Humeral periosteal spur sign with osteolysis as an early radiographic indicator of low-grade Cutibacterium acnes prosthetic shoulder infection

Abstract

The early diagnosis of low-grade Cutibacterium acnes prosthetic shoulder joint infection is challenging due to the lack of clinical and laboratory signs. Patients present with atypical symptoms such as stiffness or failure to improve shoulder function. The diagnosis is often delayed with impact on long-term outcomes. We present the case of an 82-year-old man with a surgical site erythema occurring 7 weeks after reverse shoulder arthroplasty associated with a light raise of C reactive protein (20 mg/L). At 9 weeks, radiographs revealed a ‘periosteal spur sign’ (humeral calcar) and localised adjacent osteolysis. Open surgery showed morphological signs of infection confirmed by growth of C. acnes in eight cultures. A ‘periosteal spur sign’ is a useful early radiographic indicator of low-grade prosthetic joint infection usually occurring with some delay after clinical symptoms. A high clinical index of suspicion is needed to proceed with biopsies and to initiate combined operative and antibacterial treatment.

Background

Prosthetic joint infection (PJI) is a serious complication of shoulder arthroplasty affecting 1%–2% of patients overall, 3% of patients with reverse shoulder arthroplasty (RSA) and up to 8% of men with RSA.^{1 2} PJI may require extended periods of hospitalisation, multiple reoperations, and lead to a significant morbidity and socioeconomic burden.³ Cutibacterium acnes (CA) (formerly Propionibacterium acnes), a lipophilic Gram-positive bacterium of the Cutibacterium sp family, is the most common microorganism in shoulder PJI, which is isolated in 30%–70% of cases usually forming a biofilm over prosthetic implants.^{4 5} Evaluation of patients with suspected PJI includes history and physical examination, imaging and cultures on tissue obtained by open biopsy plus/minus needle aspiration. Since CA is a microorganism with low virulence, clinical symptoms might be lacking in the early course of infection and the decision to proceed to open biopsies for a definitive diagnosis remains a challenge.⁶ The six main phylotypes (IA1, IA2, IB, IC, II and III) display differences associated with clinical disease, antibiotic resistance, virulence factors and inflammatory properties.⁷ CA that causes PJI originates from the patient’s own normal skin. Single-locus sequence typing could not identify any statistically significant differences between CA isolates from PJI and those from the skin. Traditional diagnostic criteria of infection (clinical features, erythrocyte sedimentation rate (ESR), C reactive protein (CRP) and culture on joint needle aspiration) have a low sensitivity at an early stage due to the lack of an inflammatory response. Delayed diagnosis of CA shoulder PJI is therefore a common scenario.³ There are recognised radiographic criteria of PJI which are mainly described in the late course of infection, that is, months after shoulder arthroplasty.⁸ The present case report shows that radiological signs may occur very early in low-grade CA shoulder PJI, that is, within weeks after arthroplasty. Imaging of early PJI is characterised by a ‘periosteal spur sign’ with adjacent osteolysis. This radiological indicator is useful to avoid a diagnostic delay by proceeding to open biopsies for microbiological documentation of infection followed by an early combined operative and antibacterial management.

