Implant wear occurs primarily at the articular interface, generating debris that results in the destruction of surrounding tissue secondary to inflammation. The destruction is two-fold: damage to the articulating surface of the prosthesis can be detrimental to implant stability, and the debris generated by implant wear can drive inflammation [8]. Debris may originate from multiple implant compositions, including polyethylene, metal, and cement. Generated debris can then implant on the articular surface of a polyethylene prosthesis, further exacerbating implant wear by enhancing abrasion [9].

Phagocytosis of debris less than twelve 12 µmmicrometers in diameter by macrophages underlies the primary pathogenesis of periprosthetic osteolysis; however, the specific inflammatory response is dictated by the quantity and quality of the particulates regarding size, surface area, and composition. Further, the relative concentration of debris, rather than simply the number of particles, dictates the magnitude of the inflammatory response [10,11]. Macrophage stimulation after debris phagocytosis results in the release of tumor necrosis factor-alpha (TNF-α), interleukin (IL)-1β, and IL-6. TNF-α and IL-6 are catabolic mediators in bone, and IL-1β induces the differentiation of osteoclasts and the production of matrix metalloproteinases (MMPs) that promote bone resorption [11,12]. Polyethylene debris is also associated with complement (CR3) activation, resulting in more macrophage recruitment [13].

Cement debris resulting in larger particulates not amenable to phagocytosis is associated with giant cell recruitment and toll-like receptor (TLR) stimulation, which, in turn, activates the inflammatory nuclear factor kappa-light-chain-enhancer of activated B cells (NF-κB) cascade [14]. The receptor activator of nuclear factor-kappa-Β (RANK) and its ligand (RANKL) bind on preosteoclasts, stimulating osteoclastogenesis via the NF-κB transcription factor pathway and, in turn, causing bony resorption. As such, NF-κB is the transcription factor most commonly implicated in osteolysis; it is activated by several mechanisms, including those mediated by TLRs, TNF-α, and IL-1 [15,16]. In summary, periprosthetic osteolysis, characterized by concomitant inflammation, fibrosis, and bony resorption, occurs as an aseptic chronic inflammatory response to intra- and periarticular debris.

Histologic findings of periprosthetic osteolysis include inflammatory cells (lymphocytes, histiocytes, plasma cells, giant cells, and macrophages), which may contain identifiable particulate debris; clefts containing strongly birefringent polyethylene debris; and scalloped edges where cement has been resorbed. Interestingly, Kepler et al. [9] reported no significant difference in the frequency of polyethylene debris between patients with and without osteolysis after anatomical TSA (62% vs. 67%), indicating that the presence of particles alone is not predictive of osteolysis [9]. In cases of osteolysis in the absence of debris on histologic analysis, the pathogenesis of bone loss is currently unknown.

Detritic synovitis is an inflammatory response to intraarticular debris, which causes more widespread osteolysis beyond the periprosthetic space, resulting in implant loosening or pathologic fracture [17]. Known to cause implant failure in hip, hand, and foot arthroplasty, detritic synovitis leading to osteolysis after anatomical TSA was first described in 2018 [17-19]. Guild et al. [20] described an inflammatory foreign body reaction to polyethylene implant wear resulting in osteolysis; histopathologic analysis found multinucleated giant cell and histiocyte infiltrates and polarizable debris resulting from the destruction of bone and joint tissue. Detritic synovitis and periprosthetic osteolysis share many histological characteristics; however, the scope of their consequences differs given the relative lack of geographic limitation seen in detritic synovitis.

Variations in component composition, component positioning, and stem length have been the mainstay approach to reducing implant wear and associated debris, inflammation, and osteolysis. Cemented all-polyethylene glenoid components are associated with 83% or greater survival rates at year 10 of follow-up; however, wear and revision rates vary between polyethylene models [34,35]. Cross-linked polyethylene has been associated with an 85% reduction in wear rates relative to traditional polyethylene components as well as lower revision rates at year 5 of follow-up [36,37]. Metal-backed glenoid components are associated with substantially lower survival rates at long-term follow-up than all-polyethylene glenoid components, with failure attributed to aseptic loosening in the all-polyethylene group and rotator cuff insufficiency and instability in the metal-backed group [38]. Hybrid glenoid implants, which feature porous metal central posts and no metal backing to the glenoid surface, have not been associated with significant differences in complication rates at year 3 of follow-up relative to all-polyethylene implants [39]. Friedman et al. [40] report that the hybrid glenoid component is superior to the all-polyethylene implants with regard to 3-year revision rates, though longer-term investigations on the longevity of these implants are still needed.

