**4. Discussion**

Anatomical shoulder arthroplasty is an effective method for treating degenerative joint diseases if the bone substance is enough and the RTC is intact [14]. Nevertheless, compared to knee or hip replacements, TSAs have a relatively short lifespan, averaging ten years [15]. Therefore, each component of the shoulder prosthesis should be well chosen, aiming for the longest possible survival. Any part of the prosthesis can lead to revision surgery if it is not harmonized with the other components or is flawed.

At a mean follow-up of 68.2 months (range: 1.8–119.6; SD: 27.9), complete revision in our study cohort was required in 17 patients (68%), and 71% of shoulders undergoing revision (12 of 17) had PE wear as the main reason. These results were similar to findings from a study by Gauci et al. [16], in which a total of 26 out of 69 shoulders were revised, including 16 out of 26 shoulders in the MBG group. After a follow-up period of 12 years, the survival of the implants was 24% (SD: 0.10) for the metal-backed components. PE wear with metal-on-metal contact, RTC deficiency, and instability accounted for revision in the MBG group [16]. Gauci et al. [16] had a similar patient number, indications for revision, and survival rates. However, their follow-up period was longer than ours, so their findings may indicate what we have to deal with in the future, namely, a decrease in the survival of the implants. Over time, the chances of degenerative changes in the bone and deterioration of the inlay increase due to extended use of the shoulder joint after implantation of the prosthesis.

Another study conducted by Boileau et al. [17], also showed with 46% a very poor survival rate in 165 TSAs with 2 to 16 years of follow-up and a mean age of 68 years. These patients were diagnosed with primary OA and then treated with aTSA using an uncemented MBG component. The outcome for the survival rate free of revision was 46% at 12 years, with the endpoint for the survival curve defined as either complete or partial revision. Of the study population, 61 patients, or 37%, had undergone revision surgery after a mean follow-up of 8.5 years, 49 of whom had evidence of PE inlay wear [17]. It is worth mentioning that in our study, a similar rate of revisions was caused by PE wear. For the survival rates in our study, the longevity of the prostheses until revision surgery was reported. The mean follow-up was greater than ours, and our study's indication for aTSA with MBG was also mainly OA. The revisions of the patients in our study came in earlier. However, this study has a higher patient coverage.

To date, with 570 metal-backed TSAs, the most extensive series show low survival rates with MBG implants in aTSA with 95 revisions from a total of 121 accounting for metal-backed prostheses after 15 years [18]. This shows that prostheses with MBGs lead to increasing revision rates over time, rendering MBGs inadvisable for long-term use. A systematic literature search was conducted by Papadonikolakis and Matsen [19] regarding papers stating radiographic leakage, loosening, or revision of the glenoid component in aTSAs that had been carried out in patients of all ages and with any diagnosis. They found that when comparing 1571 MBG and 3035 full-PEG implants, the revision rate was more than three times higher for MBG components (14%) than for full-PE components (3.8%), according to the authors' findings. As many as 77% of revisions of full-PE components were due to loosening, whereas 62% of revisions of MBG components occurred due to other causes, such as PE wear, metal wear, component dissociation or fracture, screw fracture, and RTC tear [20]. Our age distribution correlates with similar results of specific studies published by Gauci et al. [16] and Taunton et al. [21]. Age could also play a considerable part in the results since older patients may have a lower functional demand than younger patients or tend to have higher rates of RTC deficiency, which correlates with patient age [22]. It has been argued that, in comparison to older patients, young patients have higher functional demands and higher expectations of enhanced capacity for social interaction, participation in sports, and exercise [23].

The mean age of our patients was 70.9 years. Similar mean patient age of 68 years was found in the study by Taunton et al. [21]. It was also found that at an average follow-up of 9.5 years, the five-year survival rate in this study was 79.9%, and the 10-year survival rate was 51.9%, leading the authors to express significant concern about the utilization of metal-backed, uncemented glenoid components [21,24].

In our study, the objective and subjective clinical results were poor. The clinical outcomes of our study were, in general, worse than those in similar studies on the topic of aTSA with MBG, especially for CMS. The flat profile design of our MBG and the screws with a small diameter that we used could contribute to the worse outcome, as not all studies compared had a flat MBG or utilized screws with a small diameter. Altogether, with the ADL score, which was on average 12.8 (64%), the most important domains for patients' quality of life were not as poor as might be expected. This could be mainly due to the advanced age of the patients and the relatively inactive life they lead. Almost all study participants were retirees who did not require high mobility. A similar negative result was found in a study by Clement et al. [25]. Therein, 49 shoulders with metal-backed glenoids of 39 patients had an average CMS of 33.5 after a minimum follow-up of 132 months. However, no patient was able to abduct their arm to 90 degrees pre- or postoperatively [25]. A study by Fucentese et al. [26] that examined the clinical and radiographic outcomes associated with the use of an uncemented soft-metal-backed glenoid component found a CMS of 65.9 in 22 patients after a mean follow-up of 50 months [26]. The study from Gauci et al. [16] had a CMS of 64, similar to the one from Fucentese et al. [26], which was found in 7 shoulders with MBG from 23 MBG prostheses, of which 16 were lost to follow-up, at a mean follow-up of 10.3 years [16]. In 2017, Kany et al. [27] noted a mean CMS of 56.6 and a mean SST of 6.7 points in their study of 14 TSAs with MBG [27]. This implies a poor outcome in MBG prostheses, which was also apparent in our case. Another study by Kany et al. [28] found a mean CMS score of 60 in a total of 26 cases, 16 of whom had TSA with MBG and a mean SST score of 8 [28]. The study by Kany et al. [28] had a similar patient outcome but a moderately better CMS and SST score than ours.

