The incidence of postoperative heterotopic ossification in cervical disc arthroplasty--A systematic review and meta-analysis

Background: Heterotopic ossification has become a common disease after CDA, which has attracted much attention. Understanding the law of its occurrence and development can provide theoretical basis for the formulation of HO prevention strategies. To acquire the objective data and the change rule of HO incidence, a meta-analysis of all available evidence was performed. Methods: A search of the literature was conducted on Pubmed/MEDLINE, EMBASE, and Web of Science. Relevant studies including incidence-relative data of HO were selected according to eligibility criteria. Results: 52 studies were eligible and finally included and the quality assessment showed a relative high score of them. The results of the analysis reminded us that the incidence of HO increased with the extension of follow-up time, both based on the number of cases and the number of segments, and the increase mainly occurred after 6 years. Grade III-IV HO showed us a positive correlation with follow-up time (R2=0.218), while Grade I-II HO did not change a lot in different follow-up time points. Conclusions: The follow-up time after CDA should be long enough, so as to ensure the true clinical results. HO occurs all the postoperative time and Grade I-II HO will gradually develop into Grade III-IV, while continuous new HO keeps the number of Grade I-II in a dynamic balance. Through this study, we can preliminarily define the relatively objective incidence and change rule of HO, which provides data basis and theoretical basis for the future research of HO prevention strategy.

In this medical background, the concept of non-fusion is particularly valuable. At first, clinical scholars tried to apply artificial disc and artificial nucleus pulposus in the treatment of lumbar disc degeneration, but the clinical effect was not significantly different from that of fusion surgery [4]. Considering the significant impact of cervical mobility on quality of life, more research was devoted to the study of artificial cervical disc or cervical dynamic stabilization device. Starting with Bryan disc of Medtronic, a series of bionic artificial prosthesis are coming into the market one by one, and all of them have achieved good clinical effects. Because the artificial disc can maintain the range of motion of the surgical segment, it can prevent the occurrence of ASD to some extent [5].
However, with the deepening of research and clinical follow-up, the heterotopic ossification (HO) after cervical disc arthroplasty (CDA) compromised the clinical value of this emerging operation greatly. HO is defined as the process by which trabecular bone forms outside of the normal skeletal structure, occupying space in soft tissue where it shouldn't exist, which was firstly reported in the total hip arthroplasty-related postoperative complication [6,7]. Presently, the mechanism of HO is not clear, and there is no well-accepted effective prevention strategy. In most of these researches, they would regard ACDF as a reference but have overlooked the uniqueness of CDA. Keeping physiological ROM of neck for patients is the primary destination when CDA was firstly designed for clinical use. Therefore, only with great recover from NDI, JOA, VAS or some other clinical assessment, we cannot get the conclusion that HO is not clinically relevant.
We must first understand the occurrence of HO, which is also a theoretical premise to prevent the complication. In this paper, the incidence of HO was meta-analyzed and relevant conclusions were drawn.

Methods
The study is a systematic review and meta-analysis, and the ethics statement is not necessary.

Searching Strategy
We have searched Medline, Web of Science, and Embase for the articles published from the inception to March, 2019. The searching strategy was built with the following terms: (((cervical disc replacement[Title/Abstract]) OR cervical disc arthroplasty[Title/Abstract])) AND ((heterotopic ossification) OR HO). The references of all publications were also retrieved to obtain possible studies.

Inclusion and exclusion criteria
Studies were included if they met the following inclusion criteria: (1). Study design was cohort study or randomized controlled trials.
(3). The HO incidence of one or several time points were reported.
(4). The HO in the studies was graded by McAfee's classification system.
(5). Articles were written with English or Chinese.
Publications were excluded if they were with following characteristics: (1). Review articles, meta-analysis, cases reports, animal/cadaver studies, editorials, or letters.
(2). The type of prosthesis was inappropriate, for example, the Dynamic Cervical Implant (DCI).
(3). Article was not formally published, only with an "accepted" status.
(4). Study was with repeated or un-extractable data.
The screen process of studies was carried out independently by 2 authors.

Data extraction
For each study included, the following data and information were collected: first author, year of publication, type of prosthesis, study design, follow-up period, sample size, HO incidence. The data were independently extracted by 2 professors, and any disagreements were resolved by discussion and consensus. Data on the incidence of HO were divided into several subgroups based on the length of follow-up of every single year from 6 months to 114 months.

Quality assessment
Here we used the methodological scoring system reported by Loney et al [8] to to evaluate the included studies. The evaluation system has a maximum score of 8 points as listed in Table-1. The quality of included studies was evaluated independently by 2 authors.

