analysis was performed according to the study design, prosthesis type, and funding source. Univariate meta-regression analysis was conducted to find the source of heterogeneity. Begg test, Egger test, and funnel plot were used to evaluate the publication bias.

RESULTS

Study Selection

A total of 822 records were identified through a database search and three through the previous systematic review reference list. After removing duplicates and screening titles and abstracts, 157 full-text articles were identified to be evaluated for eligibility (Figure 1). A total of 11 studies7,8,25–33 were included for qualitative and quantitative synthesis. Among these studies, 10 reported overall HO, 11 reported severe HO, and nine reported mild HO.

Figure 1.

Flow diagram of the selection of the studies. HO indicates heterotopic ossification.

Study Demographics

The demographics of the studies are summarized in Table 1. A total of 1140 patients were included from six prospective and five retrospective studies. Five studies focused on single-level CDR and six on multilevel CDR. C5/6 was the most frequently reported cervical segment (55.5%), followed by C6/7 (21.3%), and then C4/5 (18.3%). Eight studies were not IDE studies and declared that they did not receive funding from other institutions; hence, they were classified as independent studies in this meta-analysis. Prestige LP was the most used prosthesis (43.3%), followed by Mobi-C (20.2%), Bryan (19.5%), and ProDisc-C (17.0%).

TABLE 1.

Demographics of Included Studies

References	Study design	Surgical indication	Follow-up time (years)	Follow-up rate (%)	Age, (yr) (mean ± SD)	No. patients	Funding sources	Prosthesis types	Operative levels	Surgical levels distribution	Studies centers	ROM at 10 yr (°)	HO classification system	HO incidence at 10 yr (%)	ASD incidence at 10 yr (%)
Mehren et al 7	Prospective	Symptomatic cervical disk disease; unresponsive nonoperative treatment; soft disk herniation; cervical spinal stenosis/hard disk; degeneration in fusion cases; level C3–C7; loss of intervertebral space height of <80	10.2	80.90	44.8 (range 25–67)	50	Independent	Prodisc-C	1, 2, 3	3 at C3/4 10 at C4/5 28 at C5/6 29 at C6/7	1	7.6 ± 10.3	Mehren	90	34.2
Xu et al 25	Retrospective	Cervical disk disease; unresponsive nonoperative treatment; without severe facet degeneration, OPLL or surgical segmental instability	11.3	Not reported	46.84 ± 9.39	118	Independent	Prodisc-C	1, 2, 3	10 at C3/4 32 at C4/5 90 at C5/6 22 at C6/7	1	7.3 ± 3.55	McAfee	87	NA
Lobo et al 26	Retrospective	Symptomatic cervical disk disease; level C5–C7; unresponsive nonoperative treatment	10	Not reported	Bryan group:45.4±1.49; Prestige LP: 37.81±6.97	15	Independent	Bryan or Prestige LP	1	10 at C5/6 5 at C6/7	1	NA	McAfee	60	40
Kim et al 8	Prospective	Symptomatic cervical disk disease; level C3–C7; unresponsive nonoperative treatment; disk height >3−mm	10	81	NA	231	Dependent	Mobi-C	1, 2	C3–C7	9	NA	McAfee or Mehren	The incidence of grade 3/4 was 37.6	14.1
Gornet et al 27	Prospective	Cervical disk disease; level C3–C7; unresponsive nonoperative treatment	10	83.30	44.5 ± 8.8	276	Dependent	Prestige LP	1	C3–C7	20	6.85 ± 4.96	Mehren	66.2	NA
Gornet et al 28	Prospective	Cervical disk disease; level C3–C7; unresponsive nonoperative treatment	10	86	HO 0–2 grade: 46.0 ± 8.6; HO 3–4 grade: 9.0 ± 7.6	209	Dependent	Prestige LP	2	78% of the patients were treated at C5/6 and C6/7	29	NA	Mehren	55.8	NA
Genitiempo et al 29	Retrospective	Level C3–C7; radiculopathy due to soft disk herniation; without spondylosis myelopathy; without OPLL	18	79.16	42.7±12.7	72	Independent	Bryan	1	1 at C3/4 13 at C4/5 26 at C5/6 17 at C6/7	Institutional databases	8.9 ±2.2	Mehren	50.6	77.2
Song et al 30	Prospective	Cervical disk herniation or degenerative cervical canal stenosis between C3/4 and C6/7; unresponsive nonoperative treatment; disk height loss <50%, and segmental range of motion more than 2°; without bridging osteophytes, OPLL, kyphosis or severe facet joint degeneration	10	78	55.69 ± 8.32	91	Independent	Bryan	1	2 at C3/4 14 at C4/5 45 at C5/6 9 at C6/7	1	8.6 ± 5.3	Improved McAfee	93	46.5
Zhao et al 31	Retrospective	Cervical myelopathy or radiculopathy; unresponsive nonoperative treatment; without OPLL; disk height loss <50%, and segmental range of motion more than 4°;	10.3	Not reported	44 (range 30–58)	27	Independent	Prodisc-C	1	2 at C3/4 4 at C4/5 16 at C5/6 5 at C6/7	1	6.6 ± 3.5	Mehren	74	50
Zhao et al 32	Retrospective	Cervical myelopathy or radiculopathy; unresponsive nonoperative treatment; without OPLL; without severe spondylosis at the level to be treated	10	Not reported	44.8 (range 27–54)	33	Independent	Bryan	1, 2, 3	3 at C3/4 7 at C4/5 26 at C5/6 6 at C6/7	1	4.7 ± 4.2	McAfee	69	47.6
Pointillart et al 33	Prospective	Cervical disk herniation or symptomatic cervical spondylosis; unresponsive nonoperative treatment; without OPLL	15.5	81.80	46.2 (range 26–65)	18	Independent	Bryan	1, 2	2 patients at C4/5 11 patients at C5/6 9 patients at C6/7	1	NA	Mehren	54.5	59.1

