Two-stage reimplantation for deep infection after total elbow arthroplasty

Abstract

Background

Persistent infection rate after 2-stage reimplantation complicating elbow arthroplasty has been reported to be as high as 25%. The purposes of this retrospective study were to determine the infection eradication rates, complications and outcomes in a cohort of patients treated with two-stage reimplantation for deep periprosthetic joint infection (PJI) following total elbow arthroplasty (TEA) and to determine possible associated risk factors for treatment failure.

Methods

Between 2000 and 2017, 52 elbows underwent a two-stage reimplantation for PJI after TEA. There were 22 males and 30 females with a mean age of 61 (range, 25–82) years. The most common bacterium was Staphylococcus epidermidis (28 elbows). Mayo Elbow Performance Scores were calculated at the latest follow-up. Mean follow-up time was 6 years (range, 2–14 years).

Results

PJI was eradicated in 36 elbows (69%). The remaining 16 elbows were considered treatment failures secondary to recurrent infection. The risk of persistent infection was 3.3 times higher in elbows with retained cement (p 0.04), and 3.5 times higher when the infecting organism was Staphylococcus epidermidis (p 0.06).

Conclusion

Two-stage reimplantation for PJI after TEA was successful in eradicating deep infection in 69% of cases. The eradication of PJI after TEA still needs to be improved substantially.

Introduction

Deep periprosthetic joint infection (PJI) is a particularly devastating complication after total elbow arthroplasty (TEA). The incidence of PJI after TEA has been reported to range between 1% and over 12%.^1–4 Controversy remains regarding the comparative success rate of the various surgical strategies to manage PJI after TEA.^5–7 Treatment options include chronic antibiotic suppression, ⁸ debridement with implant retention,^1,9,10 staged debridement with polyethylene exchange, ¹¹ one-stage reimplantation,^2,12 two-stage reimplantation,^1,13–15 and resection arthroplasty.^16–18

Although one-stage reimplantation has been reported to provide acceptable success rates after hip and shoulder arthroplasty,^19,20 two-stage reimplantation is considered the treatment of choice for PJI after TEA,^1,13–15 especially when one or both implants are loose. However, removal of well-fixed cemented components and complete removal of cement may lead to severe bone loss, intraoperative fracture, and nerve injury.

The reported persistent infection rate after two-stage reimplantation for PJI complicating hip and knee arthroplasty has been reported to range between 5% and 10%. ²¹ On the contrary, the reported persistent infection rate after two-stage reimplantation complicating elbow arthroplasty has been reported to be as high as 25%.^1,13–15

We hypothesized that (1) two-stage reimplantation for PJI following TEA leads to satisfactory rates of infection eradication and (2) partial component removal may lead to higher rate of persistent infection after two-stage reimplantation for PJI following TEA. As such, the purposes of this study were (1) to determine the infection eradication rates, complications and outcomes in a cohort of patients treated by two-stage reimplantation for deep PJI following total elbow arthroplasty, (2) to identify differences between partial and complete component removal at the time of resection, and (3) to determine possible associated risk factors for treatment failure.

Materials and methods

Patients

After Institutional Review Board approval, our Joint Registry Database was queried to identify all elbows treated with two-stage reimplantation for the management of PJI after TEA between 2000 and 2017. A total of 84 elbows were identified. A retrospective review of the medical records for these elbows was conducted to identify the following exclusion criteria: less than 4 weeks of intravenous antibiotic therapy in between stages (four patients), implant resection performed at another institution prior to referral (18 patients), or less than 2 years of follow-up after reimplantation (10 patients). A total of 52 elbows in 52 patients were analysed for this study. There were 30 women and 22 men, with a mean age of 61 ± 13 (range, 25–82) years at the time of their first-stage resection surgery. The mean BMI was 28 ± 7. Twenty-nine patients (56%) had never smoked, 16 (31%) were active smokers, and 7 (13%) were former smokers.