The radiological results are in concordance with the clinical results of our study. These radiographic results reaffirm the poor clinical scores. The radiological results correlate with the clinical results.

In our study, there was a huge difference in RLs from the first postoperative radiographic control to the follow-up. At follow-up, 22 patients (88%) had one or more RLs on the glenoid, and 21 patients (84%) had RLs on the humeral components, whereas none were found at the first postoperative admission. In contrast, the study by Boileau et al. [29] showed RLs in only 5 of a total of 20 shoulders with an MBG after a mean follow-up of 38.4 months [29]. For the humeral component, Gallacher et al. [14] analyzed a total of 100 shoulders from 86 patients with a mean follow-up of 35.4 months (range 24–76 months). The study found that 12% had incomplete RLs, and 4% had complete RLs [14]. Magosch et al. [30] conducted a study of 48 TSA-implanted patients with MBG who were clinically and radiologically followed up with a mean of 49 months. They found in total 4 cases of incomplete RLs, two cases with under 1 mm of thickness, and 2 cases with RLs < 2 mm. As in our study, they did not find glenoid component loosening in their cases. However, we had more cases with RLs using the same prosthesis type as Magosch et al. [30]. In summary, 8 cases from their study required revision [30,31]. Unlike other studies, ours had a large number of RLs. The high amount of PE wear could be responsible for this. However, none of the MBGs at the follow-up were loose despite RLs and rarefication due to PE wear. Furthermore, in the literature, it has been reported that the prevalence of RLs in TSA ranges from 22% to 95%, and they occur in both types of glenoids, whether they are cemented or uncemented. However, evaluation of radiolucency from radiographs is prone to error, with standardization of patient position being difficult due to scapulothoracic mobility and individual anatomic differences. The lack of a standardized scoring system makes the comparison of findings challenging [32]. RLs rates were shown to be highly variable throughout studies, complicating comparisons of related factors.

HHM was detected on the AP radiographs in all patients within our study. Clinically, proximal humerus migration is important because it implicates a disturbance of normal glenohumeral kinematics from which advanced RTC disease is often a sign [33]. There

was a mean difference between the HHM of the first postoperative radiograph and the last follow-up of 6.4 mm (range: 0.5–13.4; SD: 3.9; effect size: 1.6). This indicates a serious disturbance of normal glenohumeral kinematics. Montoya et al. [34] observed HHM in 8 of 53 patients after a mean follow-up of 64 months [34]. A comparison of HHM between our study and other findings in the literature reveals that HHM is more common in our study, but for the most part, the level of upward migration is rather moderate.

Regarding the LGHO, we see a similar pattern. Among 25 patients, 23 developed PE wear. In some patients, the MBG also showed signs of wear to some extent. The LGHO in our study had a mean difference between the first postoperative imaging and the follow-up of 2.6 mm (range: 0–4.0; SD: 1.5; effect size: 1.7). All patients who underwent revision had radiographic signs of wear of the glenoid component as well as superior HHM at follow-up. It has been shown by biomechanical and clinical studies that MBG implants used in aTSA have adverse effects on both the PE and the underlying glenoid bone [20]. In the future, finite element computational analysis (in silico) will allow implant designs to be tested and improved in advance. This could help reduce PE wear and therefore support the longevity of shoulder prostheses with an MBG [35,36].

Long-term studies of MBG implants and their LGHO measurements on radiographs are lacking, making comparative studies scarce. Nevertheless, this is an important issue that can be used to dispute the beneficial effect of MBG components in aTSA, as almost all individuals in the study population received HHM, RLs, and LGHO. Analysis of the results for clinical scores showed a similar trend as described above for radiological outcomes.

In summary, the propagated advantage of the new modular MBG components concerning prosthesis survivability cannot be confirmed in our study. The clinical results are consistent with the radiological results, which are also unacceptably poor.

The study conducted had some limitations. The number of patients was relatively low at 25. We did not include a control group for comparison purposes. A comparison group would be desirable, especially in view of the unusually high complication and revision rate. The results of the present study are also compared against old MBGs with known technical problems and a significantly longer follow-up.

Due to the fact that angles and lengths are impacted by the location of the scapula and the humerus rotation, all radiographic measures are strongly reliant on the patient's orientation during radiographic imaging. Radiographic imaging, even when standardized, can reveal massive differences. However, no patients were lost to follow-up. In addition, the same implant was used throughout the entire study population, with the same surgical technique being performed by two experienced shoulder surgeons.
