Does Heterotopic Ossification Affect The Outcomes Of Cervical Total Disc Replacement? A Meta-analysis

Objective. The purpose of this study was to answer the followingquestions:(1) Does heterotopic ossification (HO) negatively influenceclinical outcomes following cervical total disc replacement (CTDR)?(2)Should patients be classified into HO and non-HO groups?(3) Is there amore rational classification?Summary of Background Data. HO has emerged as a common complicationfollowing CTDR and has been an important reason for reoperation, thuslimiting the use of the surgery. However, the influence of HO on clinicaloutcomes following CTDR has not been well established.Methods. A search of the electronic literature databases MEDLINE,EMBASE, and the Cochrane Library was performed for publications. Thefollowing search terms were used:“heterotopic ossification”,“totaldisc replacement*”,“arthroplast*” and “artificial”. After duplicatestudies were excluded using EndNote X7, the title and abstract of eachidentified study were screened. Full-text reviews of the remainingstudies were independently performed to determine their final inclusionin the system review. The methodological quality of each study wasestimated using criteria set by the Journal of Bone&Joint Surgery,American Volume (for prognostic studies) and modified by Han Jo Kim etal. The primary outcome of interest for this study was VAS neck pain. Thesecondary outcomes included VAS arm, NDI scores and ROM. Each of the fouroutcome measures was analyzed under three classifications, based on thegrade of HO according to McAfee classification. In the subgroup analysis,the studies were divided into a non-constrained group and asemi-constrained group according to the prostheses type. The followinginformation from all eligible studies were extracted: the publicationdate, first author, study type, sample size, mean age, follow-up, HOcriteria, outcome measurement, prostheses type, surgical technique andincidence of different HO grades. For each included study, the weightedmean differences (WMD) and95%confidence intervals (CI) were calculatedfor continuous outcomes. Statistical heterogeneity was assessed using thevalue of the I2statistic. An I2value≤50%was considered to be of lowstatistical heterogeneity, The level of significance was set at P<0.05.All statistical analyses were performed using STATA, version12.0 software (Stata Corporation, College Station, TX, USA). Sensitivityanalysis was performed by altering the effect model, omitting one studyin each turn to estimate the stability of the results. Furthermeta-analysis regression was performed if the heterogeneity was high orthe results were unstable. Potential publication bias was evaluated usingthe Begg and Egger tests.Results. The initial literature retrieval identified185reports,55of which were excluded because of duplicate studies.9of these studieswere included finally. Four of the included studies were prospectivecohort studies, while the remaining three studies were retrospectivecohort studies. The levels of evidence were level I in two studies, levelII in four studies and level III in one study. The sample sizes rangedfrom21to158, representing a total of444patients; the age ranged from18to80years; the non-constrained prostheses were used in6studies,while semi-constrained prostheses were used in3studies. Mehrenclassification of HO was also used in one study, the McAfee and Mehrenclassifications are nearly identical. Publication bias was assessed usingboth Egger and Begg tests; a funnel plot was not used, because the numberof studies was too small. VAS arm pain under classification1demonstratedpublication bias when the Egger test was applied (P=0.045). Underclassification3, compared with the “low-grade HO” group, the“high-grade HO” group revealed less VAS neck pain (WMD,-1.08;95%CI,-2.06,-0.10) and less VAS arm pain (WMD,-1.38;95%CI,-2.46,-0.30).Under classification1, the HO group revealed less ROM compared with thenon-HO group (WMD,-3.74;95%CI,-4.85,-2.63). Under classification2,the “ROM-affecting HO” group revealed less ROM compared with the“ROM-preserving HO” groups (WMD,-4.91;95%CI,-6.00,-3.83). Underclassification3, the “high-grade HO” group revealed less ROM comparedwith the “low-grade HO” group (WMD,-8.86;95%CI,-9.61,-8.11). Nosignificant difference was seen in any other outcomes. In the subgroupanalysis, there were no significant difference between non-constrainedgroup and semi-constrained group in VAS neck pain, VAS arm pain and NDIunder classification1. In the sensitivity analysis, the conclusions ofall outcomes were stable when changing the effect models. We also assessedthe influence of a single study on the overall effect by removing one studyin each turn; The conclusion for VAS arm pain under classification1changed when two single studies were removed. Further meta-regressionanalyses were performed to explore the heterogeneity sources. The resultsrevealed that none of the following was the source of heterogeneity: thefollow-up period, sample size, publication time, incidence of HO,prostheses type, or level of evidence; instead, it was attributed to theactual heterogeneity between studies.Conclusions. The presence of HO is not associated with clinical outcomes following CTDR. However, the severity of HO actually impactsclinical outcomes in an inverse manner which needs further investigation.It’s inappropriate to classify patients based on the presence ofheterotopic ossification, further studies of the classifications(ROM-affecting HO vs. ROM-preserving HO; high-grade HO vs. low-grade HO)and cervical stability following CTDR are needed.