Revision Total Elbow Arthroplasty

How Bone Loss Is Actually Described at the Elbow

Unlike the hip, the elbow has no single universally adopted bone-loss classification in the way Paprosky is used for the femur and acetabulum. The practical consequence is that the deficiency has to be described rather than assigned a grade, and four questions do the describing. Answer them from a CT with metal-artefact reduction and the reconstruction largely chooses itself.

Periprosthetic Fracture: the Mayo Classification

Periprosthetic fracture is both a reason a primary arthroplasty comes to revision and the commonest single complication of the revision itself, so it needs its own framework. The classification in general use comes from O'Driscoll and Morrey's Mayo experience of more than 1,000 total elbow arthroplasties, and it is built on where the fracture lies relative to the stem - applied the same way on the humeral and the ulnar side.

The Infected TEA in Detail

Debridement with implant retention (DAIR) works far less well at the elbow than the phrase "acute infection, retain the implant" suggests. In a Mayo cohort of 26 elbows with periprosthetic joint infection managed by DAIR, the procedure failed in 17 - a failure rate of 65% at two years (95% CI 41.3-79.6%) - with a median time to failure of only 43 days. The authors' own conclusion is that a patient being offered DAIR must be told the chance of success is approximately 35%. That is the number that makes the consent honest, and it is why the staged route is the default here in a way it is not at the hip or knee.

Two findings in that series were absolute rather than relative: DAIR failed in every case with a sinus tract and in every case that was culture-negative. Either finding should therefore be read as a contraindication to retention rather than as a reason for caution. A shorter symptom duration and a monomicrobial infection characterised the patients who did well, though with those numbers the associations did not reach statistical significance.

The staged protocol. First stage: remove all components and cement, radical debridement, deep samples from several sites before antibiotics, and an antibiotic-loaded cement spacer. Organism-directed intravenous therapy follows, guided by microbiology and infection-disease input. Re-implantation is delayed until the infection is controlled clinically and biochemically. Two features specific to the elbow shape this:

• The soft-tissue envelope is thin and posterior, with little subcutaneous cover over the implant and often several previous incisions. Wound dehiscence is a prominent complication in the published revision series, so use the previous incision where possible, avoid parallel incisions that devascularise the skin bridge, handle the edges minimally, and involve plastic surgery early where cover is doubtful.
• Resection arthroplasty is a legitimate endpoint, not simply a failure. Where re-implantation is not feasible - uncontrolled infection, unreconstructable bone, or a patient unfit for further surgery - a resection gives reasonable pain relief at the cost of a flail, unstable elbow, and that trade should be described to the patient in those terms rather than as a fallback of last resort.

Revise to a Linked Implant

The revision elbow is, almost by definition, the elbow with deficient bone, incompetent collateral ligaments, or both - which is precisely the situation an unlinked (resurfacing) design cannot tolerate, because it depends on the soft tissues for stability. So a failed unlinked implant, and a failed linked implant being re-done, are in practice both revised to a linked (semi-constrained) device. The linkage is a sloppy hinge permitting about 7-10 degrees of varus-valgus and rotational laxity, which matters here because it lets load be shared with the surrounding soft tissues rather than concentrated at the bone-cement interface - the very interface that has already failed once.

Getting the Old Cement Out

Cement extraction, not implantation, is the technical crux of the revision and the step that generates most of the bone loss and most of the periprosthetic fractures.

• The canals are narrow, curved and thin-walled, and the ulna especially offers very little margin between a cement lump and the cortex. Perforation and iatrogenic fracture are made here, not at the moment of implantation.
• Work under direct vision as far as possible, using long, fine osteotomes and cement splitters down the canal rather than blind reaming, and take the cement out in fragments from the inside rather than levering against the endosteum.
• Ultrasonic cement removal tools allow deep cement to be softened and withdrawn along the canal with less cortical damage than mechanical instruments, and are the standard aid where they are available.
• A controlled cortical window or extended osteotomy is preferable to an uncontrolled perforation when distal cement cannot be reached - plan it, place it where it can be plated or strut-grafted, and repair it.
• Take deep samples before any of this if there is the slightest question of infection, and treat the case as potentially infected until they are negative.