Risk Factors and Modes for Implant Failure in the Modern Dual Mobility Implant. A Systematic Review and Meta-analysis

The aims of this meta-analysis were to: (1) validate the outcome of modern dual mobility (DM) designs in patients who had undergone primary and revision total hip arthroplasty (THA) procedures and (2) to identify factors that affect the outcome. Methods We searched for studies that assessed the outcome of modern DM-THA in primary and revision procedures that have been conducted between January, 2000 to August, 2020 on PubMed, MEDLINE, Cochrane Reviews and Embase. The pooled incidence of common failure modes and functional scores were evaluated in patients who have received: (1) primary THA, (2) revision THA for all causes or (3) for recurrent dislocation. A meta-regression analysis was performed for each parameter to determine the association with the outcome. total of 120 studies (N= 30016 DM-THAs) were included for analysis. The mean follow-up duration was 47.3 months. The overall implant failure rate was 4.2% (primary: 2.3%, revision for all causes: 5.5%, recurrent dislocation: 6.0%). The most common failure modes were aseptic loosening (primary: 0.9%, revision for all causes: 2.2%, recurrent dislocation: 2.4%), septic loosening (primary:0.8%, revision for all causes: 2.3%, recurrent dislocation: 2.5%), extra-articular dislocation (primary:0.6%, revision for all causes:1.3%, recurrent dislocation:2.5%), intra-prosthetic dislocation (primary:0.8%, revision for all causes:1.0%, recurrent dislocation:1.6%) and periprosthetic fracture (primary:0.9%, revision for all causes:0.9%, recurrent dislocation:1.3%). The multi-regression analysis identied younger age (β=-0.04, 95% CI -0.07 – -0.02) and female patients (β=3.34, 95% CI 0.91–5.78) were correlated with higher implant failure rate. Age, gender, posterolateral approach and body mass index (BMI) were risk factors for extra-articular or intra-prosthetic dislocation in this cohort. overall Harris hip score and Merle d’Aubigné score 84.87 and 16.36, respectively. 183, 184) In primary xed-bearing THA, the midterm and long-term revision rate ranged from 1.2-4.0% and 12.1-14.3%, respectively.(1, 38, 60, 73, 98, 108, 183) In revision xed-bearing THA, the mid-term and long-term revision rates can be up to 5.3-13% and 27-45%, respectively.(39,

We included original articles written in English that validated the outcome in patients who had undergone DM-THA for all kinds of indications including primary THA, revision THA or recurrent dislocation. We excluded review articles, letter to the editor, expert opinion, biomechanical studies, articles not written in English, study period earlier than 2000 or studies in which data were not obtainable. For comparative studies (e.g. hemiarthroplasty or THA vs DM-THA), we extracted data from the DM-THA group if possible. If there was uncertainty regarding the data from the study, we contacted the authors for clari cations.
Two authors (xxx, xxx) examined all relevant studies and extracted data using a predetermined form. The primary aim was to determine the overall implant failure rate and failure modes including aseptic loosening, septic loosening, extra-articular dislocation, intra-prosthetic dislocation and periprosthetic fracture. We further validated these rates strati ed by indications including primary THA, revision THA for all causes or for recurrent dislocation. The secondary aim was to identify risk factors for implant failures and to evaluate the functional outcome using Harris hip score (116) and Merle d'Aubigné score (117). We recorded the rst author, year, study design, number of THA procedures, indications, age, follow-up duration, implant brand and outcome parameters in Table 1.
Two authors (xxx, xxx) independently evaluated the methodological quality of the included studies using the NIH Quality Assessment Tool for Case Series Studies (118). The highest score on this scale is 9. A score between 7 and 9, 4 and 6, less than 4 were de ned as "good", "fair" and "poor", respectively. If there were disagreement, we consulted a third author (xxx). (Table 2) Statistical analysis A meta-analysis of proportions was conducted using the Freeman-Tukey analysis under random-effects model to determine pooled estimates with a 95% con dence interval (CI). A random-effects model was used for differences among studies such as age, sex, surgical approaches, body mass index, indications for THA procedure, implant brand and methodology. A standard multivariate linear regression analysis (β) was performed to determine potential factors for implant failure or improved functional outcome. We completed all analyses with the Comprehensive Meta-Analysis (CMA) software, version 3 (Biostat, Englewood, New Jersey, USA) and signi cance was de ned as p < 0.05.

