The reasons for ceramic-on-ceramic revisions between the third- and fourth-generation bearings in total hip arthroplasty from multicentric registry data

This study aimed to evaluate (1) the overall reasons for first revision in CoC THAs; (2) whether the reasons for revision differ between third-generation and fourth-generation CoC THAs; and (3) the specific factors associated with bearing-related problems as the reason for revision. We retrospectively reviewed 2045 patients (2194 hips) who underwent first revision THA between 2004 and 2013, among which 146 hips with CoC bearings underwent revision. There were 92 hips with third-generation ceramic bearings and 54 hips with fourth-generation ceramic bearings. The major reasons for CoC THA revisions were ceramic fracture and loosening of the cup or stem. When ceramic fracture, squeaking, incorrect ceramic insertion, and unexplained pain were defined as directly related or potentially related to ceramic use, 28.8% (42/146) of CoC revisions were associated with bearing-related problems. Among the third-generation ceramic bearings, revision was performed in 41.3% (38/92) of cases owing to bearing-related problems whereas revisions were performed for only 7.4% (4/54) of cases with fourth-generation ceramic bearings owing to bearing-related problems (p < 0.001). Younger age, lower American Society of Anesthesiologists (ASA) grade, and preoperative diagnosis of osteonecrosis were factors related to CoC THA revisions due to bearing-related problems.

Statistical analysis. Basic descriptive statistical analyses were used to describe the study population. Values were expressed as means or percentages. The independent t-test was used to compare continuous variables, and the Mann-Whitney test or Fisher's exact test was used to compare categorical variables. Statistical significance was defined at p ≤ 0.05. Statistical Package for the Social Sciences software ver. 18.0 (SPSS Inc., Chicago, IL, USA) was used for all statistical analyses.

Results
Patient demographics are shown in Table 1. Baseline demographics were similar between the third-and fourthgeneration CoC revision groups. Of the surgical characteristics, the type of surgery and interval between initial surgery and revision were significantly different between the two groups ( Table 2). In the third-generation CoC revision group, bearing change was more frequently performed [27/92, 29.3% vs. 5/54, 9.2%, p = 0.045] and the time to revision was longer (58.9 ± 47.6 vs. 27.5 ± 12.4 months, p < 0.001), compared to the fourth-generation CoC revision group.
The reasons for CoC revision in decreasing order were ceramic fracture in 34 hips, loosening of cup or stem in 34 hips, periprosthetic fracture in 24 hips, infection in 21 hips, instability in 19 hips, noise in 6 hips, malposition in 3 hips, leg length discrepancy in 1 hip, osteolysis in 1 hip, tumor in 1 hip, and unexplained pain in 1 hip.
The reasons for revision were analyzed in greater detail, with subgroups according to age, gender, and ceramic diameter. In the patients aged < 60 years, ceramic fracture was a major reason for CoC revision whereas loosening and periprosthetic fracture were major reasons in patients aged ≥ 60 years (Fig. 1). Gender differences were also noted. Male patients underwent revisions mainly due to ceramic fracture, while female patients received revision mainly due to loosening (Fig. 2). According to the diameter of the ceramic bearing, ceramic fracture was a major reason for 28 mm-head THA revisions, whereas prosthesis loosening was a major reason for 32 or 36 mm-head THA revisions (Fig. 3).
Forty-two (28.8%) CoC revisions were associated with bearing-related problems, with a significant difference in their proportion among the reasons for revision between the third-generation and fourth-generation CoC THA groups [38/92 (41.3%) vs. 4/54 (7.4%), p < 0.001] ( Table 3). Ceramic fracture was the most common cause of bearing-related problems [ www.nature.com/scientificreports/ The CoC THA revisions due to bearing-related problems were performed at an average of 50.2 months (range 1-159 months). Of 42 bearing-related revisions, 10 (23.8%) occurred within 2 years and 17 (40.5%) occurred within 5 years. Younger age, lower American Society of Anesthesiologists (ASA) grade, and preoperative diagnosis of osteonecrosis were significantly associated with CoC revisions due to bearing-related problems (Table 4).

