Introduction

Arthrofibrosis, one of the most common complications following total knee arthroplasty (TKA), is a debilitating condition that arises due to excessive scar tissue formation in a joint following injury or trauma [1,2,3,4]. With an occurrence of 1.3–19.8% [5, 6], the inconsistency regarding the definition of arthrofibrosis, especially with respect to flexion and/or extension loss, likely explains the wide range in the reported incidence [3, 5, 7,8,9]. In the USA, arthrofibrosis and stiffness account for 28% of 90-day hospital readmissions following TKA and up to 10% of revisions within years [10, 11].

Risk factors for arthrofibrosis and stiffness following TKA include genetics, low pre-operative range of motion (ROM), complexity of surgery, history of prior surgery, immobilization, and poor motivation during the rehabilitation process [12]. In general, treatment for arthrofibrosis and stiffness is initiated with intensive physical therapy, with or without bracing, and complemented with a multimodal pain regimen until 12 weeks post-operatively [12, 13]. If stiffness persists, manipulation under anesthesia (MUA) is performed between eight and 12 weeks post-op, following which, arthroscopic lysis of adhesions (aLOA) is most commonly performed [8, 13]. Failure of these interventions may warrant revision TKA (rTKA), depending on the patients’ functional requirements. Nevertheless, current literature suggests that the improvements in ROM, pain, and function are only favourable for a select few rTKA patients [14,15,16]. The challenges of treating stiffness and arthrofibrosis have spurred investigation into novel therapies and peri-operative protocols such as low-dose irradiation as a supplement to rTKA [17], as well as the use of continuous passive motion following primary TKA [18]. In the current era of value-based healthcare, there is a growing need to elucidate the most effective and cost-efficient treatment that optimizes clinical outcomes for patients.

The purpose of this systematic review was to compare the functional outcomes of MUA, aLOA, and rTKA for the treatment of arthrofibrosis and stiffness after primary TKA.

Materials and methods

Search methodology

Two databases from the US National Library of Medicine, PubMed and Ovid MEDLINE, were queried in accordance with PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) statement guidelines for all articles that were published prior to October 30, 2020. Four different search terms were utilized to screen for articles: ((total knee arthroplasty) AND (stiffness) OR (arthrofibrosis)), (manipulation under anaesthesia) AND (total knee arthroplasty), and (lysis of adhesions) OR (arthrolysis) and (total knee arthroplasty). Reference lists from each study and related citations from each search engine were explored for additional eligible studies to ensure that no studies were missed after the initial database query. Studies were then screened by two independent reviewers (AH, GG) for eligibility based on the title and abstract. The full text of studies that were deemed to meet the inclusion criteria were reviewed, after which only those that were eligible were included in the analysis. Disagreements were resolved by consensus, and by a third reviewer, if needed.

Inclusion and exclusion criteria

Articles were included if they involved the treatment of arthrofibrosis and/or stiffness following primary TKA and included patient-reported outcome measures (PROMs), ROM values, and the prevalence of treatment success/failure. The rate of treatment success/failure corresponded to the percentage of patients who experienced an increase/decrease in ROM that exceeded a threshold defined in each individual study, as well as the percentage of patients who pursued further treatment for arthrofibrosis. rTKA was defined to include the removal of at least one component from the index procedure (e.g., isolated polyethylene tibial insert exchange (IPTIE)). Only the most recent publication was included if a study published several interim results. Studies were excluded if they did not report the abovementioned outcome measures or clearly define the group the outcomes belonged to. Case reports, reviews, commentary pieces, and studies that were not available in English were also excluded.

Data extraction

All information and outcomes of interest, including author, date of publication, study design, number of patients, treatment modalities, length of follow-up, and PROMs, were recorded on a study-specific data extraction sheet. The primary outcome was PROMs after surgery with secondary outcomes including improvement in ROM and prevalence of treatment success as defined by the article.

Quality assessment

The level of evidence was determined using the Oxford Center for Evidence-Based Medicine Levels of Evidence. Assessment of quality for the included studies was performed by two of the authors (AH, GG) and scored in accordance with the methodological index for non-randomized studies (MINORS) instrument, a valid and reliable tool for study quality assessment [19]. The MINORS instrument is made of 8 and 12 items for non-comparative studies and comparative studies, respectively, with item scores of 0 (not reported), 1 (reported but inadequate), and 2 (reported and adequate). Non-comparative studies and comparative studies can achieve maximum scores of 16 and 24, respectively.

Results

In total, 2,035 articles were initially identified in the database review, and an additional 220 articles were found in relevant systematic reviews and meta-analyses (Fig. 1). Forty articles were included in this study after screening full texts, 16 with a level of evidence (LOE) of III and 24 with a LOE of IV. No level I or II studies were found in the literature. Two of the 40 studies were considered to be of moderate quality, while the 38 other studies were considered to be of poor quality (Table 1).