Results
We identi ed 1123 studies according to our search strategy. We removed 714 duplicate records and 231 studies after reading the title and abstract.
Another 58 studies were excluded after reading the full text as the studies did not meet the inclusion criteria: studies on different outcome domains (n=21), mixed etiologies (n=12), 1st generation DM designs (n=10), cemented liner to cup (n=9), cadaveric or in vitro studies (n=3), studies not written in English (n=3). After exclusion, a total of 120 studies were included   (Figure 1).

Baseline characteristics
We included 30016 patients who had undergone DM-THA for primary and revision THA procedures. The mean age was 71.9 years (range, 19 Aseptic loosening A total of 105 studies, including 28980 DM-THA procedures, have reported aseptic loosening rates. The overall pooled rate was 1.6% (95% CI 0.008 -0.032). The aseptic loosening rates in primary THA, revision THA and revision THA for recurrent dislocation were 0.9%, 2.2% and 2.4%, respectively (Table   3, Figure S1). A multivariate regression analysis revealed that a revision THA procedure for all causes (β=1.30, 95% CI 0.71 -1.89), or for recurrent dislocation (β=1.18, 95% CI 0.26 -2.10), carried a higher risk of aseptic loosening compared with a primary THA procedure (Table 4).

Intra-prosthetic dislocation
A total of 113 studies, including 20447 DM-THA procedures, have reported the intra-prosthetic dislocation rates. The overall pooled rate was 1.0% (95% CI 0.007 -0.015). The intra-prosthetic dislocation rates in primary THA, revision THA and revision THA for recurrent dislocation were 0.8%, 1.0% and 1.6%, is considered less accurate than the retrieval or simulation studies.(171, 172) Currently, there is limited evidence regarding the increased PE wear of modern DM articulation.
The non-porous alumina-coated surface, tripod anchoring system of acetabular component and polyethylene wear have been associated with a higher aseptic loosening rate in the rst-generation DM implants. (24,29,31) Several changes have been made in modern dual mobility designs, including (1) to replace UHMWPE with HXLPE to reduce wear (33,34); (2) to add bevelled edges (or chamfer) in polyethylene (PE) inserts to lower femoral neck impingement and wear (32); (3) press-t xation by bilayer coating of porous titanium and hydroxyapatite to enhance osseointegration on the outer surface(31); (4) modular metal liner design to facilitate supplementary screw xation. The long-term overall survival and aseptic loosening rate of the primary THAs using 1 st generation DM implants were 85-95.4% and 3-8.3%, respectively. (24)(25)(26)(27)(28) In this study, the primary THAs using modern generations DM implants are associated with a better overall survival (97.7%) and a lower aseptic loosening rate (0.9%). This pooled aseptic loosening rate was comparable to that of primary, xed-bearing THA from several registries, which ranged from 0.7-1.1% at 5 to 16 years. (1,173,174) The modern, modular design has an additional cobalt-chromium (CoCr) liner inserted into a titanium acetabular component allowing supplementary screw xation to enhance primary stability. However, the metal-on-metal interface between CoCr liner and titanium cup is at risk of fretting corrosion and remains a concern.(175-177) Metal ions can further lead to advance local tissue reaction (ALRT) and implant loosening.(178) The rst study regarding metal ions was conducted by Matsen Ko et al., which revealed 21% of the patient had elevated serum chromium levels.(179) Other studies reported that serum ion levels (cobalt, chromium or titanium) was elevated in 9.3-23% of the patients. (47,111) On the other hand, some studies have noted that this elevation was not associated with clinical adverse events including instability, loosening or need of revision. (64,67,72) In summary, the current evidence suggests there is a slight elevation of serum ion level but this does not negatively affect the implant survival.