Prior surgeries

The index primary TEA had been performed at our Institution in 21 patients (40%). The underlying diagnosis leading to arthroplasty included post-traumatic arthritis in 20 elbows (39%), inflammatory arthritis in 17 (33%), distal humerus nonunion or failed open reduction and internal fixation in nine (17%) (2 of these had a prior history of infected nonunion), primary osteoarthritis in four (8%) and acute distal humerus fracture in two (4%).

Twenty-seven (52%) of the fifty-two patients had at least one procedure performed on the elbow before their index TEA (mean, 1; range, 1–8 surgeries), the most frequent procedure being fixation of a distal humerus fracture (18) (Table 1). In addition, 38 (74%) had undergone one or more surgeries after TEA but prior to resection (mean, 2; range, 1–7 surgeries), including wound debridement in 28 elbows (Table 1). Overall, PJI complicated primary TEA in 32 elbows and revision TEA in 20 elbows (due to aseptic loosening or periprosthetic fracture after index TEA).

Table 1.

Summary of surgeries before and after TEA.

Surgeries before the index TEA

ORIF distal humerus fracture (18) Join debridement for arthritis (7) Interposition arthroplasty (7) Lateral collateral ligament reconstruction (5) Arthroscopic synovectomy (3) Radial head resection or replacement (3) Ulnar nerve transposition (3)

Surgeries after index TEA prior to resection

ORIF periprosthetic fracture (5) Revision for aseptic loosening (16) Wound debridement (28) Bushing exchange (3) Triceps reconstruction (2) Ulnar nerve transposition (3) Flap coverage (1)

ORIF: open reduction internal fixation.

Evaluation at the time of presentation

PJI after TEA led to various combinations of pain, loss of function, redness and drainage. A diagnosis of prosthetic joint infection was made using the guidelines of the Musculoskeletal Infection Society, ²² also adopted by the American Academy of Orthopaedic Surgeons: (1) presence of a sinus tract or fistula that communicated with the prosthesis and/or (2) two or more positive cultures for the same microorganism in samples obtained by preoperative aspiration or biopsy at the time of surgery. A high suspicion for infection was also held for elbows presenting with unexpected component loosening combined with increasing pain and swelling and elevated erythrocyte sedimentation rate or C-reactive protein. ²³ There was a positive culture from the aspiration performed preoperatively in 37 elbows (71%). Radiographic evidence of implant loosening was present in 45 elbows (86%) prior to resection arthroplasty.

Microorganisms

Table 2 summarizes the organisms cultured from fluid samples obtained prior to surgery or surgical samples obtained at the time of the resection. The most common microorganisms isolated included Staphylococcus epidermidis (28 elbows), Staphylococcus aureus (nine elbows) and Cutibacterium acnes (four elbows). In five additional elbows, S. epidermidis grew mixed with another organism. The presence of S. epidermidis was observed in 33 (64%) of 52 elbows.

Table 2.

Organisms cultured in the 52 elbows included in this study.

Staphylococcus epidermidis	28 (54%)
Staphylococcus aureus	9 (17%)
Cutibacterium acnes	4 (8%)
Others	6 (11%)
Polymicrobial including S. epidermidis	5 (10%)

Two-stage reimplantation

All elbows included in this study were treated with two procedures: removal of implants and cement with debridement followed by component reimplantation a minimum of four weeks later. All surgeries were carried out by one of three senior surgeons. The mean time interval between the index TEA and resection arthroplasty was 7 years (range, 0.14–37 years). PJI was categorized as acute when presenting within three months of TEA (1 elbow), subacute (3 months to one year, nine elbows), and chronic (over 1 year, 42 elbows).