Discussion
The reasons for revision of metal-on-polyethylene (MoP) THAs have typically related to wear and osteolysis and metal-on-metal (MoM) revision surgery has been performed mainly for adverse reaction to metal debris. In contrast, it remains uncertain why contemporary CoC THAs lead to failure. Moreover, it has not been fully clear whether the two types of currently used CoC bearings are associated with different reasons for revision. In the current study, the most common reason for first revision in CoC THA was ceramic fracture (34/146, 23.3%). When the reasons directly or potentially related to ceramic use were defined as bearing-related problems, the third-generation CoC THA revisions were associated with a higher proportion of bearing-related problems than the fourth-generation CoC THAs [38/92 (41.3%) vs. 4/54 (7.4%), p < 0.001].
While this study included a large series of CoC THA revisions, it has some limitations. First, this study was performed at nine tertiary centers, so there may have been some differences in the surgical procedures including skin incision, surgical approach, or implantation technique. Although all surgeons were experts with more than 5 years of experience in adult reconstruction, a potential bias might have been present. However, the large cohort size of this study may render the results more significant. One hundred and forty-six of CoC revisions are not a small number, considering even the national registry data include less than 100 CoC THA revisions 12 . Second, various types of prostheses were used, which limits comparisons between different prostheses. The contribution of the characteristics of prostheses to revision was neglected in this study. In addition, a majority of 28 mm www.nature.com/scientificreports/ femoral head were included in the third generation than in the fourth due to surgical skill evolution. This could be an important bias to the study. Nevertheless, our results provide valuable data that cannot be gleaned from the national registry. Specifically, reasons for revision were classified into three subgroups: directly related to ceramic use, potentially related to ceramic use, and not specific to ceramic use and all causes of failure were considered. Third, the retrospective nature of this study has an inherent risk of observer bias, including the potential for missing data and the inability to control confounding variables. Fourth, the outcome of CoC revisions was not evaluated. There are no data regarding radiographic parameters such as inclination and anteversion angles and clinical scores. This study did not aim to assess the outcome of CoC THA revisions, but sought to investigate the reasons for revision surgery. Lastly, multivariable analyses were not performed to identify factors associated with CoC revisions due to bearing-related problems because the sample size was not sufficient to consider all variables.
In the third-generation CoC revision group, bearing change was more frequently performed, and it occurred mainly due to ceramic fracture. Overall, the third-generation CoC bearing was more associated with bearingrelated problems than the fourth-generation CoC bearing. In the case of ceramic fracture, a change in the bearing type to MoP is not recommended due to concerns over metallosis 16 . The fourth-generation CoC revision group showed a shorter time to revision compared to the third-generation CoC revision group. Based on our results, bearing-independent problems such as loosening, infection, dislocation, and periprosthetic fracture were the dominant reasons for revision of the fourth-generation CoC THAs with these issues often occurring within 2 years postoperatively.
The ceramic fracture remains an ongoing issue in CoC THAs, although the fracture rates of ceramics have decreased over time with improvements in manufacturing processes and materials. In the early-generation ceramics, it was possible for crack propagation to result in fracture, but the incorporation of zirconia into the alumina matrix (Biolox Delta) was expected to prevent this from occurring. Recently, a few mid-term clinical studies on the fourth-generation ceramic have been published [17][18][19] . Hamilton et al. 17 reported 0.9% (3 hips) of ceramic fractures among 345 THAs at a mean 5.3-year follow-up. Lim et al. 19 reported 0.3% (2 hips) of ceramic liner fractures without malseating among 749 THAs at a mean 6.5-year follow-up. Overall, the rate of the fourthgeneration ceramic factures appears to be low in comparison to the reported rates of the third generation ceramic fractures ranging from < 1 to 4.4% [20][21][22] . In this study, ceramic fracture was found to be a major cause of failure in the third-generation CoC THAs, whereas it was not in the fourth-generation CoC THAs [31/92(33.7%) vs. 3/54(5.6%)].
Squeaking is another concern of CoC bearings. Previous studies have shown that the occurrence rate of squeaking in the third-generation ceramic bearing varies from 0.7% to 20.9% 23   www.nature.com/scientificreports/ recently analyzed 3689 THAs in which the fourth-generation ceramic bearing was applied. Their reports found that squeaking occurred at a rate of 3% in the fourth-generation CoC bearing, suggesting a lack of superiority of the fourth-generation ceramic in terms of squeaking. In our study, the overall incidence of squeaking was 4.1% (6/146), all of which occurred with the third-generation CoC bearing. The femoral head size plays a role in ceramic fracture and also in squeaking. A 28 mm head size has a higher fracture rate, compared to larger sized heads 26 . A short neck length similarly has a higher risk of fracture compared to longer neck 27 . These phenomenon were believed to result from a reduced distance between the corner of the bore and outer surface of the head which can predispose to crack propagation. In our series, ceramic fracture was a major reason for 28 mm-head THA revisions. According to Tai et al. 28 , in their series, 7.3% (15 of 206 hips) of the hips were recorded as squeaking. Levy et al. 29 suggested that a higher incidence of squeaking in larger heads is primarily due to the increase in the total work done at the articular interface correlated to the applied friction force.
The general taper angle of acetabular cups for ceramic liner is 18°, however, for acetabular shells with multiple options (metal, polyethylene, ceramic), the taper angle is lower. A recent study showed that the risk of malseating a ceramic liner is significantly higher for metal shells with lower taper angle compared to shells with a 18° taper 30 . This is a concern on fracture risk of ceramic liners.
An important feature of retrieved Biolox delta is the metal transfer on femoral head. It generally implies a meaningful alteration of the bearing surface 31 . The mechanisms of metal transfer were known as femoral head dislocation, closed reduction procedures, impingement, or third body entrapment in the articulating zone. In particular, it has been hypothesized that metal transfer on the femoral head is associated with joint instability and subluxation/dislocation 32 .
Nevertheless, some advantages of ceramics led to continuous increase in use of CoC bearings in THA. CoC bearings guarantees the complete avoidance of metal debris and also, reduce the risk of wear-induced osteolysis 33 . The CoC bearings showed very low friction and very low wear rates. The wear is not directly dependent on the head diameter which allows surgeons to select a larger diameter head with fewer concerns 34 .
In our series, younger age, lower ASA grade, and preoperative diagnosis of osteonecrosis were associated with CoC revisions from bearing-related problems. These variables are commonly connected to high daily activity. Younger age and higher activity may increase the risk of impingement and subsequent mechanical failure in THAs 35 , though there is not sufficient evidence supporting the relationship between high daily activity and ceramic fractures. The use of a 28 mm short-neck femoral head and component malposition is known to increase the risk of ceramic fracture 36 . www.nature.com/scientificreports/ The most common reasons for CoC THA revisions were ceramic fracture and aseptic loosening of the implanted prosthesis. The reasons for CoC revisions differed according to the generation of the ceramic bearing. In THAs with the third-generation CoC bearing, 41.3% (38 out of 92) were revised due to bearing-related problems. In contrast, only 7.4% (4 out of 54) of the fourth-generation CoC THAs were revised due to bearingrelated problems. Younger age, lower ASA grade, and preoperative diagnosis of osteonecrosis were factors related