In the present study, we utilized arthroscopy in managing PI, GS, and AF after TKA. The PJI rate was 4.1%, and the non-infectious complication rate was 1.4%. Most patients acquired remarkable improvements in ROM, KSS knee score and KSS function score, with a symptom recurrence rate of 23.0% and prosthesis revision rate of 14.9%. There were quite a few similar retrospective case series studies exploring the therapeutic efficacy of arthroscopy after arthroplasty whose main findings are summarized in Table 3. As shown in the table, the incidence of PJI ranged from 0%-3.7%. Most studies with similar patient indications but a smaller sample size and shorter duration of FU reported that there were no post-arthroscopy infections[12-16]. We reported a higher infection rate of 4.1%. There were several hypotheses for the increased risk of infection as proposed by Werner et al., including violation of the joint capsule, persistent post-arthroscopy synovitis, bacterial seeding of the joint during KA, and postoperative portal tracts as possible mechanisms for subsequent infection[26]. Lovro et al. studied 192 TKA patients who underwent subsequent knee arthroscopy using the information from the Medicare database with an FU of 5.45 years[17]. The incidence of revision for infection was 6.3%. The number (4.1%) in the current research was slightly lower than that (6.3%) in Lovro’s study. As with all database studies, Lovro’s research relied on ICD-9 and CPT codes, which were subject to inaccuracies. The authors used the code “infection”, which may possibly identify cases with infections other than PJI, which could potentially make the result falsely higher. Lovro et al. reported that the infection rate for post-arthroplasty patients without an arthroscopy was 2.1%. It was also higher than the majority of the results in other studies of the same era[21, 22]. For non-infectious complications, Heaven et al. reported in an SR that, across all 609 patients, only 1 intraoperative complication occurred —one patient had the arthroscopic instrumentation break in the knee, which was retrieved uneventfully. The overall complication rate was 0.5%, excluding those patients who received diagnostic and/or therapeutic arthroscopy for periprosthetic infection. We added one postoperative complication, and we warned that aggravation of comorbidities and life-threatening complications might occur, even after minimally invasive arthroscopic procedures.
For efficacy, Klinger et al. reported in a case series with similar patient indications that the average KSS knee score increased from 71 before arthroscopy to 85 postoperatively with an FU of 34 months, while the KSS function score increased from 69 to 83 patients[10]. In the current case series, patients gained comparable improvements in both KSS knee and function scores. For symptom recurrence, Heaven et al. reported that in an SR with approximately 2 years FU on average, of the 488 patients who underwent arthroscopy for therapeutic (not diagnostic) purposes, 85 (17.4%) patients experienced symptom recurrence after surgery[4]. In the current study, we presented a slightly higher recurrence rate of 23.0%. We think that the longer FU could potentially be the reason for this increased rate. In the SR, a more detailed statistical analysis of the pooled data was not possible because of different outcome measures used in different patient subgroups. In the current study, we found that patients with PI had the best results, in terms of lower recurrence and revision rates, while patients with GS had the highest risks of failure and prosthesis revision. Regarding the revision rate, in the current study, it was 14.9%. In the SR by Heaven et al., 19.7% of patients went on to require further operation, possibly including prosthesis revision and open arthrolysis[4]. The possible reason for the gap is that we did not include cases in which arthroscopy was applied for diagnostic purposes, whose intraoperative findings during arthroscopic evaluation may be tibial loosening, femoral prosthesis debonding, and PJI. Revision surgeries or open debridement are often needed for these cases.
Some earlier studies involved patients with other indications, for example, PE wear, apparent PF maltracking, PJI, etc. In the current study, these patients were not studied and analysed, as severe PE wear was occurring less frequently[24] and a consensus had been reached regarding the standard management of PJI and apparent patella malalignment, for which prosthesis revision is the gold standard of treatment. We also did not include patients who underwent arthroscopy for diagnostic purposes, as there was very high heterogeneity among the subgroups of these patients. Additionally, the sample size for different subgroups was rather low (less than ten), which would undermine the statistical power[25]. From the clinical perspective, the application of arthroscopy as a diagnostic procedure should also be avoided in most cases, as it carries certain complication risks, and additionally, in 8 out of 19 patients who underwent arthroscopy for diagnostic purposes, no pathology could be identified. Meanwhile, recent literature has proposed new imaging tests (i.e., magnetic resonance (MR) imaging[26], MR angiography[27], synovial fluid cell phenotype analysis[28], and SPECT/CT[29]) to serve as appropriate diagnostic tools. These analyses were not utilized in the present study. However, they would be able to facilitate identifying the major aetiology and prevent unnecessary arthroscopic examinations.