At the time of implant resection, various approaches were utilized, including the Bryan-Morrey approach (n = 19), triceps splitting (n = 16), extended olecranon osteotomy (n = 5), lateral paraolecranon approach (n = 3), Mayo-modified Kocher approach (n = 2), triceps tongue (n = 2) and bilaterotricipital approach (n = 1). In nine elbows (17%) one component was well-fixed, and the treating orthopaedic surgeon elected to retain the well-fixed component to avoid the potential for complications and morbidity related to implant removal (partial resection). In the remaining 43 elbows (83%), all components were resected using a combination of pencil tip bur to free up the cement-implant interface or dedicated extractors. In an attempt to remove all retained cement, a combination of techniques were used under fluoroscopic control including high-speed burs, sequential drilling, long grasping instruments, cannulated flexible reamers and occasionally visualization of the endosteal canal with an arthroscopic camera. We avoid the use of ultrasound devices in the humerus because of possible thermal damage to the radial nerve. When using flexible reamers, it is important to make sure that the guidewire is perfectly centred in the canal; otherwise, the reamer will create a cortical defect, and in the humerus, it may lead to iatrogenic radial nerve injury. Additionally, humeral or ulnar osteotomies were performed in 17 elbows (ulnar only, eight elbows; humeral only, eight elbows; both bones, one elbow). In those cases, in which an osteotomy has been performed, it was stabilized with metal wires or polydioxanone (PDS) sutures. All cement was successfully removed in 30 elbows (58%) while persistent cement was observed in postoperative radiographs in 22 elbows (42%). Cement removal was considered particularly challenging in those elbows where a cement restrictor had not been used and the cement extended substantially along the length of the humeral and/or ulnar canal (Figure 1).

Figure 1.

(a) Lateral radiograph of a TEA with a loose humeral component and extensive cement in the ulnar canal without cement restrictor (red arrow). (b) Resection arthroplasty with metallic pins coated with cement inserted in the canals and persistent cement in the ulnar canal (red arrow).

Antibiotic-loaded polymethylmethacrylate was placed in the articular space prior to closure, using a dose of 4 grams of vancomycin and 4.8 grams of gentamicin per batch of cement. Methylene blue was commonly added to colour the cement and facilitate later identification if cement removal was necessary. Early on, a combination of metallic pins coated with cement and cement beads was implanted. More recently, the use of external fixator components as an internal fixator was adopted: pins or bars coated with antibiotic-loaded cement are placed in the canals and linked using fixator connectors; additional cement is then moulded in the articular space over the fixator to stabilize the joint (Figure 2).

Figure 2.

Internal–external fixator. (a) External fixator pins are coated with antibiotic-loaded cement. (b) Pins or bars are fixed in approximately 30 degrees of flexion. (c) Radiograph of Internal–External Fixator.

After the first-stage resection, all patients were treated with a course of intravenous antibiotic therapy. Antibiotic selection was based on cultures and sensitivities. The mean time of antibiotic treatment was 41 days (range, 30–90 days). Antibiotics were often discontinued several days prior to reimplantation. The mean interval between stages was 21 weeks (range, 4–140 weeks). Soft-tissue flaps for coverage were performed between stages in two elbows.

At the time of the second-stage reimplantation, all components were fixed with antibiotic-loaded cement (1 gram of vancomycin and 1 mL of methylene blue per batch of cement). Components implanted at the second-stage surgery included a Coonrad-Morrey prosthesis (Zimmer, Warsaw, Indiana, USA) in 33 elbows and a Latitude prosthesis (Wright Medical, Bloomington, Minnesota) in 19 elbows. Special reconstructive techniques were required in 26 elbows (50%), including impaction grafting (n = 8), strut allograft augmentation (n = 7), internal fixation of an associated fracture (n = 1), an allograft-prosthetic composite (n = 9) and implantation of the ulnar component into the radius (n = 1). Triceps reconstructive techniques were required at the time of reimplantation in nine (17%) elbows, using an ulnar allograft prosthesis composite (APC) with triceps in six elbows, ²⁴ and an Achilles tendon allograft in three elbows.

