Articulating Knee Spacers in the Treatment of Periprosthetic Joint Infection: All Polyethylene Tibia or Tibial Insert?

Abstract

Background:

The best antibiotic spacer for periprosthetic knee joint infection treatment is unknown. Using a metal-on-polyethylene (MoP) component provides a functional knee and may avoid a second surgery. Our study investigated complication rates, treatment efficacies, durabilities, and costs of MoP articulating spacer constructs using either an all-polyethylene tibia (APT) or a polyethylene insert (PI). We hypothesized that while the PI would cost less, the APT spacer would have lower complication rates and higher efficacies and durabilities.

Methods:

A retrospective review evaluated 126 consecutive articulating knee spacer (sixty-four APT and sixty-two PI) cases from 2016 to 2020 was performed. Demographic information, spacer components, complication rates, infection recurrence, spacer longevity, and implant costs were analyzed. Complications were classified as: spacer-related; antibiotic-related; infection recurrence; or medical. Spacer longevity was measured for patients who underwent reimplantation and for those who had a retained spacer.

Results:

There were no significant differences in overall complications (p<0.48), spacer-related complications (P=1.0), infection recurrences (P=1.0), antibiotic-related complications (p<0.24), or medical complications (p<0.41). Average time to reimplantation was 19.1 weeks (4.3 to 98.3 weeks) for APT spacers and 14.4 weeks (6.7 to 39.7 weeks) for PI spacers (P=0.09). There were 31% (20 of 64) of APT spacers and 30% (19 of 62) of PI spacers that remained intact for an average duration of 26.2 (2.3 to 76.1) and 17.1 weeks (1.7 to 54.7) (P=0.25), respectively, for patients who lived for the duration of the study. PI spacers cost less than APT ($1,474.19 vs. $2,330.47, respectively; p<0.0001).

Conclusion:

APT and PI tibial components have similar results regarding complication profiles and infection recurrence. Both may be durable if spacer retention is elected, with PI constructs being less expensive.

Introduction

The incidence of prosthetic joint infection (PJI) following total knee arthroplasty (TKA) is between 1 and 2% and represents the most common cause of revision TKA at 25% [1,2]. Revision surgeries present a burden to hospitals, projected to cost United States (U.S.) health systems approximately one billion dollars in 2020 [3]. As the number of revision TKA continues to grow substantially [2,4], examination and improvement in the procedures for treating PJIs in TKA patients is imperative.

The “gold standard” treatment in the U.S. for PJI in TKA is a two-stage revision surgery [5]. The first stage is removal of infected TKA components, debridement of periarticular tissues, and placement of a drug-eluting antibiotic spacer followed by a prolonged course of antibiotics. After infection eradication, a second stage reimplantation of TKA components occurs [6]. Success rates of this technique has been reported at 82 to 91% at mean six years follow up [5,6]. This two-stage technique was first described by Insall et al in 1983 as hardware removal and debridement, followed by antibiotics, and eventual TKA reimplantation after infection eradication [7,13]. This technique was subsequently modified to include interim spacer implantation [7].

Antibiotic spacer options are categorized as either static or articulating. Static spacers are made by placing a block of antibiotic-laden cement into the joint space, but do not allow for knee range of motion (ROM) and therefore, sacrifice patient function during the antibiotic treatment phase [6]. Articulating spacers consist of TKA implants secured in place with high-dose antibiotic cement that allows local antibiotic elution while still allowing knee motion. Studies show that static and articulating spacers are equally successful at achieving infection eradication [7]. Static spacers are favored by those desiring the increased knee stability caused by the immobilization of the surrounding soft tissues [7]. However, articulating spacers have shown greater post-operative ROM compared to static spacers, although there are mixed results as to whether this causes differing functional results [7,16,17]. Articulating spacers may be constructed with cement molds, prefabricated antibiotic-laden surfaces, or actual knee arthroplasty parts. Articulating spacers allow for better function and may be a better solution over two to three years than second stage reimplantation surgery [8,9,10]. They are generally made with a polyethylene tibial component, allowing for easier removal at second-stage surgery. The tibial component may be an all-polyethylene tibial component (APT) or a tibial polyethylene insert (PI) cemented into place without the metal baseplate.