There are some specific considerations on various patient groups based on our experience from this patient series. PI was not a rare complication[12] after TKA, especially in early PS design[30] [31]. Patients with PI usually presented with anterior knee pain (AKP), and additionally, crepitus could be found in history taking and/or physical exams, which could facilitate diagnosis. CT should be done to rule out obvious component malrotations, which could also lead to AKP. There were two types of PI, with the most common type being PF impingement. In this type, the hypertrophied cyclops and synovium at the backside of the quadriceps (shown in Fig. 2a) tendon have contact with the distal edge of the trochlea, creating mechanical irritation during knee extension. A typical case of PF impingement was presented in Fig. 3. Another type was tethered patella syndrome. Adhesion and fibrous bands could be found from the inferior pole of the patellar component to the intercondylar notch, tethering the patella inferiorly[15]. Intraoperatively, surgeons need to routinely check for the existence of hypertrophic fibrous bands, synovium and potential impingement around the patella. Mild patellar maltracking might occur simultaneously due to blocking of the cyclops or the tethering effect from the fibrous bands, and normal patellar motion has to be checked in the end. The conservative treatment often is unsatisfactory[30]. Lucas et al.[12] and Koh et al.[7] reported that most individuals with isolated PF impingement could expect an excellent result from arthroscopic resection, and the symptom recurrence rate was often low, ranging from 0%-3.3%[7, 12, 15]. These results were consistent with the present study.
Patients with AF usually present with a stiff knee. In the present study, we define a stiff TKA as flexion under 90° and flexion contracture over 10°. Arthroscopy allows surgeons to release adhesions in a controlled manner. Williams et al.[15] and Mont et al.[16] reported an average improvement in ROM of approximately 30° after arthroscopic release[8, 13]. Our results were similar, with an improvement in ROM of 26.4°, and it should be noted that the improvements in the current study were generated from not only arthroscopic arthrolysis but also manipulation under anaesthesia (MUA) and intensive postoperative rehabilitation. Even though detectable clinical or radiographic abnormalities had already been ruled out, seven of twenty five patients encountered a recurrence in the limitation of ROM; three of them underwent a prosthesis revision. Frustratingly, we were unable to figure out risk factors for treatment failure due to a relatively small sample size; thus, the therapeutic efficacy for AF is less predictable. According to the literature, arthroscopy might be helpful when the stiffness is caused by a tight posterior cruciate ligament (PCL)[13] or adhesions in the superior pouch or the medial and lateral gutter due to poor rehabilitation[13, 32]. In contrast, an extension deficit caused by a tight posterior capsule is not an ideal indication. We suggest a delicate analysis of potential aetiologies. If apparent surgical errors are present, arthroscopy should be avoided.
In the current series, patients with GS after TKA usually presented with recurrent swelling and/or hemarthrosis. According to the literature, the aetiology is also multifactorial. Systemic factors include anticoagulant use or presence of a bleeding disorder. Local factors include trauma, inflamed synovium or vascular anomaly (e.g., arteriovenous malformation) and injury. Cases caused by iatrogenic vascular injury usually present with swelling within 6 months postoperatively[33]. There were no such cases in this study. A more common mechanism and pathologic entity originates from the entrapment of hypertrophic synovium (shown in Fig. 2e). Varied conditions could cause synovium inflammation and hypertrophy (shown in Fig.2f) (i.e., malaligned implants could lead to asymmetric PE wear and particle generation, causing synovial proliferation, subsequent impingement and bleeding). This is a more chronic process, usually occurring more than one year after TKA[33]. In the current study, all patients with GS belonged to this pathologic entity, and in four patients, signs of PE wear were detected. Conservative management consists of immobilization, cryotherapy, cessation of anticoagulants, and rest (and/or aspiration). However, only 30% of patients had resolution[23]. Angiography and selective embolization offers another choice, with the advantages of low infection risk, ability to be performed under local anaesthesia, and quick rehabilitation postoperatively[34]. It was reported to be effective in more than 90% of cases[35]. However, repeat embolization may be necessary[36]. In patients with contraindications to angiography, arthroscopic synovectomy could be attempted, but its success is less predictable[6, 23, 37, 38]. Ohdera et al.[6] and Kindsfater et al.[23] reported a synovitis recurrence rate of 67% and 50%, respectively. Additionally, the pathologic site was not always identified arthroscopically. These failure rates and intraoperative findings were in line with those in the current study. Additionally, the angiographic evidence of contrast “blush”, indicating the pathologic site, is arthroscopically inaccessible sometimes (i.e., at the posterior capsule)[34]. In such circumstances, an open synovectomy is indicated, with a reported resolution rate of more than 90%[23, 39]. However, infection risk, wound complications, and prolonged rehabilitation would be clinical concerns. Revision is needed when obvious clinical or radiographic abnormalities are present. According to the results of the present study and existing literature, arthroscopic synovectomy should not be the first-line treatment for GS.
Several limitations should be acknowledged when drawing conclusions. The most important limitations include the retrospective observational study design and the absence of a control group, which can lead to potential recall bias. The absence of a control group probably results in an overestimation of the specific treatment effects, as the contextual effects contributing to the overall treatment effect cannot be determined[40], so the primary outcome in this study was the feasibility but not the therapeutic efficacy of arthroscopy after TKA. Second, the study design could also induce selection bias. Third, the sample size in the PI and GS groups did not reach the number derived from the sample size calculations due to the particular patient population presenting to a single surgeon in a single centre. The lost to follow-up rate was 19.6%, which poses a potential threat to the validity of the study[41]. However, the sample size in the current study was relatively large and the follow-up duration was comparatively long. In the current study, we provided information and evidence on the safety and therapeutic efficacy of arthroscopic debridement to manage PI, AF and GS after TKA.