ASD indicates adjacent segment degeneration; HO, heterotopic ossification; NA, not available; OPLL, ossification of the posterior longitudinal ligament; ROM, range of motion.

Study Quality Assessment

We only used the first eight criteria of Methodological Index for Non-Randomized Studies (total score=16 points) as there were no comparative studies. The average score was 10.3 points (range: 9–12). All included studies clearly stated their aim and included consecutive patients. Their endpoints and follow-up periods were appropriate to the aims of the studies. However, most did not explain whether the evaluation process used a blinded method, nor did they explain the sample size calculation. In general, the included studies were of medium quality. The detailed results of the studies’ quality assessment are summarized in Table 2.

TABLE 2.

Methodological Index for Nonrandomized Studies (MINORS) of Included Studies

	Mehren et al 7	Xu et al 25	Lobo et al 26	Kim et al 8	Gornet et al 27	Gornet et al 28	Genitiempo et al 29	Song et al 30	Zhao et al 31	Zhao et al 32	Pointillart et al 33
A clearly stated aim	2	2	2	2	2	2	2	2	2	2	2
Inclusion of consecutive patients	2	2	2	2	2	2	2	2	2	2	2
Prospective collection of data	2	1	1	2	2	2	1	2	1	1	2
Endpoints appropriate to the aim of the study	2	2	2	2	2	2	2	2	2	2	2
Unbiased assessment of the study endpoint	0	0	0	1	0	0	0	0	0	0	0
Follow-up period appropriate to the aim of the study	2	2	2	2	2	2	2	2	2	2	2
Loss to follow-up <5%	1	0	0	1	1	1	1	1	0	0	1
Prospective calculation of the study size	0	0	0	0	0	0	0	0	0	0	0
Total	11	9	9	12	11	11	10	11	9	9	11

The items are scored 0 (not reported), 1 (reported but inadequate), or 2 (reported and adequate). The global ideal score is 16 for noncomparative studies.

