Introduction

Cervical total disc replacement (cTDR) has been established as a reasonably safe and effective alternative treatment for degenerative cervical radiculopathy and myelopathy, with the potential to reduce the incidence of adjacent segment disease by providing motion preservation at the treated level [1, 2]. Prospective, randomized clinical trials have established the safety profile of cTDR with long-term follow up, and based on the supportive clinical evidence, increased acceptance by spine surgeons and predictable reimbursement by insurance carriers has expanded utilization [3]. First approved for use in the U.S. by the FDA in 2007 [2], there are now multiple discs available on the U.S. market . By 2019, prior to healthcare disruptions caused by the global pandemic, there were approximately 31,700 cTDRs performed in the US, as compared with 352,000 cervical fusions [4].

The rate of complications for cTDR is reasonably low based on published clinical trials [3]. However, as the number of trained cervical arthroplasty surgeons expands, so too will the incidence of complications and revision surgery as part of routine clinical practice under real world conditions, outside the purview of strictly regulated FDA trials. There is ample published literature for the indications and outcomes of primary cTDR patients [5, 6]. Less is known about the clinical management of cTDR patients who experience complications and their outcomes.

Anterior bone loss (ABL; Fig. 1) around cTDR has received increased attention in recent studies, though questions regarding etiology, clinical presentation, and terminology remain [7]. Multiple processes may impact the nature, diagnosis, and treatment of bone loss adjacent to cTDR including stress shielding, avascular necrosis caused by surgical insult, periprosthetic infection, and the biological response to implant wear debris [8, 9]. Depending on the underlying cause(s) and degree of bone loss, patients may experience severe and persistent symptoms or may be asymptomatic [8,9,10]. While the characteristics and treatment of bone loss are well-reported for large joint arthroplasty [11, 12], comparable research for disc arthroplasty may be lacking.

Fig. 1
figure 1

Reprinted from Keiser et al. [13]

Radiographs of the lateral cervical spine of a representative case showing the natural history of moderate anterior bone loss in the upper endplate and severe anterior bone with collapse in the lower endplate (arrow).

Full size image

In response to uncertainty in the assessment of bone loss, our group developed and validated a classification system for periprosthetic bone loss based on plain radiographs, as the front-line assessment tool for cTDR patients in follow up [14]. The purpose of this current work was to develop a clinical management plan for cTDR patients once radiographic evidence of bone loss was established. To guide the development of our clinical management recommendations for periprosthetic bone loss, we undertook a review of the literature for bone loss, osteolysis, infection, and revision cTDR. We sought to address the following research questions: (1) how was bone loss identified and further worked up preoperatively with diagnostic imaging; (2) were preoperative and/or intraoperative assessments of infection performed; and (3) if revised, was histological and/or explant confirmation of bone loss performed; and (4) what decision-making and revision strategies were employed? The aims of this investigation were to (1) use the findings from this review to form a basis for treatment recommendations, and (2) identify gaps in the literature to inform the design of future prospective studies of cTDR revision outcomes.

Methods

Literature search

A systematic search of the literature was performed using PubMed and Embase. The searches were performed on July 17, 2023, using the following search terms: (cervical AND spine AND (disc OR disk) AND (replacement OR arthroplasty) AND (revision OR infection OR “implant failure” OR “bone loss” OR remodeling OR osteolysis OR wear OR debris OR particles OR “failed osseointegration” OR “failed osteointegration”). Additional details of the search terms are available in the supplementary material.

Duplicates, conference abstracts without full papers, and non-English publications were excluded, and the remaining titles and abstracts were screened. Full texts were obtained for all studies meeting the inclusion criteria, specifically, those which report instances of bone loss in cTDR. Title/abstract screening and assessment of full texts according to the established inclusion and exclusion criteria were carried out by two independent researchers. The complete list of inclusion and exclusion criteria can be found in Table 1.