Case presentation

An 82-year-old man presented with right-sided shoulder pain progressively worsening over several years due to osteoarthritis with high-grade partial rotator cuff tear and failure of non-operative treatment with corticosteroid infiltrations, the last performed more than 3 months prior to surgery. He had no history of diabetes and no other risk factors for immunosuppression.⁹ The patient underwent uneventful right-sided RSA with a standard deltopectoral approach (RSA; Ascend flex 3B; Perform reversed 25 mm base plate p with long peg and two screws; centric glenosphere of 42 mm; standard 1.5 mm offset tray in position 6; insert of 42 mm with 6 mm height) with 10 mm bio-increased offset bone grafting. A strictly standardised aseptic perioperative protocol was applied, including: (1) chlorhexidine solution to wash and shower in the 48 hours prior to surgery, (2) fourfold standard concentric centripetal disinfection with chlorhexidine in the operating theatre and (3) skin cover with an impregnated film (Ioban). The surgical time was 90 min. A control anteroposterior (AP) radiograph in internal rotation showed a well-implanted RSA without any signs of concern (figure 1, left). A drain was left in place and removed on day 2. The initial recovery and physiotherapy with active-assisted exercises without limitations of the range of motion (ROM) were normal. The patient presented on week 7 after surgery for a wound check because of a low-grade geographic erythema around the proximal surgical scar of 6×4 cm (figure 2, left). The wound was well healed and dry. The patient experienced painless heaviness of the arm and reported stagnation in progress of active shoulder flexion. Passive flexion was possible to 110°, extension to 30°, abduction to 80°, external rotation with the arm at side to 40° and internal rotation with the thumb reaching the buttock. Active ROM showed flexion to 80°, extension to 20°, abduction to 60°, external rotation to 25° and internal rotation to the proximal thigh. The patient had no fever, no chills, and no general symptoms such as fatigue, loss of appetite or increased sweating. Laboratory studies showed a normal leucocyte count and a CRP of 20 mg/L (normal: <10 mg/L). An AP radiograph showed a suspicion of speckled osteolysis at the medial side of the humeral osteotomy (calcar) at the entrance of the effective joint space (by definition the surface of the osteotomy which is covered by the implant) (figure 1, middle). We did not prescribe antibiotic treatment or any medication at this stage. By the 8th postoperative week, the erythema spontaneously disappeared (figure 2, right) and CRP was 31 mg/L. On week 9, there was no recurrence of the erythema, the patient remained pain-free but reported persistence of the heaviness of the arm with failure to progress with active shoulder function. CRP was 28 mg/L. The third postoperative AP radiograph in internal rotation (figure 1, right) showed a cortical reaction (spur sign) adjacent to the lysis at the humeral calcar. Due to a high index of suspicion for low-grade PJI, we proceeded to open biopsies through the previously used deltopectoral approach. We found a turbid fluid in the shoulder joint, an abundant scarring in the subacromial space, a sloughy yellowish tissue suggesting necrosis, a membrane in the effective joint space under the tray of the humeral implant and under the glenosphere. At the calcar, we observed a sloughy membrane with lysis and a periosteal reaction. These perioperative findings strongly suggested a low-grade PJI.

Figure 1.

This shows the postoperative anteroposterior radiographs on day 1 (left), after 7 weeks (middle) and after 9 weeks, with appearance of the ‘spur sign’ at the humeral calcar (right).

Figure 2.

This shows the clinical findings with appearance of a light skin erythema 7 weeks after surgery (left) and its spontaneous resolution 8 weeks postoperatively (right).

Six biopsies (see results below) were performed with different instruments according to the Oxford protocol and the implant was removed for sonication prior to insertion of a handmade antibiotic-loaded (gentamicin and vancomycin) spacer (figure 3).¹⁰

Figure 3.

This shows anteroposterior radiograph after spacer implantation 9 weeks postoperatively.

The 10 mm glenoid-sided bone graft was solid, well integrated and remained preserved after thorough lavage.

Investigations

Day 1 after surgery:

• AP radiograph with the arm on the body in internal rotation (figure 1, left).

Seven weeks after surgery:

• Leucocytes: 5.6 x10⁹/L; CRP: 20 mg/L.

• AP radiograph: suspicion of osteolysis on the medial side of the humeral osteotomy (calcar) at entrance of the effective joint space (figure 1, middle).

Eight weeks after surgery:

• Leucocytes: 4.9 x10⁹/L; CRP: 31 mg/L.

Nine weeks after surgery:

• Leucocytes: 4.9 x10⁹/L; CRP: 28 mg/L.

• AP radiograph in internal rotation: cortical reaction (spur sign) adjacent to lysis at the humeral calcar (figure 1, right).

Invasive samplings (n=8) at revision surgery, 9 weeks after the index procedure, showed turbid joint fluid which was positive for leucocytes and culture positive for CA. All open biopsies (n=6) were culture positive for CA. Cultures where obtained from the joint capsule (n=2), from tissue slough under the glenosphere (n=2), tissue slough at the site of the calcar lysis (n=1) and tissue from inside the humerus after stem removal (n=1; culture positive after enrichment). According to the criteria of Asseray et al, more than two positive cultures with local signs of infection are highly indicative (>90%) of CA infection.¹¹ The authors have defined criteria to differentiate between real CA infection and contaminations.