Intra-prosthetic dislocation (IPD) is a rare complication of DM design, which occurs as a result of retentive failure of the inner articulation. Long-term, homogenous PE wear or impingement at extreme range of motion between neck and PE liner leads to loss of PE retentive rim and IPD.(180, 181) The incidence of IPD ranged from 0.7%-4.3% in rst generation of DM cup and(29, 30) modi cations have been made to the 2 nd generation DM implants. These changes include a thinner, more polished femoral neck to reduce impingement with the liner and the use of HXLPE to reduce wear during contact.
(32) In our pooled study for modern DM cup, IPD rate in primary THA and revision THA was 0.8% and 1.0% respectively, which is much lower than the 1 st generation. (29,30) Another form of IPD has been observed in modern generation DM implants, which often occurs in the short-term. This form of IPD results from a secondary decapsulation of the liner followed by reduction for dislocation.(182) During close reduction of a dislocated DM-THA, impingement occurs between PE liner and posterior edge of the acetabular component. The excessive loading during reduction maneuver may "decapsulate" the femoral head from PE liner. Therefore, the reduction should be performed gradually under general anesthesia with completely relaxed muscle. (29) Our meta-analysis showed that the mid-term revision rates in primary and revision DM-THA were 2.3% and 5.5-6.0%, respectively. These results were comparable to the reported outcome of primary or revision, xed-bearing THA. (1,38,39,60,73,98,108,183,184) In primary xed-bearing THA, the midterm and long-term revision rate ranged from 1.2-4.0% and 12.1-14.3%, respectively. (1,38,60,73,98,108,183) In revision xed-bearing THA, the mid-term and long-term revision rates can be up to 5.3-13% and 27-45%, respectively. (39,184) This meta-analysis revealed promising mid-term outcomes and a reduction in dislocation rate, but the long-term implant survival of modern DM-THA is still lacking. In addition, the regression analysis showed that revision THA procedures, younger age and female patients were associated with a higher risk of implant failure. Younger patients have been established as a risk factor for failure after primary THAs. However, whether female sex is a risk factor remains controversial.(185-188) This can be attributed to the representativeness of the study cohort, follow-up duration and type of implant. Although female patients have been associated with increased risk of dislocation, aseptic loosening, periprosthetic fracture and overall implant failure after primary THA(187, 188), the same was not seen in DM-THA aside from overall implant failure. Potential confounders and inadequate follow-up duration are important considerations when interpreting this result.
We should recognize several limitations. First, we included only studies written in English. In addition, due to the nature of our research question, the level of evidence of the included studies was low (III or IV). Furthermore, we included studies that reported outcome of modern DM (the 2 nd and 3 rd generation) implants over a time span of 12 years between 2008 to 2020. We could only analyze factors that were clearly described in the studies, including age, sex, surgical approach, BMI and indication for hip arthroplasty. Factors such as surgeons' experience, patient activity level or implant designs could have affected the outcome but were unavailable could not be analyzed. Nonetheless, this review provides an updated information about the outcome of modern DM implants and factors that might affect the outcome.