Table 1 Inclusion/exclusion criteria for study screening
Full size table

Data extraction and analysis

Following final paper selection, basic study details including patient number and follow-up time were captured along with procedure characteristics (i.e., cTDR device design, surgical levels) and patient outcomes (i.e., reports of bone loss, instances of revision/reoperations). Methods for imaging, characterizing, and treating bone loss were also captured. Some cohort studies also detailed individual patient cases, which are summarized along with the case studies. Data extraction was performed by one author and reviewed for accuracy by an independent researcher.

Results

Study selection

The database searches yielded 260 studies from PubMed and 266 from Embase, for a total of 353 unique papers. Twenty-one additional articles were found from searching source bibliographies, resulting in 374 papers for initial screening. Ultimately, 31 studies were included in our systematic review (Fig. 2). This included 11 case studies and 20 cohort studies. Summary data from the reviewed case reports and cohort studies are provided in Tables 2 and 3, respectively.

Fig. 2
figure 2

PRISMA [15]flow diagram for study selection

Full size image

Study characteristics

Case reports provided the most detail regarding bone loss management. Out of the 15 cases presented, the first indication of bone loss was most frequently noted via CT scan (n = 8) versus radiograph (n = 3). Secondary imaging with CT, radiograph, and/or magnetic resonance imaging (MRI) was common. Preoperative assessment of infection was reported in 8 cases, and 2 cases also included nuclear medicine studies. Of the 13 cases revised, 6 reportedly included intraoperative infection workup and 7 had histological and/or explant assessment. For cohort studies, it was not clear how many individual cases involved these methods, but some histological and/or explant analysis findings were reported in 10% (n = 2/20) of the studies and infection workup was reported in 15% (n = 3/20).

The reviewed case reports and cohort studied studies represented 2073 patients (one study represented a subcohort of an earlier study and was excluded from this summary [13]). The most frequently used devices were Prestige-LP (Medtronic, Minneapolis, MN, n = 828), Prodisc-C Vivo (Centinel Spine, West Chester, PA, n = 285), and Bryan (Medtronic, n = 266). Notably, some of the implants included in the reviewed studies are no longer clinically available. There were 821 patients reported to have bone loss, 5.7% (n = 47/821) of which underwent revisions/reoperations for reasons directly related to the condition (this number may be higher, as some studies did not clearly report the occurrence of revisions).

Table 2 Summary of case studies reviewed. Case report details included in the cohort studies from Ebinu et al. and Kim et al. are also included
Full size table
Table 3 Summary of cohort studies reviewed
Full size table

Discussion

Periprosthetic bone loss around cTDR has been receiving increasing attention, but clinical management of the complication is not yet well-defined. In this study, our aim was to propose a clinical management plan for cTDR patients with evidence of bone loss. A review of the literature was performed to guide the development of our plan. We sought to address how bone loss was identified and imaged, what pre- and intraoperative assessments of periprosthetic tissue were undertaken, and what decision-making or revision strategies were employed.

Our study has limitations. Given the limited and poor quality of the literature, a robust evidence-based clinical management plan is not feasible. There was significant heterogeneity across the literature in terms of clinical methods and reporting nomenclature, making it difficult to identify standard practices in the treatment of cTDR bone loss. For example, we were limited in our summarization of bone loss severity given the multiple classification methods used. This overall inconsistency highlights the usefulness of our proposed clinical treatment paradigm [14]. Additionally, the cohort studies, which were mostly retrospective analyses of routine radiographs, generally lacked a meaningful analysis of the bone loss cases reported, leaving their clinical significance unclear. We instead rely on the case studies to provide context for clinical relevance, while the cohort studies provide a more global view of how clinicians classify (i.e., grade) the loss and how often revisions occur due to bone loss. Finally, most of the cohort studies reviewed involved short-term (< 5 year) follow-up, and the results may not be representative of long-term clinical performance. While short-term cTDR performance is of particular interest due to the occurrence of early and nonprogressive bone loss, the potential for osteolysis to develop makes longer-term results equally as important. Increased analysis with longer follow-up will be necessary as the data becomes available.

How was bone loss identified and further worked up preoperatively with diagnostic imaging?