The goal of our study was to compare complication rates, treatment efficacies, in-situ durabilities, and implant costs of articulating spacer constructs using either an APT or PI component with a standard metal femoral implant. We hypothesized that while the PI would have lower costs, the APT spacer would have lower complication rates and higher efficacies and durabilities.

Methods:

After institutional review board approval, we conducted a retrospective review to identify patients who underwent placement of an articulating spacer for PJI treatment after TKA at a single tertiary care academic center from September 1, 2016 to August 31, 2020. Inclusion criteria were placement of an articulating knee spacer for infection and an age greater than 18 years. Exclusion criteria was placement of the spacer for native knee infection. Data was collected from the electronic medical record on patient demographic information, spacer constructs, microorganism profile, infection eradication, complication details, spacer longevity, and spacer cost.

We identified 136 patients who underwent placement of 144 articulating antibiotic knee spacers during the study period. Of those, 18 patients were excluded for placement of spacers for a native knee infection. The final study population included 118 patients who had 126 articulating antibiotic knee spacers placed for infected TKA, with 64 spacers in the APT group and 62 in the PI group (Figure 1). All patients were operated on by a group of four surgeons who were all fellowship-trained in knee arthroplasty. No statistically significant differences were evident between the groups regarding age, sex, body mass index (BMI), or the presence of diabetes mellitus (Table 1).

Figure 1:

Flow diagram for articulating antibiotic knee spacer case selection

Table 1:

Comparison of demographics for the Polyethylene Insert group versus the All-Polyethylene Tibia group.

Demographic	All Spacers (n=126)	Polyethylene Insert (n=62)	All-polyethylene Tibia (n=64)	P-value
Mean age (range)	67 (32 to 96)	67 (32 to 87)	67 (43 to 96)	0.9864
Sex
Men	63 (50%)	32 (51.6%)	31 (48.4%)	0.8587
Women	63 (50%)	30 (48.4%)	33 (51.6%)
Diabetes
Yes	44 (34.9%)	17 (27.4%)	27 (42.2%)	0.0948
No	82 (65.1%)	45 (72.6%)	37 (57.8%)
Mean BMI (range)	32.72 (19.3 to 68.7)	33.35 (19.3 to 59.7)	32.12 (20.8 to 68.7)	0.4095
Side
Right	66 (52.4%)	33 (53.2%)	33 (51.6%)	0.8605
Left	60 (47.6%)	29 (46.8%)	31 (48.4%)

BMI = body mass index

All infected TKAs underwent implant removal with thorough irrigation and debridement according to established best-practice protocols [11]. All patients completed six weeks of broad-spectrum or organism-specific intravenous antibiotics after the explant and spacer placement. Antibiotic choice was based on culture results and management was directed by our musculoskeletal infectious disease service. The most common infecting organism based on cultures obtained at the time of spacer placement was methicillin-sensitive Staphylococcus aureus (MSSA) (20.6%, 26 of 126). Other isolated organisms included methicillin-resistant Staphylococcus epidermidis (MRSE) (11.1%, 14 of 126), methicillin-resistant Staphylococcus aureus (MRSA) (10.3%, 13/126), and methicillin-sensitive Staphylococcus epidermidis (MSSE) (9.5%, 12 of 126). Cultures were negative in 18.3% (23 of 126) of knees and polymicrobial in 11.1% (14 of 126).