Fig. 1
figure 1

Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram outlining the study selection process

Full size image
Table 1 Included studies and methodological index for non-randomized studies (MINORS) for included studies
Full size table

Revision total knee arthroplasty

In the 21 studies, the mean time to surgery was 1.81 (95% CI (1.29–2.34)) years, and the average follow-up was 3.86 (95% CI (3.41–4.32)) years. Between 15 and 73% of patients underwent some form of treatment before rTKA, although this was only reported in seven studies [3, 15, 16, 31, 40, 48, 49]. Demographics, definitions of stiffness used, and indication of treatment are summarized in Table 2.

Table 2 Demographics, range of motion information, and patient-reported outcome measures following revision total knee arthroplasty
Full size table

rTKA improved arc ROM in all the included studies, with average ROM increasing from 54.60° (95% CI (46.85–62.34°)) to 82.92° (95% CI (78.21–87.62°))—an increase of 28.32°. However, the mean or median post-op arc ROM remained below 90° in 14 of 20 studies [3, 16, 20, 24, 27, 30, 36, 39, 47,48,49,50,51, 54]. Three of the four studies that reported pre-operative and post-operative Knee Society [56] Pain (KSSP) scores showed improvement after surgery [16, 39, 49]. All studies that reported KSS clinical scores (KSSC) or KSS functional scores (KSSF) showed improvement over time but only one reported excellent KSSF, defined as a score > 70 [56].

Two publications compared the outcomes of IPTIE to full rTKA [20, 47] (Table 3). Patients who underwent IPTIE had greater post-operative ROM, flexion, and clinical outcomes than those who underwent complete revision [47]. Another study found that in patients with idiopathic arthrofibrosis, IPTIE leads to greater improvement in ROM and flexion with lower revision rates compared to complete revision. Idiopathic arthrofibrosis was defined as a painful limitation in knee ROM without an underlying etiology. For patients with non-idiopathic arthrofibrosis, complete revision was a better option as patients had lower pain but equivalent values of post-operative ROM compared to those who underwent IPTIE. However, no other studies distinguished between rTKA performed for idiopathic and non-idiopathic causes of stiffness.

Table 3 Sub-analysis of isolated polyethylene tibial insert exchange vs. complete revision TKA
Full size table

Two studies [26, 27] compared the outcomes of patients who underwent rTKA using either a rotating-hinge (RH) device or a traditional constrained condylar device (CC) (Table 4). rTKA patients with a RH device experienced a greater increase in arc ROM. While one study [27] showed that patients with a RH rTKA fared better than those with non-RH TKA in terms of KSS, the report by Bingham et al. [26] contrasted these findings, suggesting that non-RH patients fared better in all outcome measures.

Table 4 Sub-analysis of rotating hinge vs. constrained condylar revision TKA for stiffness
Full size table

In the twelve rTKA studies that reported the percentage of patients who experienced an improvement in ROM, four studies [15, 26, 50, 54] found that more than 10% of patients did not experience an improvement in ROM, while eight studies reported that 0–6.5% of patients did not experience improvement (Supplementary Table 1) [5, 31, 36, 39, 40, 48, 49, 51]. Furthermore, 14 of 18 studies reporting the percentage of patients who pursued further treatment for stiffness showed that at least 10% of patients required additional treatment.

Arthroscopic lysis of adhesions

Seven studies that evaluated aLOA for the treatment of stiffness and arthrofibrosis following TKA were included (Table 5). Patients were treated at an average of 1.03 (95% CI (0.77–1.30)) years after primary TKA and were evaluated at an average of 2.26 (95% CI (1.57–2.95)) years after aLOA. The percentage of patients who had failed prior treatment for stiffness was reported in six studies, ranging from 27.3 to 100%, in three studies [38, 45, 55]. All studies reported an increase in arc ROM from an average pre-operative value of 61.97° (95% CI (57.32–66.61°)) to a post-operative average of 101.78° (95% CI (86.97–116.58°)). Only one study reported Oxford Knee Scores [42]. Of the studies that reported pre-operative and post-operative KSS, all demonstrated improvements from baseline [38, 45, 46, 52].

Table 5 Demographics, range of motion, and patient-reported outcome measures following arthroscopic lysis of adhesions
Full size table

Six studies reported values for the percentage of patients who did not improve with respect to ROM (Supplementary Table 2) [5, 38, 42, 46, 52, 55]. Four of 6 studies reported that all patients had an increase from their pre-operative arc ROM [5, 38, 52, 55]. Two studies reported that 4.5% and 5.6% of patients did not have an improvement in arc ROM [42, 46]. Four studies analyzed the percentage of patients who pursued further treatment for stiffness, with three of those reporting 0% and one reporting 25% (2 of 8 patients) [5, 40, 47, 54].