HO Incidence

A total of 10 studies reported an overall HO rate of 960 cervical levels based on at least 10 years of follow-up after CDR. The pooled incidence was 70% for overall HO (95% CI, 60%–81%) at 10 years postoperatively (Figure 2A), 60% (95% CI, 44%–75%) at five or six years postoperatively (Figure 2B), and 50% (95% CI, 27%–72%) at one or two years postoperatively (Figure 2C). Considering HO severity, the pooled incidence was 37% for severe HO (grade 3–4) (95% CI, 29%–45%) (Figure 3A) and 30% for mild HO (grade 1–2) (95% CI, 17%–44%) (Figure 3B) at 10 years of follow-up.

Figure 2.

The pooled incidence of overall heterotopic ossification at 10 years postoperatively (A), five or six years postoperatively (B), and one or two years postoperatively (C).

Figure 3.

The pooled incidence of severe HO (A) and mild HO after 10 years of follow-up (B). HO indicates heterotopic ossification.

Subgroup analyses were performed based on study design, prosthesis type, and funding source. There was no significant difference in pooled overall HO between prospective (pooled rate: 72%, 95% CI, 56%–89%) and retrospective (pooled rate: 68%, 95% CI, 53%–84%) studies (Figure 4A). In addition, ProDisc-C had the highest pooled overall HO rate (pooled rate: 86%, 95% CI, 81%–91%) at 10 years of follow-up. Bryan (pooled rate: 64%, 95% CI, 42%–86%) and Prestige LP (pooled rate: 62%, 95% CI, 53%–71%) prostheses had a lower pooled overall HO rate (Figure 4B). As for the subgroup analysis of funding sources, the overall HO of independent studies (pooled rate: 74%, 95% CI, 64%–84%) was higher than that of dependent studies (pooled rate: 61%, 95% CI, 51%–71%) (Figure 4C).

Figure 4.

Subgroup analyses of the incidence of overall heterotopic ossification based on study design (A), prosthesis type (B), and funding source 10 years postoperatively (C).

ROM and Incidence of ASD

Eight studies reported ROM at the surgical level preoperatively and 10 years postoperatively. The pooled ROM was 8.59° (95% CI, 7.10°–10.08°) preoperatively, and it remained at 7.40° (95% CI, 6.46°–8.34°) after 10 years of follow-up (Figure in Supplementary Material 1, Supplemental Digital Content 1, http://links.lww.com/BRS/C78). A total of eight studies reported the incidence of ASD at 10 years postoperatively. One study only reported severe ASD8. The pooled incidence of ASD was 45% (95% CI, 27%–63%) (Figure in Supplementary Material 2, Supplemental Digital Content 2, http://links.lww.com/BRS/C79).

Meta-regression

Meta-regression analyses were performed to determine sources of heterogeneity and to explore the effects of the potential variables related to HO. More recent studies had a lower incidence of severe HO (grade 3–4) (P=0.015). No significant differences were found in age, sex, study design, prosthesis type, funding source, CDR at C5/6, ROM, or ASD rate. Detailed results are shown in Table 3.

TABLE 3.

Meta-regression Analysis for the Incidence of HO

	Meta-regression coefficient	95% CI	P
Overall HO
Year of publication	−0.06	−0.19 to 0.07	0.301
Age	0.02	−0.02 to 0.06	0.218
Male sex	−0.19	−2.32 to 1.95	0.845
Design (retrospective vs. prospective)	−0.07	−0.42 to 0.27	0.645
Prosthesis type
Prestige LP vs. others	−0.13	−0.54 to 0.27	0.470
Prodisc-C vs. others	0.27	−0.07 to 0.61	0.106
Bryan vs. others	−0.14	−0.52 to 0.24	0.431
Funding source	0.137	−0.29 to 0.56	0.487
CDR at C5/6	−0.78	−2.33 to 0.77	0.264
ROM-pre	0.017	−0.14 to 0.18	0.788
ROM 10-year	−0.05	−0.39 to 0.29	0.706
ASD	−0.34	−2.03 to 1.34	0.634
Severe HO
Year of publication	−0.27	−0.47 to −0.06	0.015 *
Age	0.01	−0.10 to 0.11	0.906
Male sex	−0.38	−4.89 to 4.13	0.853
Design (retrospective vs. prospective)	−0.24	−0.92 to 0.45	0.460
Prosthesis type
Prestige LP vs. others	−0.03	−0.89 to 0.82	0.930
Prodisc-C vs. others	0.37	−0.46 to 1.17	0.347
Bryan vs. others	−0.33	−1.09 to 0.42	0.346
Mobi-C vs. others	0.15	−1.17 to 1.46	0.805
Funding source	0.10	−0.70 to 0.90	0.787
CDR at C5/6	−0.58	−4.10 to 2.93	0.698
ROM-pre	−0.11	−0.60 to 0.37	0.567
ROM 10 yr	−0.31	−0.70 to 0.07	0.091
ASD	−0.80	−3.43 to 1.82	0.492