Articulating antibiotic spacers were fabricated from metal femoral components and either APT or PI components. Two batches of plain Cobalt ^® bone cement (DJO Surgical, Austin, Texas) were used intraoperatively to provide fixation for all spacers with the manual addition of 2 to 3 grams of vancomycin and 2.4 to 3.6 grams of tobramycin per batch based on surgeon discretion. Standard cementing technique was used by all surgeons. Adjunctive fixation was based on surgeon preference and included Steinmann pins or screws for improved stability of the components and/or placement of antibiotic cement dowels in the femoral and tibial intramedullary canals [12]. When utilizing the PI, the surgeon would roughen the back of the liner using a saw or burr to create grooves for cement interdigitation. Antibiotic cement was used to fabricate tibial and femoral augments as indicated. Augments were added to help balance the flexion and extension gaps found in the knee during trialing. Examples of the two spacer constructs are shown in Figures 2 and 3. APT spacers used included Genesis II (Smith & Nephew, Memphis, Tennessee; n=31) or Triathlon ^® (Stryker, Kalamazoo, Michigan; n=33) components; PI spacers included Empowr 3D Knee^™ (DJO Surgical, Austin, Texas; n=25), Evolution ^® Medial Pivot (Microport, Irvine, California; n=36), or Optetrak Logic ^® (Exactech, Gainesville, Florida; n=1) components.

Figure 2:

Radiograph of all-poly tibia (APT) spacer placement

Figure 3:

Radiograph of polyethylene insert (PI) spacer placement

Postoperatively, patients were allowed to weight bear and perform range of motion exercises as tolerated. Patients were evaluated in clinic at two weeks to assess wound healing. Patients were assessed again at six-to-eight weeks after spacer placement to assess candidacy for the timing of reimplantation surgery based on infection control and functional status of the spacer. Of the four surgeons, some were more comfortable leaving the spacer if tolerated by the patient rather than proceeding directly to the second-stage surgery, leading to different reimplantation rates among the surgeons. However, there were no differences between rates of reimplantation each surgeon had for APT vs. PI components (Table 2). Complications were reviewed during the lifetime of the spacer and were classified as spacer-related (e.g., dislocation, periprosthetic fracture, component loosening), antibiotic-related (e.g., adverse medication reaction, intravenous catheter issues), infection recurrence/persistence, medical complications, emergency room (ER) visits, and hospital readmissions. Comparisons of complications and infection eradication rates were made using Fisher’s exact tests.

Table 2:

Surgeon rates for spacers and reimplantation

	All-polyethylene tibia		Polyethylene insert
Surgeon	cases (n=64)	cases not reimplanted (%)	cases (n=62)	cases not reimplanted (%)	P-value
A	34	6 (18)	0	0	n/a
B	20	12 (60)	9	8 (89)	0.20
C	10	2 (20)	26	6 (23)	1.00
D	0	0	27	5 (19)	n/a

Spacer lifetime was calculated as the difference between date of reimplantation and date of spacer placement. For patients who had not yet undergone, or declined, reimplantation, spacer longevity was calculated as the difference between the date of last follow up and the date of spacer placement and grouped further based on patients who survived the length of the study and those who died due to other causes during the study’s timeframe. Data was analyzed using Welch’s t- and Student’s t-tests and analyzed to ascertain if the APT spacer had a longer time before reimplantation or a longer follow-up time for patients who have not received the second-stage surgery. Given the sample size and variance of data obtained during the study, as well as a significance level of 0.05, there was 99.7% power to detect a 3-month (91 day) difference between those who had the second stage and a 36.3% power to detect a 3-month difference for those who did not. Spacer cost was defined as the cost of the implants used at our hospital system, including the femoral component, tibial component, and any pins or screws used.

Results:

No significant differences were observed in spacer-related complication rates between the APT (10.9%, 7 of 64) and the PI group (9.7%, 6 of 62, P=1.0). Spacer-related complications for the APT group included spacer dislocations/instability (n=2), extensor mechanism disruption (n=1), patellar instability (n=1), significant arthrofibrosis (n=1), femoral condyle fracture (n=1), and ipsilateral ankle fracture after a fall (n=1). Spacer-related complications in the PI group included spacer dislocation/instability (n=2), patellar instability (n=2), significant arthrofibrosis (n=1), and femoral condyle fracture (n=1). No significant difference was found in rates of overall complications, antibiotic-related complications, infection recurrences, medical complications, ER visits, and hospital readmissions between spacer types (Table 3).