Sonication of the implant showed >500 CFU/mL CA (>50 CFU/mL: implant-associated infection). Sonication provides the option for further valuable culture samples after dislodging bacteria from the biofilm and implant. After multiple culture-positive biopsies, this optional process was not necessary.

The antibiogram of CA showed sensitivity for penicillin, amoxicillin–clavulanic acid, imipenem, clindamycin, levofloxacin and rifampicin.

Differential diagnosis

The differential diagnoses considered in this early setting after RSA were the following:

1. Aseptic postoperative frozen shoulder with stiffness.

2. Superficial wound infection with associated erythema.

3. Low-grade prosthetic joint infection.

Heaviness of the arm, stiffness and failure to progress with active flexion can occur at 6 weeks after surgery and might be associated with a form of aseptic frozen shoulder.

Transient skin erythema is rare and might be associated with suture granuloma or superficial wound infection, this differential diagnosis would be consistent with a light raise of CRP values.

The combination of a transient skin erythema, a light CRP raise, a functional status differing from the expected clinical course 6 weeks postoperatively and radiographic signs of early osteolysis with a periosteal reaction on the humeral side consistently suggest the diagnosis of a low-grade PJI and imply an indication for surgical exploration with perioperative samplings for microbiological documentation.

Until proven otherwise, every report of pain (not present in this case), prolonged stiffness and loosening of the shoulder prosthesis should be regarded as an indication of infection.

Treatment

After removal of the prosthesis and thorough 360° debridement followed by repetitive lavage with a 50 mL bulb syringe avoiding pulsatile pressure lavage, a handmade cemented spacer (Palacos with gentamicin and vancomycin) was inserted. The patient received 3 weeks of intravenous antibacterial therapy with amoxicillin–clavulanic acid (2.2 g three times per day), followed by oral switch to clindamycin (600 mg three times per day).

Five weeks after this first-stage revision under continuous antibiotic treatment, the spacer was removed and sent for sonication and culture, biopsies for cultures were performed and the shoulder debrided and washed. The leucocyte count, ESR and CRP were all normal at the time of second-stage revision. Intraoperatively there were signs of moderate mechanical erosion of the spacer on the glenoid associated with passive pendulum exercises and use of the hand in front of the body with the arm supported in a sling but no signs of persistent infection. We proceeded with second-stage implantation of a new non-cemented modular press/fit revision stem (Aequalis revive flex 13/90; proximal body 13/132.5°; spacer 13/20; reversed tray +6 mm, 3.5 mm eccentric; insert 39 mm+9 mm; Perform reversed base plate with a 9.5 mm central screw and two additional screws and a 39 mm+3 mm eccentric glenosphere). For biomechanical reasons, a distal press-fit stem was chosen over a cemented stem with possible antibiotic cement loading because of a highly probable eradication of the sensitive organism at second-stage surgery. We reamed 3 mm of the metaphyseal bone and the remaining glenoid-sided bone graft was still partly present after removal of the spacer. Figure 4 shows the AP radiographs on day 1 after second-stage revision. Intravenous amoxicillin (2 g four times per day) was administered for 1 week postoperatively, followed by oral switch to clindamycin (600 mg three times per day). Two weeks after reimplantation, under dry wound conditions, clindamycin was switched due to poor tolerance (unpleasant taste) to levofloxacin (500 mg two times per day) and rifampicin (300 mg two times per day) for a total postoperative antibacterial treatment duration of 6 weeks.

Figure 4.

This shows the second-stage reverse shoulder arthroplasty on day 1.

Outcome and follow-up

All the intraoperative cultures (spacer sonication and biopsies) were negative. The clinical course of the surgical wound was unremarkable (dry and without any erythema). CRP was normal 2 weeks after the last surgery. The patient progressed after 6 weeks to 140° active-assisted painless shoulder flexion and without heaviness of the arm. Figure 5 shows the well-inserted revision RSA after 3 months.

Figure 5.

This shows the radiographs of the well-inserted uncemented revision reverse shoulder arthroplasty after 3 months.