Postoperative management

After surgery, the elbow was immobilized between 2 and 6 weeks depending on the need to protect an associated reconstruction of the extensor mechanism. Passive and active assisted range of motion exercises were initiated once the immobilization was discontinued.

Evaluation

All medical records were reviewed retrospectively. If the patients had been examined at our institution (43 patients), data from that evaluation were used to calculate the Mayo Elbow Performance Score (MEPS). ²⁵ Patients who did not return for final follow-up were called (nine patients). The MEPS in these latter nine patients should only be considered an estimate since it was based on the patient's own assessment of their elbow stability and range of motion. A retrospective review of the medical record was conducted to record this information and also identify complications and reoperations. For the purposes of this study, the PJI was considered eradicated for elbows with negative cultures at the time of reimplantation and no additional surgeries for persistent infection.

The mean follow-up time was 6 years (range, 1.8–14 years) and no patients were lost to follow-up.

Statistical analysis

Data are reported using standard summary statistics, including means and standard deviations for continuous variables, and counts and percentages for categorical variables. Comparisons between patients in whom infections were eradicated and those with persistent infections at last follow-up were performed using Wilcoxon rank-sum tests for continuous or ordinal variables, and chi-square tests for categorical variables. Additional analyses were conducted using logistic regression to estimate the odds of having a persistent infection. All analyses were performed using SAS version 9.4 (SAS Institute Inc., Cary, NC).

Results

Eradication of infection

At most recent follow-up, the PJI was considered to be eradicated in 36 elbows (69%). The remaining 16 elbows (31%) were considered treatment failures secondary to recurrent infection. No statistically significant differences were observed in failure rates between patients with partial component resection (33%) and those with resection or both components prior to reimplantation (30%, p = 0.7). Four of the elbows (8%) in which the PJI was initially considered to be eradicated, eventually developed a secondary PJI by a different microorganism (Enterococcus spp).

We analysed risk factors for treatment failure by defining as either persistent or late infection (i.e. the four elbows with late infection were also considered to be failures). The risk of treatment failure was 3.3 times higher in those with retained cement, and 3.5 times higher when the infecting organism was Staphylococcus epidermidis. (Tables 3 and 4).

Table 3.

Comparative analysis to identify risk factors for persistent or late infection.

	Infection eradicated (N = 32)	Persistent or late infection (N = 20)	p Value
Age mean ± SD	59 ± 14	64 ± 10	0.2
Age >65 n (%)	10 (31)	11 (55)	0.09
BMI mean ± SD	28 ± 6.5	30 ± 8.7	0.5
Gender n (%)			0.4
Male	12 (38)	10 (50)
Female	20 (63)	10 (50)
History of smoking			0.3
Never/Former	24 (75)	12 (60)
Active	8 (25)	8 (40)
Underlying diagnosis n (%)			0.1
Post-traumatic arthritis	11 (34)	9 (45)
Inflammatory arthritis	10 (31)	7 (35)
Primary degenerative arthritis	4 (13)	0 (0)
Nonunion distal humerus fracture or failed ORIF	7 (22)	2 (10)
Distal humerus fracture	0 (0)	2 (10)
Primary/Revision TEA n (%)			0.7
Revision	13 (41)	7 (35)
Primary	19 (59)	13 (65)
Time between TEA and one stage (months) mean ± SD	96 ± 92	75 ± 105	0.3
Time in between stages (weeks) mean ± SD	24 ± 32	15 ± 12	0.4
Number of surgeries prior to TEA mean ± SD	1.3 ± 1.7	1.2 ± 1.5	0.8
Number of surgeries prior to two stage mean ± SD	1.8 ± 1.8	1.4 ± 1.2	0.6
Presence of SCN	17 (53)	16 (80)	0.05
Type of two stage			0.7
Total two stage	27 (84)	16 (80)
Partial two stage	5 (16)	4 (20)
Cement removed			0.04
Retained cement	10 (31)	12 (60)
All cement removed	22 (69)	8 (40)
Reconstructive techniques at two stage			0.6
No	17 (53)	9 (45)
Yes	15 (47)	11 (55)
Positive cultures after two stage			0.2
No	28 (88%)	15 (75%)
Yes	4 (13%)	5 (25%)
Chronic antibiotic suppression			0.3
No	24 (75%)	12 (60%)
Yes	8 (25%)	8 (40%)

SCN: S. epidermidis.