^* P-value<0.05.

ASD indicates adjacent segment degeneration; CDR, cervical disk replacement; HO, heterotopic ossification; ROM, range of motion.

Publication Bias

Begg test and Egger test were used to identify negligible publication bias of overall HO (Begg test, P=0.721, Egger test, P=0.495) and severe HO (Begg test, P=0.350, Egger test, P=0.190). Funnel plots are shown in Figure 5A, B. Owing to the heterogeneity of the included studies, some studies were outside of the funnel.

Figure 5.

Funnel plots of the meta-analysis for overall heterotopic ossification (HO) (A) and severe HO (B).

DISCUSSION

HO as a complication after CDR may progress over time. Therefore, short-term follow-up is not sufficient to evaluate its incidence. This study is the first systematic review and meta-analysis to determine the incidence of HO 10 years after CDR. A total of 11 studies with 1144 patients were included. We provided detailed information on the pooled 10 years incidence of HO and analyzed the role of prothesis, funding source, and study design.

The incidence of overall and severe HO 10 years after CDR were 70% and 37%, respectively. Three systematic reviews have analyzed the incidence of HO; however, they included very few studies with a follow-up period of >10 years. Kong et al 16 observed that the overall incidence of HO 5 to 10 years after CDR was 53.6%, and that of severe HO was 47.5%. The incidence of severe HO is higher than that of this meta-analysis due to the small sample size, only including 120 patients in four 5–10-year studies, which made it difficult to evaluate the long-term incidence of severe HO. Similarly, Hui et al 17 observed a HO incidence of 24.8% one to two years postoperatively and 50.1% over six years postoperatively. However, they did not report the incidence at 10 years. Dowdell et al 15 found the incidence of serious HO to be 22.8% (a meta-analysis method was not used). Compared with independent studies, they observed that IDE studies underreported the incidence and severity of HO. Nevertheless, they included only one 10-year follow-up study. In addition, we observed that the incidence of HO increased with the length of follow-up. This progression can be observed in almost every original study. However, we are still unable to observe the end of this progression. In addition, the pooled incidence of HO showed significant heterogeneity. This heterogeneity may be due to differences in the determination of HO between studies. In 2003, McAfee et al 21 first proposed the classification of HO after lumbar artificial disk replacement, and in 2006, Mehren et al 22 proposed HO classification after cervical artificial disk replacement based on the McAfee classification. Although these are two of the most commonly used scales today and are similar, there are still some differences between them. The Mehren classification22 grade I emphasizes ossification occurring at the anterior edge of the vertebral body, while the McAfee21 classification grade I does not limit HO at the anterior or posterior edge of the vertebral body. Mehren may underestimate the incidence of HO. On the other hand, the modified McAfee classification23 could potentially increase the incidence of HO diagnoses. Moreover, different observers may have different evaluation results, although they use the same scale. In addition, heterogeneity may also come from different prostheses, research regions or ethnic groups, and different funding sources. Therefore, we conducted subgroup analyses and meta-regression for these parameters. Furthermore, a high degree of ossification before surgery,34 insufficient endplate coverage of the prosthesis,35–37 intervertebral space height change before and after surgery ≥1.8 mm,35 excessive intervertebral space distraction,38 and disk space angle change during operation >5°39 may lead to higher incidences of HO. The above patient characteristics and surgical details may vary between studies. Unfortunately, no specific data on these factors are reported in the included studies, which makes it difficult for us to further analyze. Therefore, readers should carefully interpret these results.