It was clear across the cohort studies that most bone loss was identified during routine radiographic examination, often performed at intervals of at least 3-, 6-, and 12-months. In some cases, additional imaging with CT or MRI was also noted. By contrast, in many case reports the bone loss was identified through CT imaging, with radiograph and MRI serving as secondary methods (though earlier radiographs with unremarkable or inconclusive findings may have been collected). The differences in initial bone loss identification methods likely relate to the difference in study types; most cohort studies involved retrospective radiographic analyses while most case reports detailed investigations of patients already presenting with symptoms (e.g., progressive pain, paresthesia) or suspected bone loss. Regardless of how bone loss was first identified, some commonalities emerged including the preference of CT imaging over plain radiography for its sensitivity and 3-dimensional format [27, 41] and the recommendation that immediate postoperative radiographs be taken to serve as a baseline for later images [13, 23, 36].

Along with identification, bone loss was generally characterized in terms of severity, with the scale proposed by Keiser et al. or derivatives thereof widely used [31, 32, 40, 44]. This method is based on subchondral bone loss percentage (as opposed to other classifications that use implant exposure) in order to account for differences in implant size and positioning [13]. Bone loss is classified as mild (1–5%), moderate (6–10%), or severe (> 10%) with or without collapse. The scale relates the severity of bone loss to potential cause, radiograph presentation, implant exposure (No/Potential/Yes), endplate compromise (No/Yes), and treatment (Nil/Monitor). Alternatively, some studies determined the extent of bone loss using qualitative descriptions of bone resorption and implant exposure [29, 33, 36, 45] or quantified bone loss without classifying the extent [28, 43].

Were preoperative and/or intraoperative assessments of infection performed?

Periprosthetic bone loss may have multiple causes including stress-induced bone remodeling (stress shielding), periprosthetic infection, an inflammatory reaction to wear debris (periprosthetic osteolysis), avascular necrosis, and hydrostatic pressure. Thus, evaluation of bone loss via imaging is not enough to determine the underlying etiology. In the spine and other joints, infection has been identified as a potential contributor to bone loss [8, 46, 47], making it an important consideration in cTDR bone loss management. In the reviewed studies, infection was assessed preoperatively via complete blood counts and intraoperatively via culture collection. In the cases where an infection workup was initiated, active infection was seldom present, an unsurprising finding considering the reportedly low rate of cTDR and anterior cervical discectomy and fusion (ACDF) infection in general [48, 49].

Some additional diagnostic workups unrelated to infection are also worth noting. Two cases included bone scintigraphy to determine whether the observed regions of bone loss were related to malignancies; one showed hyperfixation in the vertebral bodies surrounding the implant but not in the bone defect itself, while the other revealed no increased activity [16, 26]. The case report from Tumialán and Wayne also included dermal patch testing for CoCr sensitivity, which was negative [26]. Finally, another study used single photon emission CT (SPECT) and found “no osteoblastic activity” [41].

If revised, was histological and/or explant confirmation of bone loss performed?

In addition to infection assessment, histological and explant analyses at the time of revision can provide a better understanding of the etiology of bone loss. Examination of periprosthetic tissue and retrieval analysis were reported primarily in cases with late onset bone loss (suspected osteolysis), implant failure, and/or suspected hypersensitivity [16, 19,20,21, 24, 25]. For other cases of revision that did not report similar analyses, it was not clear whether the evaluation was just unreported or explicitly not performed since most cases of bone loss were not suspected to be a result of foreign body response. When noted, the specific method used was hematoxylin and eosin (H&E) staining [19, 20, 25, 27]. Some studies reported that, like cultures for infection, tissue samples for histology were collected from multiple sites around the implant [27]. Histological examination revealed findings of granulation tissue with implant wear particles, foreign body giant cell reactions, and fibrinous inflammation [16, 19,20,21, 24, 27].

Some studies reported explant analysis or overall macroscopic findings of device damage, wear, and suspected failure modes [16, 19, 24, 26, 27, 41]. Ebinu et al. conducted a visual macroanalysis and photo-documentation of the explant to determine in vivo damage [27], while Tumialán and Gluf reported a dimensional assessment by the implant manufacturer [26].