Table 4.

Logistic regression analysis to identify risk factors for eradicated vs persistent or late infection.

	Odds ratio	95% CI odds ratio	P value
Age	1.04	(0.99, 1.1)	0.11
Age >65	2.7	(0.9, 8.5)	0.09
Gender	1.8	(0.5, 5.2)	0.4
BMI	1.03	(0.95, 1.1)	0.4
History of smoking	1.9	(0.6, 6.6)	0.3
Primary/Revision TEA	0.8	(0.3, 2.5)	0.7
Time between TEA and one stage	1.00	(0.99, 1)	0.4
Time in between stages	0.98	(0.9, 1)	0.2
N° surgeries prior to TEA	0.96	(0.7, 1.4)	0.8
Nª surgeries prior to one stage	0.8	(0.6, 1.2)	0.4
Presence of SCN	3.5	(0.96, 13)	0.06
Type of two stage	0.7	(0.2, 3.2)	0.7
Retained cement	3.3	(1, 10.6)	0.04
Reconstructive techniques at two stage	1.4	(0.4, 4.3)	0.6
Positive cultures after two stage	2.3	(0.5, 10)	0.3
Chronic antibiotic suppression	2	(0.6, 6.6)	0.3

SCN: S. epidermidis.

Complications and reoperations

A total of 20 patients (38%) required debridement (n = 1), implant resection (n = 11) or additional two-stage reimplantation (n = 8) for persistent infection or late infection by a different microorganism (Figure 3). The mean interval time between the second-stage reimplantation and further surgery for persistent or late infection was 37 months (range, 1.6–136 months). Soft-tissue flaps for coverage were required in seven elbows, five of them with persistent infection and 2 complicated with a late infection caused by different microorganisms.

Figure 3.

Flowchart summarizes all elbows included in this study.

Additionally, surgery for non-infectious complications was performed in 14 of the 32 elbows free of infection (44%) (Figure 3). Reasons for reoperation included aseptic loosening (n = 7), triceps weakness (n = 1), ulnar periprosthetic fracture (n = 4), allograft nonunion (n = 1) and radial nerve palsy (n = 1). Additional procedures performed for non-infectious complications included revision arthroplasty (n = 12, with APC reconstruction in five elbows and strut augmentation in one elbow), triceps reconstruction (n = 1), and tendon transfers for radial nerve palsy (n = 1). At the most recent follow-up, 24 (46%) of the prosthesis implanted at the time of the second stage remained well-fixed.

Clinical results

Amongst the 32 elbows with no evidence of either persistent infection or late infection, there was adequate information in the medical record to calculate or estimate the MEPS in 26 elbows. The most recent mean MEPS was 72 ± 22 points (range, 10–100 points). According to the MEPS, results were considered excellent in 5 (19%), good in 11 (42%), fair in 5 (19%) and poor in 5 (19%) elbows. For elbows with surviving implants, the mean extension was 28 degrees (range, 0–90 degrees), and mean flexion was 129 degrees (range, 60–150 degrees).

Discussion

The results of our study seem to indicate that two-stage reimplantation may lead to the eradication of infection in approximately 70% of the elbows. However, due to high rate of complications and reoperations after PJI complicating TEA, some elbows may later develop PJI due to a different microorganism. Finally, even for elbows with surviving implants, good or excellent results according to the MEPS are obtained in only 60% of the elbows.