Different types of prostheses have different biomechanical properties, which may affect HO incidence. The natural motion of the cervical spine includes six degrees of freedom, with translations and rotations around three independent axes.40 Fixed core prostheses (e.g. ProDisc-C) only have rotations in three degrees of freedom without translations. This may cause greater stress between the prosthesis and the endplate and lead to bone chips, contributing to HO development. In contrast, an unconstrained prosthesis (e.g. Bryan) and a semiconstrained prosthesis (e.g. prestige LP) allow segmental motion with translations and rotations in six degrees of freedom, which better simulates the natural motion of the cervical spine. Yi et al 41 compared the HO incidence of ProDisc-C, Mobi-C, and Bryan prostheses in a retrospective study of 170 patients with at least 12 months follow-up and observed that Bryan had the lowest HO incidence, while ProDisc-C had the highest, similar to our results. A previous study,42 with an average of four years of follow-up, found that ProDisc-C had the highest HO incidence, followed by Mobi-C, and Bryan had the lowest HO incidence, which is consistent with this meta-analysis. A previous meta-analysis17 also observed that ProDisc-C had the highest grade 3 to 4 HO incidence compared with other prostheses. Although the results of the prosthesis subgroup analysis in our meta-analysis are consistent with the above studies, the CI of the Prodic-C and Bryan subgroups overlap (Prodic-C: 0.86, CI, 0.81–0.91 vs. Bryan 0.64, CI, 0.42–0.86). In addition, there was no significant difference in the meta-regression, potentially due to aggregate bias and the number of studies on prostheses limiting the statistical efficiency. Therefore, readers need to interpret this result carefully. Furthermore, patient characteristics and surgical details may confound prosthesis selection. However, as previously mentioned, these factors were not available in the included studies.

The most immediate purpose of CDR is to preserve segmental ROM. However, HO has been shown to decrease ROM7–9, especially high-grade HO. Few studies have summarized ROM outcomes in the 10 years following CDR. We found ROM decreased 1.19°, but remained at 7.40°, 10 years after surgery. Since the included studies rarely reported ROM per year after surgery, it is difficult to observe trends. Meta-regression analysis demonstrated that HO had no significant effect on ASD and ROM. As meta-regression can only include the average ROM of each study rather than the ROM of each patient, this will lead to aggregation bias. For instance, patients with severe HO may experience reduced ROM, while the patients without HO may experience larger ROM. However, meta-regression can only include the average ROM of each study, making the difference undetectable. Therefore, although we do not find significant differences in meta-regression, we should remain neutral about their relationship. In addition, there are currently no studies comparing the long-term results between CDR with ROM-limited HO and primary fusion surgery, which may be a direction of future research. Furthermore, it should be clarified that the included studies did not provide detailed results on patient-reported outcome measures according to HO/non-HO groups. Therefore, this study does not allow for the analysis of the relationship between HO and patient-reported outcome measures. The revision rates after CDR are lower than after ACDF, according to previous studies. Gornet et al 9 found the rate of index segment reoperation (4.7% vs. 17.6%) and adjacent segment reoperation (9.0% vs. 17.9%) was significantly lower in the CDR group than in the ACDF group 10-year after surgery. Radcliff et al 43 found that the incidence of secondary surgery at the index level was significantly lower in the CDR group than in the ACDF group (4.4% vs. 16.2%, P=0.001) seven years postoperatively in a multicenter RCT with 599 patients, and the same results were also obtained at the adjacent segment (4.4% vs. 11.3%, P=0.03).

