Our prior post,When should a failed shoulder arthroplasty be treated for infection? brought up a number of questions about the management of the possibly infected shoulder. One of our shoulder fellows, Brad Graefe, summarized a recent case illustrating some of the challenges. The case is presented with the consent of the patient and the operating surgeon. Here's the story in brief
A patient did well for years after a shoulder arthroplasty and then developed unexplained pain that did not go away. At first, the clinical examination was essentially unchanged, and the radiographs were read as unchanged. The sedimentation rate was 3 mm/h, the C-reactive protein 2.8 mg/L, and the peripheral white count 5.74 ×10³/µL. A joint aspirate returned 3,295 nucleated cells per microliter with 12% neutrophils, 57% lymphocytes and 31% macrophages. At revision surgery, a frozen section of the collar membrane showed two neutrophils per high-power field and no organisms.
Nothing remarkable.
Within days of that surgery, all five specimens grew Cutibacterium acnes.
This post is about the delayed presentation of a Cutibacterium periprosthetic infection after a “honeymoon” period of good function.
The full story
A man in his late fifties came in with advanced glenohumeral arthritis in one shoulder. Years before, he had an os acromiale treated with ORIF. He wanted to keep lifting weights, so he chose a ream-and-run arthroplasty — humeral hemiarthroplasty with concentric reaming of the glenoid — together with an open biceps tenodesis.
Left shoulder radiographs at his 6-month post-operative visit were unremarkable
He did well. More than four years later he reported no pain in that shoulder, was lifting weights and doing push-ups, and his radiographs remained unremarkable. On the strength of that outcome, he had the same operation on the other side, and that shoulder did well too.
Left shoulder radiographs at 4-year post-operative visit were again unremarkable.
About six and a half years after the first operation he developed soreness in both shoulders after a day of yard work. The right shoulder settled. The left did not. His examination was unchanged. His radiographs were interpreted as showing no significant interval change compared with prior films. In retrospect there was some evidence of osteolysis of the calcar as seen below.
Left shoulder radiographs at the 6-year post-operative visit
Three months later, his examination had changed: he had pain at the end of active elevation and of rotation in both directions. Active and passive elevation were both 130 degrees, external rotation at the side 20 degrees, external rotation in abduction 30 degrees, internal rotation in abduction 80 degrees, and internal rotation behind the back to the back pocket. Supraspinatus strength was 4/5 and external rotation strength 5−/5.
Nonoperative treatment was continued, but the pain persisted, so his joint was aspirated and serum markers were drawn, with these results:
· Synovial nucleated cell count 3,295 per microliter, with 12% neutrophils, 57% lymphocytes and 31% macrophages
· Erythrocyte sedimentation rate 3 mm/h
· C-reactive protein 2.8 mg/L
· Peripheral white blood cell count 5.74 ×10³/µL
The serum markers were normal. The synovial count was slightly above 3,000 cells per microliter, but the differential was lymphocyte-predominant rather than neutrophil-predominant — not the pattern of a septic joint.
Notwithstanding these unimpressive results, the patient had a revision for persistent unexplained pain and because of new radiographs showing osteolysis along the medial calcar, the greater tuberosity, and along the glenoid as shown below.
Radiographs of the left shoulder before revision, showing humeral and glenoid osteolysis
What the operation showed
At surgery, aspiration of the shoulder yielded no joint fluid. There was no gross purulence. There was extensive osteolysis of both the humerus and the glenoid, with bone that looked necrotic. The glenoid had multiple deep areas of osteolysis.
A frozen section of the collar membrane showed two neutrophils per high-power field and no appreciable bacteria — below the threshold usually taken as evidence of infection. Five specimens were each sent for aerobic, anaerobic, acid-fast and fungal culture.
So at the moment a decision had to be made, the serum markers were normal, the aspirate was not inflammatory, the frozen section was negative, and there was no pus. What was abnormal was the bone.
What was done
The appearance of the tissue was enough to proceed as though this were an infection. The humeral components were removed, the shoulder was extensively debrided and irrigated, and a wound VAC was placed. A postoperative CT was obtained to assess the remaining bone stock.