The results of our study are consistent with persistent reinfection rates reported previously in the literature, ranging between 20% and 28%.^13–15 When confronted with such a high rate of unsatisfactory outcomes, a number of questions come to mind, including why is PJI so difficult to manage after TEA, what could be done to improve current outcomes, and how can we best prevent this devastating complication.

A number of factors may play a role in the relatively high failure rate of two-stage reimplantation for PJI after TEA. The soft-tissue envelope of the elbow is fragile, complete removal of cement from the ulnar and humeral canals is difficult and fraught with complications, and the reconstructive techniques required at the time of the second stage increase the length of time to perform and the complexity of the procedure. ¹⁶ In our study, 52% had undergone one or more surgeries prior to their index arthroplasty. In addition, soft-tissue coverage was required in several elbows, and complex reconstructive techniques were required in half of the elbows.

Staphylococcus epidermidis was particularly prevalent amongst those elbows with persistent or late infection. This association between Staphylococcus epidermidis and a higher failure rate has been previously documented in another study. ¹ Furthermore, aspiration was able to recover an organism 70% of the time prior to staged treatment. As such, every attempt should be made to identify the species of organism prior to surgery as prior positive cultures for Staphylococcus epidermidis in particular can guide the ultimate treatment. If the clinical presentation is suggestive of infection, but laboratory studies are negative, due to low very negative predictive value of aspiration, ²³ arthroscopic or open biopsies may be considered to obtain tissue for culture and debridement and synovectomy can be performed concurrently.

Another layer of complexity is introduced by the fact that removal of well-fixed components and complete removal of cement at the elbow are fraught with complications. Despite no statistically significant differences were observed in this study in terms of treatment failure between patients with partial or complete implant resection prior reimplantation, higher rates of reoperation were reported in elbows patients with partial two-stage either due to infectious or non-infectious complications. As such, the treating orthopaedic surgeon must make a judgment call to balance the benefits and risks of complete debridement in the light of potential catastrophic complications that decrease the chances of a successful reconstruction at the time of the second stage.

It is important to emphasize that the rate of reoperation for non-infectious reasons after two-stage reimplantation cannot be neglected. Cheung et al. ¹⁴ reported that between 25% and 30% of the elbows with no evidence of persistent or late infection needed additional surgery for other reasons. The most common complications were aseptic loosening (10%) followed by periprosthetic fracture (8%). Not surprisingly, many elbows with surviving implants but multiple complex surgeries do not achieve a satisfactory clinical outcome despite eradication of infection.

Considering the suboptimal results of two-stage reimplantation for the management of deep PJI after TEA, every effort must be made to prevent this complication at the time of primary or revision elbow arthroplasty. Strategies that might help in that regard would include avoidance of TEA in patients with modifiable risk factors, such as retained cement until those are corrected. As far as possible, a homogeneous treatment protocol should be attempted. However, given these patients’ complexity, treatment must often be individualized according to the microorganism involved, state of soft tissues and bone stock to avoid additional complications.

Our study has several limitations, most notably its retrospective nature and the lack of a standard protocol for the management of PJI after TEA. Additionally, data for the calculation of MEPS were obtained through a phone call in some patients included in this study. Such calculations should only be considered an estimate since they were based on the patient's own assessment of their elbow stability and range of motion. At the same time, strengths of the study include a relatively high sample size compared to previously published studies, strict inclusion criteria, and very low attrition rate, with no patients lost to follow-up.

In conclusion, two-stage reimplantation for deep PJI after primary or revision TEA in our patients had a high rate of recurrent infection with either the same organism (70%) or a different one (8%). Many of these procedures require soft-tissue coverage of complex reconstructive techniques. Non-infectious complications were common after the successful eradication of PJI. Every effort should be made to prevent infection after primary or revision TEA, and further changes in surgical practice and additional research are required to improve the outcome of two-stage reimplantation in the management of PJI complicating primary and revision TEA.