CDR still seems to have a lower incidence of ASD compared with ACDF in long-term follow-up studies. Ghobrial et al 44 and Lavelle et al 45 found the incidence of symptomatic adjacent level disease requiring surgery was significantly lower in the CDR group compared with the ACDF group seven years postoperatively (6.9% vs. 11.7%, P=0.23) in an RCT with 494 patients. At 10 years of follow-up, the rates were 9.7% and 15.8%, respectively, although this difference did not reach a significant difference at 10 years (P=0.146). Lei et al 46 also found a significantly lower incidence of ASD in the CDR group than in the ACDF group (28.6% vs. 58.6%) in a controlled study of 66 people with eight years of follow-up. Phillips et al 6 found the incidence of ASD in the superior segment was significantly higher in the ACDF group than in the CDR group (33.1% vs. 50.9%, P=0.006) seven years after surgery in an RCT with 110 patients and that ASD was the main reason for secondary surgery in the ACDF group. Boody et al 47 found a higher incidence of high-grade adjacent segment ossification in the CDR group compared with the ACDF group (68.2% vs. 11.1%). However, some studies48,49 still show similar rates of distant ASD for both procedures. Although the cause of ASD is unclear, several studies have made observations. Yang et al 50 observed that ASD incidence was comparable between fusion and arthroplasty groups in a cohort of 253 patients, and ASD was unrelated to ROM in the arthroplasty group. Hilibrand et al 51 demonstrated that the effect of natural degeneration should not be ignored in ASD development. Further, Matsumoto et al 52 observed that >60% of 201 asymptomatic volunteers had cervical degenerative changes within 10 years, and the incidence of ASD at C5/6 was similar between operation and control groups. Similarly, Park et al 53 observed that 85% of patients with ASD had the condition before anterior cervical surgery, indicating that preoperative degeneration increased the risk of ASD. In a 10-year follow-up magnetic resonance imaging study, Pesce et al 54 divided 71 patients with surgical indications into the ACDF group and a nonoperation control group and observed no difference in ASD incidence between the groups.

CDR was in the exploratory stage over a decade ago, and there was a limited experience on how to avoid HO during surgery. In this meta-analysis, the higher pooled incidence of severe HO was related to an earlier year of publication. It is possible that as surgeons have become more aware of the factors influencing HO, studies performed more recently take into account approaches to reduce HO, such as selecting patients with less degeneration and maintaining a certain level of endplate coverage intraoperatively. We also conducted a subgroup analysis of funding sources and observed that the pooled incidence of overall HO in dependent studies was lower than those reported by independent studies. However, given concerns for aggregation bias and only 2 studies being dependent, this finding should be interpreted with caution.

This study had some limitations. First, we observed significant heterogeneity in our study, which persisted in the subgroup and meta-regression analyses. Second, the follow-up rate in the included studies was low (range: 78%–86%); however, this may be an inevitable problem in long-term follow-up investigations. Third, several factors influence HO development or contribute to heterogeneity, such as obesity, positioning of the implant, endplate sclerosis, diagnosis, number of fusion levels, and revision status. However, few studies report these factors, which obviates the capacity for further analysis. Finally, four of the included studies reported that a few patients were converted to fusion during follow-up. Because of the nature of the data, it is difficult to separate out these patients, but we emphasize that such cases only accounted for a small proportion of the overall cohort.

CONCLUSIONS

The literature included in this study is of medium quality, and several analyzed variables have obvious heterogeneity. This study provided detailed information on HO incidence 10 years after CDR. The incidence of HO increased with the length of follow-up.

Key Points

• The combined incidence of overall HO at 10 years postoperatively was 70%, and the combined incidence of severe HO was 37%, which seems to be higher than we expected. However, this is subject to a large heterogeneity and needs to be interpreted with caution by the reader.

• ProDisc-C had the highest pooled overall HO rate (86%) at 10 years of follow-up. Bryan (64%) and Prestige LP (62%) prostheses had a lower pooled overall HO rate.

• The pooled incidence of overall HO was higher in independent studies (74%) compared with that in dependent studies (61%).

• The incidence of HO increased with the extension of follow-up time. According to the current original study, the endpoint of HO incidence progression could not be determined.

Supplementary Material

brs-48-e203-s001.eps ^{(2.2MB, eps)}

brs-48-e203-s002.eps ^{(2.1MB, eps)}