Statistical analysis
We calculated the overall incidence of HO with 95% confidence intervals (CIs), and obtained corresponding forest plots. The I 2 statistic and Q tests were used to quantify the heterogeneity (I 2 25% as low heterogeneity, 25-75% as moderate heterogeneity, and 75% as severe heterogeneity). In this meta-analysis, random-effects model was used. We use the following transformation to merge and analyze the two-category data[9]: While the x and (n-x) are both bigger than 5: While one of the x or (n-x) is equal to or smaller than 5: All of the analyses were performed with RevMan 5.3.

Searching results
The initial database search identified a total of 327 records and duplicate-check removed 123 of them. After the titles and abstracts were reviewed, 14 of them were eliminated. A full-text review was evaluated in the 190 records maintained, and 138 of them were excluded because of they didn't meet the inclusion criteria. Finally, 52 articles, meeting the inclusion criteria, were included in the present meta-analysis. Figure-1 shows the selection process and the characteristics of included studies were listed in Table-2. After the quality assessment of the included studies, 15 of them scores 7, 16 of them scores 6, 12 of them scores 5, and 9 of them scores 4. Hence, the quality of included studies is relatively satisfactory.

Annual incidence of HO
According to the follow-up period, we classified the included data with the interval of 12 months. When we based our assessment on the number of cases, as shown in Figure-2A and Figure-2B, the total incidence of HO, without limitation of follow-up period, was 39% (95% CI: 32%-47%, I 2 =94% among the whole part, I 2 =71.6% among the subgroups), and with the extension of follow-up period, the incidence increased year by year. Similarly, based on the number of levels, as shown in Figure-3A and Figure-3B, the total incidence of HO was 39% (95% CI: 33%-45%, I 2 =94% among the whole part, I 2 =64.8% among the subgroups), and with the extension of follow-up period, the incidence also increased year by year. The two statistical methods ended up with similar results. Actually, within 6 years, the HO incidence did not obviously change, while after 6 years, the incidence, no matter case-based or level-based, significantly increased.

Annual incidence of low-grade (Grade I-II) and high-grade (Grade III-IV) HO
According to the McAfee's grading system, as listed in Table-3

Discussion
CDA have already been the classical surgical procedure in anterior cervical approach, whose effect of treatment is the same as that of traditional ACDF. What's more, the ROM of index level could also be remained and the living quality of patients was improved. But unfortunately, the superiority of CDA has been seriously compromised by the occurrence of HO, especially in Grade III-IV group.
In large joints replacement, for example the total hip arthroplasty (THA), HO is a frequent complication and the exact mechanism of it remains unknown[60, 61]. Several inflammatory processes including a series of mediators and growth factors are possibly involved in the formation of HO, which leads to the recruitment of mesenchymal stem cells and the formation of bone. In recent years, with the gradual development of CDA, more and more clinicians have realized the significance of HO after cervical spine surgery. The mechanism of HO is still unclear, and many studies have suggested that it may be related to soft tissue injury, improper prosthesis size, and other factors. Accordingly, many clinicians have put forward corresponding prevention strategies, such as thorough irrigation of the surgical field, postoperative use of NSAIDs drugs, etc., but with little effect, and there is no well-accepted prophylaxis.
In different reports, the incidence rate of HO varies greatly, which may be due to the different degree of understanding and diagnostic method of HO by different authors [62].
Through meta-analysis of existing data, this study attempted to obtain a relatively objective incidence of HO and explore its variation rule during follow-up period.
The results of the analysis showed that the incidence of HO increased with the extension of follow-up time, both based on the number of cases and the number of segments, and the increase mainly occurred after 6 years. This result suggests that the follow-up time after CDA should be long enough, especially for young patients, so as to ensure the true clinical results. On the other hand, based on the McAfee's grading system, we analyzed Grade I-II and Grade III-IV respectively, and found that the incidence of HO in Grade I-II did The deficiency of this study mainly lies in that, in order to observe the principle of "prefer less to more", we added the entry of "exclude polycentric study" in the exclusion criteria, which is to prevent possible data duplication. However, this standard also makes the amount of data available reduce greatly. In some follow-up points, only 1-2 articles are included, so the data obtained are not objective enough. Moreover, we haven't registered a protocol for the SR/MA, which should be perfected in the future.
Through this study, we can preliminarily define the relatively objective incidence and change rule of HO, which provides data basis and theoretical basis for the future research of HO prevention strategy.

Supplementary Files
This is a list of supplementary files associated with the primary manuscript. Click to download.

PRISMA checklist.pdf
List of all excluded studies.xlsx