Research investigating the means to reduce osteolysis surrounding humeral implants has largely focused on stem length and implant composition. Bell et al. [41] demonstrated decreased rates of radiolucent lines and humeral osteolysis in stemless ceramic humeral components when compared to long-stem metal-head alternatives. Long-stem designs are associated with stress shielding of the proximal humeral metaphysis, resulting in increased bone resorption, while the opposite is true for stemless humeral component designs. Indeed, stemless designs have been demonstrated to better mimic intact bone [42,43]. Investigations of humeral implant composition have demonstrated a decreased wear rate associated with ceramic humeral heads when compared to metallic components [44].

Comprehensive postoperative follow-up and physical evaluation should be performed in the setting of new-onset pain following TSA regardless of the time from the index procedure. The most sensitive indicator of osteolysis following TSA is new-onset or persistent pain [9]. However, postoperative pain is non-specific and should prompt a comprehensive evaluation of other etiologies. Other considerations that may induce pain after TSA include periprosthetic infection, periprosthetic fracture, stiffness, rotator cuff pathology, heterotopic ossification, bursitis, and malalignment. It should be determined whether the pain is associated with weakness or decreased motion, as this may leverage insight into additional shoulder stabilizer involvement and displacement of the glenoid or humeral components.

Importantly, the timing, quality, responsiveness, location, and duration of symptoms can provide more insight into the potential pathology. For example, pain secondary to glenoid or humeral osteolysis is generally experienced when sleeping or first initiating movement (start-up pain) of the shoulder and is diffuse in nature, whereas well-localized pain over the posterosuperior aspect of the shoulder may represent an acromial stress fracture. Pain in the proximal part of the upper extremity can indicate humeral component loosening. Patients should also be asked about wound issues and drainage after the index surgery, as this may elevate indolent infection as a cause of symptoms. Concern for possible osteolysis and aseptic loosening should be raised for patients who report years of symptom-free shoulder function postoperatively followed by new-onset pain or reduced function.

The physical exam should be performed systematically and include inspection, palpation, range of motion, strength, and provocative maneuvers where appropriate. Specifically, the surgical incision and skin around the shoulder should be assessed. The presence of effusion, erythema, or swelling may indicate chronic inflammation or infection. Diffuse tenderness to palpation around all areas of the shoulder in the absence of other findings may signify a chronic pain syndrome.

The extent of passive and active range of motion should be assessed, with particular attention directed towards instability, impingement, or pain along short arcs of motion. Patients with osteolysis that begin to experience early subsidence may experience loss of function. Atrophy or deformity in the setting of a primary anatomical TSA may suggest compromise of the rotator cuff. Finally, a thorough neurovascular exam should be assessed to rule out neurovascular compromise as the etiology of pain and dysfunction.

Though osteolysis is characteristically a chronic process associated with night or start-up pain, it is notable that early osteolysis may manifest non-characteristic symptoms. Therefore, in the setting of painful TSA, early osteolysis should still be considered with a thorough evaluation of routine radiographic imaging. Indeed, early osteolysis that is rapidly progressive without identification and treatment can result in glenoid loosening, subsidence, and early failure.

In all scenarios where a patient presents with a painful TSA, standard laboratory testing including complete blood count, erythrocyte sedimentation rate, and C-reactive protein measurement should be obtained. If these raise suspicion for infection, such as if the synovial leukocyte count exceeds 2,000 and is composed of at least 70% polymorphonuclear leukocytes [45], an ultrasound-guided shoulder aspiration is warranted. However, in cases with a high index of suspicion for infection but negative laboratory and aspiration work-up, arthroscopic or open tissue biopsy is considered a gold standard diagnostic tool for infection. If periprosthetic joint infection has been ruled out, osteolysis and aseptic loosening then rise among the differential diagnoses as the culprit of shoulder pain [46].

Postoperative radiographs are the first-line imaging modality to evaluate for osteolysis in the proximity of either the glenoid or humeral components. Standard views of the shoulder, including anteroposterior, Grashey, lateral, and axillary views, should be obtained. The examiner should evaluate radiographs for radiolucencies and stress shielding adjacent to the glenoid and humeral components. Comparison to prior radiographs should be made when available, particularly when monitoring the progression of previously diagnosed osteolysis. The examiner may observe implant loosening, malpositioning, or subsidence. Particular attention should be focused on the location of the humeral head, as proximal migration may indicate a supraspinatus tear, and anterior displacement may suggest a tear of the subscapularis.