Manipulation under anaesthesia

Fourteen studies evaluated the outcomes of MUA. The average time to surgery was 0.24 (95% CI (0.14–0.33)) years with an average follow-up of 4.70 (95% CI (3.12–6.28)) years. The average pre-operative ROM was 78.69° (95% CI (64.14–89.23°)), and this increased to an average of 99.65° (95% CI (91.79–107.50°))—an increase of 20.97°. In six of the ten studies that reported post-operative ROM, the average value was greater than 100° [5, 22, 25, 34, 35, 43], and all studies that recorded the post-operative arc ROM showed means or medians of at least 80° (Table 6). Three studies reported both pre-operative and post-operative KSSP, of which, all three showed improvement [25, 37, 44]. In addition, all studies that reported pre-operative and post-operative KSSC and KSSF showed improvement at final follow-up [22, 25, 35, 37].

Table 6 Demographics, range of motion, and patient-reported outcome measures following manipulation under anesthesia
Full size table

Only one study compared patients who underwent MUA with or without a glucocorticoid adjunct (Table 7) [43]. Sub-analysis showed that patients who received this adjunct better retained the gains in ROM at 2-year follow-up despite achieving equivalent ROM during MUA.

Table 7 Steroid adjunct vs. non-steroid adjunct cohort sub-analysis
Full size table

Four studies reported the percentage of patients who did not experience an improvement in arc ROM (Supplementary Table 3) [5, 21, 34, 43], three [21, 34, 43] of which reported values less than 10%, with one study [5] reporting 15.2%. Five of seven studies found that less than 10% of patients needed to seek additional treatment for stiffness [33, 35, 37, 43, 44], with two studies reporting that at least 15% of patients had to undergo further treatment [21, 25].

Discussion

Stiffness following TKA is a challenging complication to treat. Multiple definitions of stiffness have been proposed, and the indications for treatment are not uniform [12]. This problem is further compounded by patients who perceive a limitation in knee ROM and are dissatisfied with their knee replacement but who nonetheless do not fit the established criteria for stiffness.

Our findings suggest that rTKA for stiffness and arthrofibrosis, even in the absence of component malposition, leads to inferior clinical outcomes and range of motion compared to MUA and aLOA. A greater percentage of patients who underwent rTKA needed to undergo further treatment for stiffness and arthrofibrosis, with figures reaching up to 42.9% [50]. While the need for additional treatment was greatest for rTKA patients, three of 4 aLOA studies reported a 0% incidence of patients requiring further treatment, while another reported an incidence of 25% [38]. The need for further treatment was also lower in MUA, with most studies reporting values less than 10% [25, 33, 35, 37, 43, 44]. Nonetheless, the reduced incidence of secondary surgery after MUA or aLOA may not indicate treatment success. It could instead possibly suggest a sense of realistic futility by the patient and/or the care team with the prospects of rTKA.

Analysis of KSSP showed that the three different procedures provided similar levels of pain relief following treatment, despite lower overall post-operative KSSC and KSSF after rTKA compared to both MUA and aLOA. While the majority of rTKA studies reported KSSC less than 70, the lowest reported KSSC was 70.7 for MUA and 77 for aLOA. This finding could be attributed to the nature of the procedures—MUA and aLOA preserve the index arthroplasty and bone stock, while rTKA is associated with bone loss during implant removal, precluding optimal fixation of implants [57]. Furthermore, during rTKA for stiffness, constrained condylar (CC) prostheses are often used due to the difficulty in achieving soft tissue balance [58]. As multiple studies having shown that the outcomes of rTKA are inferior to those of primary TKA [59,60,61,62,63], these factors could explain, in part, the more limited knee functionality among patients who undergo rTKA. Nonetheless, the poorer outcomes after rTKA could be indicative of a higher incidence of component malposition or failure of prior procedures such as MUA and aLOA. In other words, it is unclear how many patients in the rTKA had failed prior interventions and thus represented a more intrinsically complicated group.

Patients who underwent aLOA had greater improvement in total arc ROM from baseline compared to patients who underwent rTKA. With the exception of Campbell et al. [55], all the included studies on aLOA reported an increase of least 29° in ROM, whereas patients who underwent MUA had an increase in mean ROM of 21°. While rTKA significantly increased ROM by at least 40° in some studies [5, 27, 40, 49, 51], others reported only a modest increase of 4° at follow-up [24]. Improvement in mean ROM across all rTKA studies was 28°, and a subgroup analysis of 5 studies [20, 27, 39, 50, 51] that specified the indication for revision showed an increase in mean ROM of 27° in rTKAs performed for idiopathic arthrofibrosis alone. It is important to note that despite the higher pre-operative ROM in MUA compared to rTKA patients (79° vs. 55°), MUA patients had a similar, albeit slightly smaller increase in mean ROM (21°) compared to patients who underwent rTKA for idiopathic arthrofibrosis (27°) [20, 27, 39, 50, 51] (Table 8).