Conclusions
In conclusion, the mid-term implant survival of modern dual-mobility designs was satisfactory. Aseptic loosening continued to be the most common failure mode after DM-THA. Younger age and female sex were correlated with implant failure.

List Of Abbreviations
Dual mobility (DM); total hip arthroplasty (THA); body mass index (BMI); intra-prosthetic dislocation (IPD); Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA); con dence interval (CI); Comprehensive Meta-Analysis (CMA); ultra-high molecular weight polyethylene (UHMWPE); highly cross-linked polyethylene (HXLPE); polyethylene (PE); cobalt-chromium (CoCr); advance local tissue reaction (ALRT) Authors' contributions FYP and SWT were responsible for conception and design, publication screening, acquisition of data, analysis and interpretation, and drafting and revising the manuscript. HHM and TFAC were initial analysis and prepared tables. TWH and KCH prepared gures. CFC and WMC were responsible for reviewing and revising the manuscript. All authors were involved with interpretation of the data. All authors discussed the results and commented on the manuscript.
The author(s) read and approved the nal manuscript.
Was the study population clearly and fully described, including a case de nition?
Were the outcome measures clearly de ned, valid, reliable and implemented consistently across all study participants?
Quality of the cohort study (score) 8 9 7 7 8 9 9 7 8 7 Y= Yes, N= No; The maximum possible score on this scale is 9. "Good" was de ned as a total score of 7-9; "fair" as a score 4-6, and "poor" as a score of less than 4.
Was the study population clearly and fully described, including a case de nition?
Were the outcome measures clearly de ned, valid, reliable and implemented consistently across all study participants?
Quality of the cohort study (score)  8  7  7  7  7  8  8  8  8 Y= Yes, N= No; The maximum possible score on this scale is 9. "Good" was de ned as a total score of 7-9; "fair" as a score 4-6, and "poor" as a score of less than 4.
1. Was the study question or objective clearly stated?
Was the study population clearly and fully described, including a case de nition?
Were the outcome measures clearly de ned, valid, reliable and implemented consistently across all study participants?
Were the statistical methods welldescribed?
Quality of the cohort study (score) 6 7 7 9 8 7 8 7 8 8 8 Y= Yes, N= No; The maximum possible score on this scale is 9. "Good" was de ned as a total score of 7-9; "fair" as a score 4-6, and "poor" as a score of less than 4.
Was the study population clearly and fully described, including a case de nition?
Were the outcome measures clearly de ned, valid, reliable and implemented consistently across all study participants?
Was the length of follow-up adequate?
Were the results welldescribed?
Quality of the cohort study (score )   7  8  8  7  7  7  7  8  7  8 Y= Yes, N= No; The maximum possible score on this scale is 9. "Good" was de ned as a total score of 7-9; "fair" as a score 4-6, and "poor" as a score of less than 4 1. Was the study question or objective clearly stated?
Was the study population clearly and fully described, including a case de nition?
Were the outcome measures clearly de ned, valid, reliable and implemented consistently across all study participants?
Was the length of follow-up adequate?
Quality of the cohort study (score) 7 6 9 8 8 6 7 8 7 7 8 Y= Yes, N= No; The maximum possible score on this scale is 9. "Good" was de ned as a total score of 7-9; "fair" as a score 4-6, and "poor" as a score of less than 4 1. Was the study question or objective clearly stated?
Was the study population clearly and fully described, including a case de nition?
Were the cases consecutive?
Quality of the cohort study (score) 8 7  6  8  8  8  6  8  7  8  8  8 Y= Yes, N= No; The maximum possible score on this scale is 9. "Good" was de ned as a total score of 7-9; "fair" as a score 4-6, and "poor" as a score of less than 4 1. Was the study question or objective clearly stated?
Was the study population clearly and fully described, including a case de nition?
Were the cases consecutive?
Were the outcome measures clearly de ned, valid, reliable and implemented consistently across all study participants?
Was the length of followup adequate?
Quality of the cohort study (score)   8  8  7  6  7  7  8  6  7  6  9 Y= Yes, N= No; The maximum possible score on this scale is 9. "Good" was de ned as a total score of 7-9; "fair" as a score 4-6, and "poor" as a score of less than 4 1. Was the study question or objective clearly stated?
Was the study population clearly and fully described, including a case de nition?
Were the outcome measures clearly de ned, valid, reliable and implemented consistently across all study participants?
Quality of the cohort study (score )   7  7  8  8  7  7  8  7  7  7  7  6 Y= Yes, N= No; The maximum possible score on this scale is 9. "Good" was de ned as a total score of 7-9; "fair" as a score 4-6, and "poor" as a score of less than 4 1. Was the study question or objective clearly stated?
Was the study population clearly and fully described, including a case de nition?
Were the subjects comparable?
Were the outcome measures clearly de ned, valid, reliable and implemented consistently across all study participants?
Were the results well-described?
Quality of the cohort study (score )   6  7  8  8  6  7  5  8  9  7  8 Y= Yes, N= No; The maximum possible score on this scale is 9. "Good" was de ned as a total score of 7-9; "fair" as a score 4-6, and "poor" as a score of less than 4 Tarasevicius et al.
Was the study population clearly and fully described, including a case de nition?
Were the outcome measures clearly de ned, valid, reliable and implemented consistently across all study participants?
Was the study population clearly and fully described, including a case de nition?
Were the cases consecutive?
Were the subjects comparable?
Was the length of follow-up adequate?
Were the results welldescribed?
Quality of the cohort study (score) 7 7 7 6 7 9 8 7 7 7 7 Y= Yes, N= No; The maximum possible score on this scale is 9. "Good" was de ned as a total score of 7-9; "fair" as a score 4-6, and "poor" as a score of less than 4