Total shoulder arthroplasty (TSA) remains the gold standard for treating osteoarthritis of the shoulder in patients with a functional rotator cuff and sufficient glenoid bone [1-3]. Outcomes and survivorship data from available registries support the use of cemented glenoids over uncemented ones [4,5]. However, there is growing interest in uncemented polyethylene (PE) glenoids, which feature a thin layer of titanium sprayed on the back, allowing uncemented insertion. This specific implant has held a CE mark since 2013 and has been utilized in selected centers in Europe since 2018. This concept has a long and successful history in total hip arthroplasty [2,6]. These implants have significantly lower elasticity than solid metal-backed glenoids, with elasticity levels that are more similar to those of bone [7].

Loosening of the glenoid component remains the primary issue in the long-term assessment of TSA [8,9]. Radiolucent lines (RLLs) can be observed immediately after implantation of cemented glenoids, likely due to technical factors such as inadequate cementation techniques (e.g., incomplete cement mantle, movement of the implant before the cement has hardened, or incomplete seating). The immediate appearance of RLLs around uncemented glenoids can occur due to factors such as insufficient reaming or a mismatch in convexity between the glenoid and the implant. These immediate RLLs are distinct from those that develop over time, which are believed to arise from different mechanisms, including wear, PE debris, and loosening.

Although RLLs are primarily a radiologic finding, their progression over time can indicate implant loosening, potentially occurring with a delay [8]. This ultimately leads to implant migration, bone loss, and pain. Even when cemented glenoids are radiographically classified as loose, revision surgery remains relatively uncommon. This phenomenon can be attributed to the lower functional demands of older patients and the relatively forgiving nature of cemented implants, especially in comparison to rigid metal-backed components. However, as patients undergoing TSA are becoming younger and more active, there will likely be an increasing burden in the future.

To address these issues, a novel titanium-sprayed glenoid implant (RM Pressfit Vitamys) was developed. However, evaluation of RLLs in x-rays of this new type of implant is needed. The thin radiopaque titanium layer on the convex PE component, which faces the sclerotic convex reamed bone, may simulate RLLs. Additionally, tilting during imaging can obscure the presence of RLLs. Furthermore, in a clinical setting, anteroposterior (AP) radiographs are rarely obtained in a perfectly straight AP view, often exhibiting some degree of craniocaudal and mediolateral tilt [10]. Therefore, we conducted a cadaveric study in which we implanted these novel implants into human glenoids and examined the appearance and disappearance of RLLs at various tilts of the glenoid relative to the x-ray beam. We hypothesized that minor tilting of the glenoid would affect the presence or visibility of RLLs on plain radiographs of uncemented PE glenoids. These findings will assist healthcare providers interpret x-rays.

The study was approved by the Ethics Committee of Heidelberg University (No. S-476/2022). Donors had provided written informed consent for body donation and scientific use prior to death. Specimens were obtained from the Institute of Anatomy at Tübingen University.

Both scapulae of four specimens (three males, one female, age ranging from 72 to 85 years) were explanted from whole body specimens preserved with an ethanol based embalming solution (70% ethanol, 30% glycerol). Exclusion criteria were diseases like malignant disease, bone disease or cysts, a metabolic disease that affected bone quality, or a bone mineral density <0.6 g/cm³. Bone density was measured using the dual x-ray absorptiometry (Horizon Wi, Hologic) method. x-rays confirmed that all glenoids used had a type A configuration according to Walch et al. [3]. First, scapulae were embedded in an orthograde fashion into epoxy resin with the glenoid surface parallel to the surface of the epoxy resin block. Then, an experienced shoulder surgeon (MB) implanted n=8 RM Pressfit Vitamys glenoids (Mathys) according to the manufacturer’s manual and a recent publication [9]. The size of the glenoid implants was determined using AP radiographs and TraumaCad software version 2.5 (Brainlab). This software is used to digitally template and plan orthopedic surgeries.

Inter-rater reliability was adequate (Cohen’s=0.369; 95% CI, 0.248–0.491), while average intra-rater reliability of the two evaluators was good (Cohen’s=0.667; 95% CI, 0.639–0.743). Tilting of the glenoid away from the exact orthograde x-ray beam can obscure signs of incomplete seating of the implant on the bony surface as an x-ray is a summation image. In all zones, tilting from 0° to 10° inferiorly (P=0.028), 10° retroversion (P<0.001), and 20° anteversion (P<0.001) caused RLLs to appear or disappear (Table 1, Figs. 3 and 6). Tilting from 0° to 20° inferiorly (P=0.087), 10° (P=0.218) and 20° superiorly (P=0.484), and 10° anteversion (P=0.126) did not cause any RLLs to disappear or appear. RLLs always disappeared with a tilt of 20° retroversion (Fig. 3).

Evaluation of individual zones 1, 2, 4, and 5 did not reveal any significant findings. RLLs were mainly detected in zone 3. RLLs only appeared or disappeared significantly in zone 3 when the glenoid was tilted 10° inferiorly (P=0.014), in 10° retroversion (P=0.032), or in 10° (P=0.015) and 20° anteversion (P=0.004). There was no significant change when tilted at 20° inferior tilt (P=0.317) or 10° (P=0.135) and 20° superior tilt (P=0.082).

Our data demonstrate that tilting the novel uncemented titanium-sprayed PE glenoid in relation to the x-ray beam can cause RLLs to appear or disappear in some of the examined positions. This finding has a direct impact on the evaluation of this novel uncemented glenoid: if you cannot be sure that your x-ray projection is perfectly AP—which can be a challenge in daily practice—then you should suspect that RLLs are under- or overestimated.