Table 8 Average follow-up, time to surgery, and range of motion
Full size table

It has been reported that idiopathic stiffness rTKA patients have worse outcomes than non-idiopathic stiffness rTKA patients [22]. Although the distinction between idiopathic and non-idiopathic stiffness was not made in most studies, one study by Xiong et al. suggested that IPTIE may be more efficacious for these patients compared to complete rTKA, as IPTIE patients reported greater improvement in ROM and flexion with lower rates of revision [22]. However, approximately 17% of IPTIE patients with idiopathic arthrofibrosis still required further treatment, higher than reported figures for MUA and aLOA.

In rTKA, implant choice may also influence treatment outcomes. RH prostheses do not rely on ligamentous or soft tissue balancing for stability, allowing for the release of collateral ligaments and posterior capsule as well as elevation of the joint-line [26]. Patients who undergo rTKA with a RH prosthesis may experience less pain, better knee function, and greater improvement in extension and flexion compared to those who underwent rTKA with CC prostheses [26, 27]. Despite these benefits, patients who undergo rTKA with a RH device are at greater risk of implant failure and re-revision; hence, the trade-offs between ROM and implant durability should be evaluated and tailored to the needs of the patient [26].

Few of the included studies investigated the impact of timing from the index TKA on the outcomes of rTKA or aLOA. Mont et al. reported that patients who underwent aLOA at less than one year post-op had higher gains in ROM than those who underwent aLOA at greater than 1 year post-op [46]. One study found no difference in flexion, extension, and KSS gains in patients who underwent rTKA before (n=22) or after two years post-op (n=24) [33]. In contrast, Donaldson et al. reported that patients who underwent rTKA before two years post-op had superior improvement in ROM, and outcomes were reported [3]. However, the authors suggested that these results be interpreted with caution due to the small sample size and heterogeneity of the study population [3]. Nonetheless, further studies evaluating the effect of timing on the outcomes of rTKA for stiffness are warranted.

The timing of MUA is another important consideration that has been shown to influence the outcome of treatment. MUAs performed later in the post-operative period need to overcome a greater degree of fibrotic [5]. Several studies have shown improved function and ROM in MUAs performed before 12 weeks post-op compared to those performed after 12 weeks post-op [32, 64]. Further, of the included studies in this systematic review, only one compared the outcomes of MUA with and without a glucocorticoid adjunct. Despite achieving similar ROM during manipulation, MUA patients who received a glucocorticoid adjunct better retained the gains in ROM and had higher final ROM at two years (111° vs. 92°) [43]. Further studies are needed to confirm these clinical benefits in a larger cohort and explore the mechanisms behind the synergistic effect of steroid treatment.

The advent of value-based models has increased the pressure on healthcare systems to provide high-quality and cost-efficient care. As of 2018, the inpatient National Average Medicare CMS payment for performing rTKA was $17,115.36, while the average outpatient costs for MUA and aLOA were $1209 and $2041, respectively. Future cost-effectiveness studies should be performed to directly compare the three options.

There were several limitations of our study. This systematic review only contained studies with a LOE of III and IV, likely resulting in cohorts that were unmatched in terms of demographics, aetiologies, indications, and other confounding variables such as time from TKA, prior surgeries, procedural techniques, and peri-operative protocols. Furthermore, patients in the three cohorts varied with respect to pre-operative KSSF and KSSC (Table 9). We also could not include studies that evaluated the efficacy of physical therapy as none of these studies analyzed PROMs, which was our primary endpoint. Finally, it is unclear whether patients in the rTKA group were similar to those in the other cohorts, as patients who underwent rTKA presented with significantly lower pre-operative ROM and at a significantly longer time following the index surgery. The extent of component malalignment, inaccurate implant sizing, or instability is poorly reported in the literature, and thus, their incidence in each group also remains unknown. Hence, it is uncertain how these factors could have influenced the results of each treatment modality. Furthermore, given the vagueries in the literature, our study could not elucidate potential differences for either MUA or rTKA done for idiopathic arthrofibrosis alone or with implant malposition. Nevertheless, we believe that the findings of this study still provide a valuable synthesis of the literature to guide clinical practice.

Table 9 Preoperative patient-reported outcome measures
Full size table

Conclusion

Manipulation under anesthesia (MUA) and arthroscopic lysis of adhesions (aLOA) patients had superior clinical outcomes, knee functionality, and range of motion after intervention compared to patients who underwent revision total knee arthroplasty (rTKA). As this was a systematic review of studies with LOE III and IV, additional prospective randomized studies are warranted to elucidate any differences among these treatment modalities. Future studies should also compare the cost-effectiveness of these treatment options, especially given the greater cost and unpredictable results of rTKA.