In non-cemented humeral components, radiolucent lines often occur at the tip of the prosthesis, whereas radiolucencies commonly develop along the proximal and midbody aspects of the stem in cemented humeral components. In some smaller series with 10 years of follow-up, over 50% of patients developed radiolucencies, most often in association with glenoid wear and polyethylene debris [47]. However, the clinical significance of osteolysis remains unclear in certain populations, as asymptomatic patients with osteolysis do not always require a revision procedure [48].

It also appears that the choice of humeral fixation technique is not associated with osteolysis on radiographs. A recent randomized controlled trial with a mean 38-month follow-up period reported a 0.74% incidence of radiolucencies ≥2 mm in three or more zones, which did not significantly differ between cemented and non-cemented humeral component cohorts [49]. Scapular notching may be a more obvious finding of progressive osteolysis. In a 10-year follow-up study of patients treated with a Grammont-style RSA, 73% of patients developed scapular notching on radiographs, with 12% having undergone revision surgery [50].

Unfortunately, radiolucent lines on plain radiographs do not always reliably diagnose loosening, particularly during long-term follow-up, as some series report the presence of radiolucent lines in up to 80% of radiographs at 10 years of follow-up [51]. Therefore, it may be more appropriate to evaluate radiolucent line progression over time, as opposed to making a definitive plan of care based on a single observation. Advanced imaging appears to be more sensitive than radiography at detecting radiolucency. Recent studies have demonstrated that the reliability of computed tomography (CT)-based assessments of radiolucencies is three times higher than that of radiographs, and up to 40% of radiolucent lines and 74% of osteolysis cases not observed on radiographs are detectable by CT [51,52].

In cases of osteolysis following RSA, notching of the polyethylene liner against the inferior border of the scapula should be assessed. This mechanical impingement can potentially lead to a high level of particulate debris, leading to osteolysis in both the glenoid and the humerus. In severe cases, significant osteolysis can occur at the inferior glenoid, directly affecting baseplate fixation. In the evaluation of osteolysis after RSA, component malposition should be recognized early and potentially revised to prevent further osteolysis.

For patients with symptomatic osteolysis and evidence of glenoid loosening following anatomical TSA or RSA, nonoperative treatment is generally reserved only for patients that are poor surgical candidates and medically unfit for surgery. This approach relies on secondary stabilizers to maintain the functional integrity of the shoulder. To solidify the surrounding soft tissue architecture, nonoperative treatment consists of a 4–6-week period of sling immobilization during which active and passive range of motion are deferred. Whereas all surgical treatment options to be discussed in this article carry a significant risk for complications, non-surgical management mitigates the risk of surgery-related complications. In a retrospective analysis of 79 patients diagnosed with aseptic glenoid loosening following RSA, Ladermann et al. [53] demonstrated that a sub-group of 29 shoulders treated nonoperatively had similar clinical improvements and fewer associated complications compared to a group of 27 shoulders that underwent revision. Furthermore, in similar cohorts of patients, nonoperative treatment resulted in better clinical outcome scores than revision to hemiarthroplasty.

In postoperative anatomical TSA patients with isolated aseptic glenoid loosening and suspected infection, arthroscopic removal of the polyethylene glenoid component offers an appealing surgical option [54,55]. Given the high suspicion for periprosthetic infection and concurrent difficulty in diagnosing indolent Cutibacterium acnes infection in this clinical scenario, an arthroscopic procedure enables the clinician to obtain intraoperative tissue samples to aid in diagnosis while performing a minimally-invasive glenoid resection that may provide significant symptomatic relief. Removal of the polyethylene component reduces debris created by contact between the glenoid component and the adjacent metal and bone [54]. To address cavitary bone defects caused by prior osteolysis, bone graft, in the form of corticocancellous bone chips, can be introduced arthroscopically through an enlarged anterosuperior portal and tamped into the glenoid cavity until the void is filled [56]. The use of a human dermal allograft patch has been described to help contain the bone graft within the defect [56]. Clinical outcome data regarding dermal allograft procedures have yet to be published. However, the available literature supports the use of isolated arthroscopic glenoid resection. Indeed, a cohort of 15 patients who underwent glenoid component resection with or without bone grafting had comparable pain relief and satisfaction at 2 years postoperatively in comparison to patients that underwent glenoid reimplantation [57]. Given that the authors noted a selection bias in that lower-demand patients were more likely to undergo an isolated glenoid component resection, further evidence is needed to delineate the optimal patient characteristics for this intervention. Nonetheless, in patients whose clinical presentation remains concerning but non-diagnostic for infection, an arthroscopic glenoid component resection with tissue culture procurement is a reasonable temporizing option.