Table 3:

Comparison of Complication rates between polyethylene insert spacers and all-polyethylene tibia spacers. P-values are calculated using Fisher’s exact test.

Complications	All Spacers (n=126) (%)	Polyethylene Insert (n=62) (%)	All-polyethylene Tibia (n=64) (%)	P-value
All	53 (42)	24 (39)	29 (45)	0.48
Antibiotic-related	13 (10)	4 (6)	9 (14)	0.24
Spacer-related	13 (10)	6 (10)	7 (11)	1
Recurrent infection	19 (15)	9 (15)	10 (16)	1
Medical complications	31 (25)	13 (21)	18 (28)	0.41
Emergency Department visits	6 (5)	4 (6)	2 (3)	0.44
Readmissions	17 (13)	5 (8)	12 (19)	0.12

There was no difference in the rate of recurrent infection between the APT (14.1%, 9 of 64) versus the PI group (14.5%, 9 of 62, P=1.0). Overall infection eradication was achieved in 84.9% (107 of 126) of patients following spacer placement.

Average time to reimplantation for the APT group was 19.1 weeks (4.3 to 98.3) vs. 14.4 weeks (6.7 to 39.7) for the PI group (P=0.09). Twenty of 64 spacers (31.3%) in the APT group had not undergone reimplantation at the time of data collection, with an average follow up of 22.1 weeks (2.3 to 76.1), while 19 of 62 spacers (30.7%) in the PI group were not reimplanted, with an average follow up of 16.6 weeks (1.3 to 54.7). Of those patients, 5 from the APT group and 4 from the PI group died with an articulating knee spacer in place. Those subjects still living with a retained articulating knee spacer had an average longevity of 26.2 weeks (2.3 to 76.1) for APT constructs compared to 17.1 weeks (1.7 to 54.7) for PI constructs (P=0.26).

The average total cost of all spacers in the study was $1,909.13. The mean cost of the PI spacers was $856.28 less than APT spacers ($1,474.19 vs $2,330.47, P<0.0001). Steinmann pins were used as part of spacer constructs in both groups, while screws were used only in the PI group (Table 4). Tibial cement augments were used at similar rates (22 of 64 APT vs. 19 of 62 PI spacers; P=0.66).

Table 4:

Spacer components and adjunctive implants utilized

Spacer Type	Steinmann pin(s)	Screw(s)	No adjunctive implants	Tibial Augments
All-polyethylene tibial component (n=64)	35	0	29	22
Polyethylene insert (n=62)	36	26	0	19

Discussion:

We compared the results and complications of two MoP articulating spacer types including an APT and a tibial PI placed without the modular metal tibial baseplate. We compared complication rates, treatment efficacy, spacer duration, and implant costs, hypothesizing that while the APT would have lower complication rates and higher efficacy and durability, the PI would be less expensive. Our study showed no significant differences in performance of the two spacers and a lower cost of components for PI spacers, allowing for surgeon preference to play a role in the choice between the two spacers.

Overall, the spacer-related complication rate in our study was 10.3% (13 of 126) with no significant differences between the APT and PI cohorts. Both articulating knee spacers experienced a low but clinically meaningful rate (7 cases for APT versus 6 for PI) of spacer-related complications related to the first-stage explant or mechanical complications from the spacer components. Johnson et al previously reported that mobile articulating knee spacers were subject to a higher spacer-related complication rates than static spacers, namely prosthetic instability or fracture in 4 of 34 patients [20]. Some of these patients had molded, prefabricated polymethylmethacrylate femoral components, so the concern with femoral component fracture can essentially be eliminated using metal implants. Within the articulating spacer subgroup, MoP techniques have statistically fewer (p<.043) mechanical complications compared to other articulating spacer subtypes including cement-on-cement handmade, cement-on-cement prefabricated, and cement-on-cement molded techniques [15]. However, in each of our construct types, two subjects experienced a dislocation, underscoring the importance of achieving ligamentous balance even when placing a temporary articulating spacer device. Additionally, both groups experienced patello-femoral complications including one case of extensor mechanism disruption and one case of patellar instability in the APT cohort, and two cases of patellar instability in the PI cohort. This represented a lower incidence of patello-femoral complications than reported in previous studies. Maheshwari et al stated that patello-femoral complications may occur in up to 10% of TKAs [21]. Both groups reported one case each of arthrofibrosis and femoral condyle fracture; the APT group also had one patient who experienced an ipsilateral ankle fracture.