CT after debridement, obtained to assess the remaining bone stock
Two days later a revision to a reverse total shoulder arthroplasty was carried out using a monoblock baseplate, a hooded glenosphere to load share with the remaining native bone, and a cemented humeral component. The greater tuberosity fractured during instrumentation of the humeral canal which was not surprising given the amount of bone loss from osteolysis. The greater tuberosity was securely repaired by suturing to the implant.
Radiograph after revision to a reverse total shoulder arthroplasty
Three days after the first debridement, one of the operative cultures began to grow Cutibacterium acnes; by post-operative day 8, all five were positive.
Intraoperative Culture Results
Specimen | Culture result | Growth |
Humeral head explant | Cutibacterium acnes | 1+ |
Proximal humeral bone | Cutibacterium acnes | 1+ |
Explanted humeral stem | Cutibacterium acnes | 1+ |
Collar membrane | Cutibacterium acnes | 1+ |
Periglenoid tissue | Cutibacterium acnes | 1+ |
Total | 5 of 5 specimens positive (100%) | All 1+ |
He was discharged on oral amoxicillin with a planned course of about three months and follow-up with an orthopaedic infectious disease specialist. He is being followed closely.
What a new multicenter model says about this case
The ASES Revision Shoulder Arthroplasty and PJI Multicenter Research Group has reported a prediction model for clinically apparent shoulder periprosthetic joint infection, built on 861 consecutive revision arthroplasties performed by 40 surgeons at 20 institutions [1]. Ninety-seven of those shoulders (11%) met the study’s definition of infection: a sinus tract communicating with the prosthesis, or intra-articular purulence. Seven variables entered the final model, which had a concordance statistic of 0.90 in the development cohort. Using this prediction model, a nomogram was constructed, and weights were assigned to each variable.
The model was built to predict infection that is clinically apparent. This patient had neither a sinus tract nor purulence, so by the study’s outcome definition he would have been counted among the shoulders without infection — and he grew Cutibacterium from five of five specimens. The limitations section of the article anticipates patients like this one: low-grade or indolent infections may follow a different biological pathway and may be associated with a different set of predictive factors. What those factors are is not yet known.
What we can take from this case
· Pain that persists without an explanation is a finding that may suggest infection, even if it surfaces months or years after the index arthroplasty.
· The lack of a host response by lab values (normal serum inflammatory markers, synovial values, and frozen section) cannot exclude a Cutibacterium PJI.
· Progressive osteolysis was the principal sign of infection; the laboratory tests did not support it.
· The necrotic-appearing bone seen at surgery was striking, especially in the absence of purulence and a better explanation for osteolysis such as polyethylene wear.
· Indolent infections may follow a different physiologic pathway that is not fully captured by a predictive model based on pathognomonic signs of infection as a reference standard.
The bottom line
The objective preoperative signs were a changed examination and progressive osteolysis, first visible at the calcar on films initially read as showing no significant interval change. Every laboratory test was falsely reassuring.
Two questions:
(1) Is there a better way to determine the risk of infection prior to revision so that treatment can be modified accordingly?
(2) Where did these Cutibacterium come from – were they present for six years after the index arthroplasty?
Getting things straight
Follow on twitter/X: https://x.com/RickMatsen
Follow on facebook: https://www.facebook.com/shoulder.arthritis
Follow on LinkedIn: https://www.linkedin.com/in/rick-matsen-88b1a8133/
Reference
[1] Hsu JE, Ricchetti ET, Tashjian R, Namdari S, Collins AP, Fedorka C, Kim HM, Zhang C, Presson AP, Whitson AJ; American Shoulder and Elbow Surgeons (ASES) Revision Arthroplasty/Periprosthetic Joint Infection (PJI) Multicenter Group. Predictive modeling of clinically apparent shoulder periprosthetic joint infection (PJI). J Shoulder Elbow Surg. Published online October 12, 2026. doi:10.1016/j.jse.2026.10.004