Another commonly employed surgical treatment option for aseptic glenoid loosening following anatomical TSA is revision anatomical TSA with reimplantation of another polyethylene glenoid [58]. The greater potential shoulder range of motion conferred by an anatomical TSA, in comparison to that of an RSA, has made this an enticing revision option, particularly in younger patients [59,60]. In an early report of outcomes from this intervention, Cheung et al. [61] compared 33 shoulders with glenoid reimplantation to 35 shoulders with glenoid implant removal and bone grafting. Five-year postoperative outcome data demonstrated a 91% reoperation-free survival and higher satisfaction in patients that underwent reimplantation compared to those that underwent implant removal and bone grafting. However, in subsequent work, the authors conceded that these findings might have been confounded by the inclusion of infected arthroplasty cases in their analysis [62]. Subsequent studies have demonstrated that revision anatomical TSA indicated for glenoid loosening has been fraught with complications. In one analysis of 42 patients with symptomatic glenoid loosening following a primary anatomical TSA who underwent an isolated glenoid exchange, 67% of patients had recurrence of glenoid loosening, and 17% required reoperation at approximately year 6 of follow-up [63]. Sheth et al. [64] corroborated the disappointing results of revision anatomical TSA in a cohort of 20 patients, reporting that 35% required reoperation within 3 years of revision surgery. The poor clinical outcomes and increased complications associated with revision anatomical TSA and isolated glenoid component exchange make its contemporary use in the current treatment of glenoid component loosening relatively minimal.

In the setting of a failed anatomical TSA due to osteolysis or glenoid component loosening, conversion to RSA affords several advantages over revision anatomical TSA (Fig. 2). Whereas anatomical TSA requires an intact and functioning rotator cuff for optimal outcomes, RSA outcomes can be satisfactory even with rotator cuff deficiency, which is often present in patients undergoing revision shoulder arthroplasty [65]. Second, on the glenoid side, RSA allows for both stronger fixation with screws and posts, as well as bony ingrowth for greater potential implant longevity [65]. Third, RSA allows the surgeon to not only address the aforementioned bone loss with the increasing popularity of augmented baseplates [66]. The available clinical evidence bores out these advantages. A multicenter study of 37 anatomical TSAs revised to RSA for aseptic glenoid loosening demonstrated an 86% patient satisfaction rate at approximately year 4 of follow-up [67]. The authors reported a 21% reoperation rate and, therefore, cautioned that, despite a high satisfaction rate, patients must be counseled on the elevated risks of reoperation compared with an index operation. Currently, RSA affords the most predictable surgical solution for symptomatic aseptic loosening and osteolysis following shoulder arthroplasty. Further elucidation of optimal patient candidates and long-term clinical outcomes of RSA used in this setting are needed.

Humeral component loosening secondary to osteolysis surrounding the humeral implant is exceedingly rare. In a radiographic study of 395 shoulders that previously underwent either hemiarthroplasty or total arthroplasty, 43% of shoulders demonstrated evidence of osteolysis at either the greater tuberosity or calcar [48]. Despite this, humeral component loosening was not observed in any of the uncemented stems, and only one cemented stem was deemed to be at risk for humeral loosening based on the morphology of radiolucent lines surrounding the implant. In the single published case series on the management of humeral component aseptic loosening, Cil et al. [68] reported on 38 cases over a nearly 30-year period. The authors used cancellous bone grafting to treat contained proximal humerus osteolysis in approximately one-third of cases. Due to more extensive bone loss, custom humeral stem implants were employed in two cases. At revision, cement humeral fixation was utilized in approximately 75% of cases. Some authors postulate that the expanded use of stemless humeral implants in shoulder arthroplasty will further minimize the risk of proximal humerus osteolysis; however, further studies are needed to evaluate the impact of stemless humeral designs [69].

There are limited data available to guide clinicians in the management of aseptic glenoid loosening following RSA. As mentioned, nonoperative treatment should be pursued as a first-line treatment option in minimally symptomatic patients. For patients unable to tolerate nonoperative management, glenoid loosening should be treated with revision of the glenosphere. Ladermann et al. [53] reviewed 79 patients treated for aseptic glenoid loosening. Among this cohort, patients treated nonoperatively, and those treated with glenoid revision experienced similar improvements in pain, range of motion, and clinical outcomes scores at a minimum 2-year follow-up. As the number of RSA procedures continues to grow, so too will our collective experience with managing its associated complications, including osteolysis and aseptic loosening. A select example demonstrating our institutional experience with the management of osteolysis after RSA is described in Fig. 3.