**4. Discussion**

This study showed that IB-RSA is a safe and effective procedure and does not present specific implant-associated complications at mid-term follow-up.

Scapular notching remain the most common complication associated with RSA [1,3]. In this series, notching occurred in 27% cases. Even though this rate still is not completely satisfying, it is lower compared to the notching rates (40–68%) of similar standard bearing RSA with a classic Grammont humeral stem with or without a lateralized glenoid [1,8,15,16,25]. Moreover, it must be underlined that the notching observed in this series has peculiar features. First, at this follow-up, no grade higher than 2 was observed (Figure 5).

As postulated by Friedman et al., grade 3 and 4 extending over the inferior screw are likely the results of a biologic response to polyethylene particles and osteolysis [3]. Secondly, in all cases with notching, a bone spur on the scapular neck was present. Third, the notching was almost only present in cases with a high position of the baseplate, a condition that is proven to be associated with mechanical contact of the prosthesis with the scapula [5,8,11,26]. All these features seem to be linked to a pure mechanical notching, proving that IB-RSA with a hard humeral liner leads to a distinct type of scapular notching and avoids PE wear-induced osteolysis at mid-term follow-up. Similar findings were shown by Kohut et al. using a different IB-RSA [27]. Based on our findings, optimal (as low as possible) and secure (optimal preparation of the subchondral bone) positioning of the glenoid is mandatory to avoid scapular notching (Figure 6).

Further studies are needed in order to analyze the notching evolution with IB-RSA at longer follow-up and verify if the notching remains mainly mechanical, or if the osteolysis will spread over the screw reaching the central peg, with a potential risk of loosening [12]. Moreover, histological studies on retrieved implants will be useful to clarify this phenomenon in vivo. In contrast with other authors [8,16], we did not find a statistical difference in baseplate inclination between patients with and without notching in our series. This finding may have different explanations. First, the limited number of patients included in

this series compared to other series [8,16]. Second, the design of the glenosphere presenting an inferior extension may compensate for a slight superior inclination of the baseplate (Figure 2). Third, we believe that scapular notching is more linked to the inclination of the baseplate relative to the scapular neck [11] or to the intrinsic neck morphology [3] compared to the inclination relative to the supraspintus fossa.

**Figure 6.** Correct position of the glenoid.

Another interesting observation was that in this series, GT and calcar resorption were not associated with scapular notching, which contrasts the finding of Mazaleyrat et al. [28] using a standard bearing RSA. This observation can be related to the fact that with IB-RSA PE, wear-induced osteolysis is minimized with a potential effect on proximal humerus resorption. The high rate of humeral stress shielding observed in this series is likely due to the methaphyseal fixation of the stem, which is associated with these radiographical changes [24,28]. Regarding the higher rate of glenoid RLL among patients with notching, we believe that this finding is due to the presence in this group of the three patients with early subsidence with the development of non-progressive RLL and notching. However, further studies at longer follow-ups are necessary to clarify the evolution of this observation.

The impact of scapular notching on clinical function is controversial. Some authors did not find an influence of notching on clinical function [2,7,8,12], while other authors clearly stated that scapular notching is associated with lower functional scores and decreased range of motion [6,9,10,29], especially in case of high grades. In this series, notching development has a significant negative influence on functional scores and anterior elevation at mid-term follow-up. We believe that this finding is due to the fact that the notching observed with this prosthesis is mainly mechanical and strictly linked to the incorrect high position of the glenoid. This leads to premature contact of the humeral component with the scapula and the acromion and consequent limitation of the motion [26].

Clinical scores and range of motion were overall improved at mid-term follow-up and were comparable with other series of IB-RSA [27,30] or standard RSA [6–10,16]. Interestingly, patients with a 44 mm diameter showed significantly better external and internal rotation, better CS, and a positive trend for better anterior elevation. Biomechanical studies showed that increasing the glenosphere diameter may have a favorable effect on range of motion [31–33]. Clinically, some authors found similar improvements when increasing the diameter to 42 or 44 mm [34,35]. The use of bigger glenospheres is advisable in order to improve the results. However, this finding should be taken with caution, since it is not possible to use a 44 mm glenosphere in all patients due to technical issues.

Regarding complications, only one patient was revised because of an infection and two patients had an acromial fracture. The reported rate of acromial fracture with classic reverse design is generally lower [1,36], but due to the low number of patients included in this series, it is impossible to evaluate whether increasing the glenosphere diameter could be associated with a higher risk of acromial fracture. However, Kohut et al. [27], in a larger series comparing 40/44 mm with 36 mm glenosphere, did not report an increased risk of acromial fracture. One important phenomenon that we observed in three cases is an initial glenoid migration, which stabilized within the first year and remains in a high and severe superiorly tilted baseplate (Figure 2). We noticed that in all these cases, the superior part of the glenoid was not perfectly in contact with the subchondral bone, probably due to an incorrect technique and uncompleted cartilage removal. Based on this observation, we recommended a good preparation of the subchondral bone in order to match the baseplate profile.

This study presents the following strengths: the series included all consecutive patients prospectively enrolled during the index period; the same prosthesis and the same technique was used in all the cases; radiographical and clinical evaluation was performed by two surgeons (R.R. and L.D.) not involved in the surgical procedure. However, this study presents some limitations. Firstly, its retrospective nature. Secondly, the lack of a control group. To definitively confirm our observation, comparative studies comparing IB-RSA and standard RSA will be needed in the future. Thirdly, because of the limited number of patients included it is impossible to draw definitive conclusions regarding complications and revisions rate. Fourth, other anatomical measures (i.e., scapular neck angle, neck-shaft angle, glenoid inclination) that may have a role in scapular notching were not measured.