What decision-making and revision strategies were employed?

Throughout the literature, some treatment and revision recommendations were provided. Decisions on whether to revise were often guided by the severity of symptoms and the progression and extent of bone loss [26, 27, 41, 42]. For patients with early bone loss, monitoring was recommended with the understanding that non-osteolytic bone loss often becomes nonprogressive after the first year and does not affect clinical or functional outcomes (based on patient-reported outcomes measures and complication rates)(Fig. 3) [13, 28,29,30,31, 33, 34, 36, 40]. The propensity for complications, infection, or osteolysis to lead to further complications during revision should also be considered [20, 21]. It was often unclear how soon revisions were performed after bone loss was confirmed, though a case with potential mechanical implant complications was treated especially urgently [23]. The involvement of multidisciplinary teams in decision-making was mentioned [21], and patient comorbidities, while influential to procedure outcomes, were mostly unreported.

Fig. 3
figure 3

Reprinted with permission from Elsevier [44]

Serial radiographs of a representative case underwent CDA with Prestige-LP disc showed the natural process of anterior bone loss (yellow arrow). (A) Preoperative lateral radiograph of the cervical spine. (B) 1-week postoperatively lateral radiograph showed CDA at C4/5 and C5/6 levels. (C) Moderate bone loss developed at both C4/5 and C5/6 levels at 3 months after CDA (arrow). (D–F) Postoperative radiographs showed the process of bone loss had stopped.

Full size image

Once the decision was made to reoperate or revise, the specific surgical strategy depended on the details of the case, specifically the extent of bone loss [23, 42]. Revisions primarily involved implant removal and conversion to ACDF [16, 21,22,23,24, 26]. Corpectomies were performed in some instances [19, 20, 25, 27]. Brophy and Hoh also described a case in which the implant was retained and the segment fused posteriorly [17]. Many studies reported the improvement or full resolution of symptoms following the revision. Choi and Lee also noted the resolution of the osteolytic process in a case [18].

Our proposed clinical management plan

Although the systematic review of the literature did not provide the robust evidenced-based approach to clinical management that we were seeking, the authors’ own extensive clinical experience (Level of Evidence V [50]; the surgeon co-authors together perform an estimated 500 + TDRs annually) can provide guidance for the management of patients with this increasingly prevalent clinical problem. Clinical presentation is paramount; the literature review as well as authors’ experience suggest that many patients are asymptomatic despite bone loss. Plain radiographs are recommended as the first method of assessment. Based on extensive arthroplasty experience with large joints, the authors believe that cystic erosion adjacent to (in continuity with) the endplate with diffuse margins, and cystic erosion not adjacent to the endplate are the most concerning plain radiographic observations and if identified, recommend close follow-up with serial radiographs every 6 months to assess the progression and severity of bone loss, as well as implant changes over time. If rapid bony changes are observed in serial radiographs, CT is recommended as a next step. CT findings can be used for clinical decision-making and further follow-up care. We also recognize that infection, implant migration, or subsidence can also affect clinical decisions regarding further non-operative follow-up versus surgical intervention.

Ultimately, even in the absence of symptoms, progressive bone loss that may compromise implant stability warrants a discussion about revision. The following is the list of clinical scenarios when revision surgery is suggested by the authors: In patients with high suspicion of infection, either as a result of clinical presentation or diagnostic work-up; rapid radiographic progressive cystic bone loss with diffuse margins either adjacent or non-adjacent to the end plate further confirmed by CT scan; substantial bone loss involving more than 1/3 of the bone/implant interface; substantial and/or progressive migration of the implant or mechanical failure of the implant posing a pending threat to the dorsal neural or ventral vital structures.

Conclusions

Our findings demonstrate a critical gap in the literature regarding the optimal management of patients with bone loss following cTDR. Multicenter, prospective outcome studies of patients undergoing revision surgery for bone loss following cTDR, which include implant retrieval analysis, histopathology, and microbiological assays, are needed. Such future studies may help elucidate the etiologies of bone loss and how surgical, patient, and implant factors may contribute.