Discussion

As shoulder arthroplasty becomes more and more frequent, complications of PJI are likely to result in increasingly significant individual morbidity and socioeconomic burden.¹² Low-grade infections play an important role in PJI of the shoulder.¹³ They are most often caused by CA, a low-virulent skin coloniser.^{4 14} The diagnosis of PJI due to these microorganisms is often documented late after arthroplasty. In one large series, the average time from arthroplasty to diagnosis of PJI was 3.5 years.¹⁵ Since clinical signs and symptoms are poor and mainly atypical for infection, the decision to proceed with invasive biopsies remains a challenge.⁶ The Musculoskeletal Infection Society in 2014 adapted criteria for identifying PJI.

Major criteria include two positive cultures with identical organisms from (1) periprosthetic tissue samples and (2) a fistula or sinus. Minor criteria are (1) elevated ESR and CRP, (2) elevated leucocyte count or ++ change on leucocyte esterase test strip, (3) elevated synovial fluid polymorphonuclear neutrophil percentage, (4) one single positive culture and (5) positive histology. One major or three minor criteria are indicative of PJI.¹⁶

Pottinger et al analysed revisions of shoulder arthroplasty without obvious classic signs of infection and reported that radiographic humeral osteolysis was an important prognostic indicator with a more than 10-fold increase in the OR for positive cultures for CA. Other factors significantly associated with a positive culture (300%–600% increase in likelihood) were male sex, cloudy fluid at revision surgery, membrane formation and glenoid wear. It is important to point out that these data were derived from patients undergoing revision arthroplasty without obvious clinical signs of infection for mechanical reasons for revision after long time intervals after the primary shoulder arthroplasty.¹⁷

Radiographic signs indicative of PJI of the shoulder include endosteal scalloping, generalised bone resorption, periosteal reaction and periprosthetic radiolucency.5 8 15 17–21

In our case, we found subtle humeral osteolysis with periosteal reaction as early as 8 weeks after the primary arthroplasty.

These signs are non-specific as they might be caused by non-infectious conditions such as mechanical loosening and osteolysis. They have been reported in studies with long time intervals between the primary arthroplasty and the radiographic description at the time of surgical revision. Radiographic signs with high specificity are gas formation and active, immature periostitis. In cases with non-specific clinical signs and symptoms and normal-looking radiographs, surgeons should have a high index of suspicion and employ other diagnostic modalities like serum inflammatory markers, assessment of synovial biomarkers, white cell scintigraphy, nuclear imaging studies, culture, PCR and biopsy to rule out PJI.

To our knowledge, this report is the first description of a low-grade CA infection leading to early radiographic signs of osteolysis with periosteal reaction after shoulder arthroplasty. Without meticulous scrutiny of radiographic images, the patient could have been left untreated for months to years or even be lost to follow-up, since the clinical presentation was non-specific for PJI. However, the radiographic cues combined with a high index of clinical suspicion led us to undertake multiple open biopsies which confirmed CA infection allowing an early appropriate management combining surgery and antibacterial therapy.

Patient’s perspective.

At first, everything went well after my shoulder operation, with evident improvement of the pain and limited movements which were present before operation. Later, I was worried about a discrete redness around the scar, which however disappeared soon without any action. I never had pain at this time but lifting up my arm was unexpectedly difficult. Three to six weeks after the exchange of the prosthesis I have been able to lift up my arm again. This was very different compared to the progress of my rehabilitation after the first surgery.

Learning points.

• Diagnosis of low-grade prosthetic joint infection (PJI) of the shoulder is often delayed by months to years.

• Clinical signs depend on the phylotypic class of Cutibacterium acnes, however they are often non-inflammatory and atypical with stiffness, heaviness or reduced functional status.

• Postoperative radiographs should be scrutinised for fine changes including osteolysis and periosteal reactions.

• An early humeral periosteal spur sign with osteolysis can be an important indicator for PJI.

• An abnormal postoperative functional recovery combined with a radiographic humeral spur sign and osteolysis should be considered as low-grade PJI until proven otherwise and mandate invasive samplings for microbiological documentation followed by prompt appropriate surgical and antimicrobial treatment.