Infection eradication was achieved in approximately 86% of patients in both groups. This represents a similar rate as reported by Romanó et al, who described an average infection eradication rate of 91% following two-stage procedures using an articulating knee spacer, while static knee spacers had a slightly lower eradication rate of 87% [22]. Overall, two-stage procedures have resulted in higher eradication rates compared to one-stage procedures (89.8 and 81.9%, respectively) [22].

Articulating knee spacers have potential for two to three years of function in patients who are medically unfit or elect to forego reimplantation [8,9]. A total of 31% of patients (39 of 126) in our study did not undergo revision procedures, with an average follow up of 136 days within this sub-group, which is similar to previously published results [8]. Furthermore, our data demonstrates that for the patients with retained spacers, the APT components lasted an average of 39.3 days longer than the spacers with tibial PI components, though this may not be clinically relevant in the setting of two-stage treatment.

We found that the PI construct is less expensive than the APT ($1,474.19 versus $2,330.47; p <0.0001). These values included the use of adjunctive fixation (screws or pins) if utilized. Roof et al. compared the costs of MoP component spacers and all-cement spacers regarding the entire episode of care for PJI and found that MoP spacers may lead to an overall cost savings of approximately $12,000 [14]. Kalore et al. [18] studied the costs of autoclaved original femoral components, new femoral components, and silicone molded cement spacers. The authors reported an average cost of $3,589.00 for articulating spacers consisting of new femoral components, but this value included the cost of the implant, cement, and antibiotics [18], which may account for the difference observed in our study. When considering implant cost, using PI components as part of the articulating spacer construct is less expensive at our institution. These cost findings may be different at other institutions, as Robinson et al. demonstrated that implant costs for joint arthroplasty can vary between cases by a factor of over five, and 36 to 61% of variation may be attributed simply to the hospital in which the case was performed [23]. A potential advantage of APTs is rotational control from the stem and keel, which is not present in modular inserts, as APT implants were originally intended to provide long-term stability and fixation [19]. An advantage of the PI is the availability of larger implant thicknesses and varying degrees of constraint; APTs usually do not have as many options for component thickness and bearing surface constraint.

Our study has several potential limitations. It is retrospective, while a prospective, randomized study comparing APT versus PI components could more directly address the study question by better controlling for surgeon bias. Furthermore, our rudimentary cost analysis focused solely on implant-related pricing and did not account for variable costs related to home health services, needs for post-acute care facility placement, or costs of intravenous antibiotic therapy or monitoring. Surgeon bias regarding the timing for second stage reimplantation and willingness to use an articulating spacer as a potential permanent solution may skew the results in terms of spacer longevity. Shared decision making between the patient and physician for reimplantation occurred during follow-up based on patient condition and satisfaction with the articulating spacer. Additionally, allowing spacers to remain in place longer may decrease the difference observed in spacer longevity. Also, a major number of articulating knee spacers were placed just prior to the COVID-19 pandemic. Patients may have been more likely to decline second stage reimplantation during the pandemic if their spacer was functioning well and the infection was clinically treated, skewing patient bias towards retention of antibiotic spacer. Initial plans for a second surgery for reimplantation were potentially further hindered by hospital policies to delay non-emergent surgery during the study.

Conclusion:

Metal-on-polyethylene antibiotic knee spacers are effective for two-stage treatment of PJI in TKA. Complications, infection eradication, and spacer retention length were similar between the APT and PI constructs. Polyethylene insert spacers are less expensive and are potentially more cost effective if reimplantation is performed in the standard two-stage treatment paradigm. Both constructs may be durable if spacer retention is elected and will provide satisfactory results in the treatment of PJI in TKA patients.