Evaluating the ideal seating of the glenoid and the incidence of RLLs are important for long-term survival [12]. Therefore, this evaluation has always been at the focus of survivorship analysis [13]. However, the true clinical impact of the presence of RLLs is not clear [8,13,14]; the clinical outcome can still be sufficient, even if RLLs are present around the glenoid implant [15-18]. However, one study reported an increase in RLLs at the back of cemented and uncemented glenoids over time [14]. With a delay of 5-8 years, the incidence of loosening of the glenoid, as defined by migration and/or subsidence, also increases [14]. Patients with radiologic signs of loosening have been shown to have inferior clinical outcomes [15].

If surgeons observe RLLs on the postoperative x-ray, they might infer that the glenoid was not perfectly implanted in terms of 100% bone contact between the titanium and bone. This does not mean that bone cannot grow toward the titanium in the future. This can be possible in areas of adequate micromotion [19,20], around the center of the glenoid where, for example, pegs or keels are placed. Biomechanical studies have shown that the limit of 150-micrometer micromotion is regularly exceeded at the edges of the implant [21]. In our study, RLLs were seen not around the pegs but at the edges and the surface of the glenoid. This finding makes it unlikely that bone will grow in-between an RLL at the edge of the implant.

A previous study of cemented glenoids highlighted the problem of exact evaluation of RLLs [10]. Zwingenberger et al. [10] examined which landmarks are most reliable if the scapula is tilted. The authors showed that the medial margin of the scapula was best and that the lateral margin of the scapula had acceptable reliability. To measure medial migration, the coracoid baseline has been proven to have acceptable reliability, whereas the glenoid fossa line is subject to change if osteolysis occurs at the glenoid [10]. The amount of tilting in the current study was derived from the study of Zwingenberger et al. [10] who examined craniocaudal tilting at ±15° and ±30° and mediolateral tilting at ±10° and ±20°; higher degrees of tilting >20° did not lead to higher incidence of RLLs but did lead to lower detection rates. Therefore, tilting angles greater than 20° were not examined. Zone 3 (between the pegs in the AP view) was the most sensitive to concealed RLLs due to implant tilt. One possible explanation for this finding is that the implantation technique more frequently resulted in RLLs in this zone, achieving statistical significance.

There are limitations to this study: a sample size of n=8 might be considered low. However, this sample size yielded sufficient power to assess the importance of anatomic landmarks subjected to tilting in x-rays in a similar study (Zwingenberger et al. [10]). Interrater reliability was calculated as adequate, which might indicate observer-dependent variability. Automated image analysis tools might enhance consistency in future studies.

The current study was a time-zero cadaveric investigation conducted under ideal conditions, which ensured an optimal fit and the absence of bony ingrowth. However, our study did not aim to address the long-term implications of RLLs. Rather, we were able to offer new and valuable insights based on our observations from postoperative and follow-up x-rays focusing on the radiologic findings associated with new uncemented PE glenoid implants that feature a thin layer of titanium on their convex surface.

Despite the ideal implantation conditions, we noted some minor gaps between the implant and the bone that might be challenging to detect in tilted glenoids on two-dimensional x-ray images. It can be argued that the convex titanium layer, in conjunction with the PE glenoid, might mimic an RLL in certain projections; however, our findings indicate that this is not the case.

Our study aims to assist surgeons and healthcare providers in counseling their patients regarding the observation of RLLs. It is important to emphasize that RLLs do not always lead to implant loosening. Previous studies examining cemented glenoids have demonstrated that the number of RLLs tends to progress over time [14,22]. Consequently, the progression of RLLs in both number and thickness can raise concerns regarding cementless implants. The findings from this study cannot be generalized to other solid metal-backed glenoid designs currently available on the market. [23]. The implant we examined features only a thin layer of titanium on the back of the PE, resulting in an elasticity comparable to that of bone, which is significantly lower than that of the solid metal-backed glenoids available on the market. Several solid metal-backed glenoids have demonstrated failure in registries during mid-term and long-term observations, primarily due to issues related to loosening [23,24].

Reasons for failure of glenoid components include backside wear, overstuffing, and stress shielding. This new generation of implants could potentially reduce these issues for three reasons. First, the titanium is directly bonded to the PE, eliminating the possibility of backside wear. Second, the PE is sufficiently thick, allowing proper anatomical reconstruction without over-tensioning the rotator cuff and preventing overstuffing. Third, stress shielding is minimized because the Young's modulus of the entire implant is comparable to that of bone.

However, correct implantation of both the humeral and glenoid components, with an anatomic neck cut, is crucial to ensure proper anatomical reconstruction and avoid overstuffing. Additionally, we emphasize that the present study is a time-zero cadaveric experiment conducted under ideal conditions, without clinical follow-up. Therefore, the theoretical advantages of the novel glenoid have yet to be validated. Clinical data must be provided in separate studies. 

Postoperative radiologic interpretation of a novel PE glenoid with a thin layer of titanium sprayed onto its posterior surface poses challenges. Tilting the glenoid away from the true orthograde x-ray beam can obscure proper seating of the implant on the bony surface. When evaluating all zones, it is recommended to avoid tilting 10° inferiorly, ±10° in anteversion and retroversion (the medial-lateral plane), or 20° anteversion during radiological examinations. Careful attention should be paid to obtaining a true AP image during the x-ray check following implantation of an uncemented glenoid component. Additionally, a CT scan is advised to confirm or rule out RLLs.