SEPTIC FAILURE

This is mentioned here to emphasize the need to exclude a septic process as the cause of implant failure. A high index of suspicion must be developed. The value of preoperative screening as with sedimentation rates and C-reactive protein (CRP) is well accepted. One recent study demonstrated a sedimentation rate over 23 mm/hr and a CRP in excess of 13.5 mg/L correlates with infection with a specificity and sensitivity of 85% and 90% respectively.¹¹ Radiolucency of the bone-cement interface may be due to either mechanical failure or sepsis, but implant loosening is not usually associated with a septic joint unless the process is a chronic one.²¹ Infection of a well-fixed device is particularly bothersome, because removing the prosthesis may result in fracture of bone that poses major reconstructive problems for the surgeon, but commonly sepsis is managed with a staged reimplantation. The important consideration is to avoid excessive use of bone graft in this setting. This topic was reviewed recently by Cheung and colleagues, and is discussed in detail in Chapter 62.³

DEVICE FAILURE

Device failure may involve the stem or the articular coupling elements of the device. High-density polyethylene articular components do have a tendency to wear,¹⁵ but this typically does not cause loosening of the implant. Because the semiconstrained implant is being used for broader indications and loosening is much less of a problem, articular wear is increasingly being recognized over long-term follow-up.^10,15 Lee et al¹⁵ assessed the experience of almost 900 linked Coonrad/Morrey devices. Only 1.5% required reoperation for isolated bushing wear. The authors make the distinction between bushing wear that is managed by simply exchanging the bushing and loosening that typically occurs as a result of loosening and osteolysis that was seen principally in the precoat ulnar component design. This design has not been used for more than 7 years (Fig. 65-1).

FIGURE 65-1 Osteolysis around the Coonrad/Morrey “precoat” ulnar surface component (A and B). This type of loosening is not due to bushing wear, as some have suggested.

Implant fracture is an uncommon complication that usually follows excessive activity or a single significant trauma. This has occurred with the Coonrad-Morrey device owing to the stress-riser effect of the surface treatment of the titanium implant and in patients with excessive physical demands (Fig. 65-2).²⁵

FIGURE 65-2 Fracture of the ulnar component of a Mayo modified Coonrad device removed from a patient who repeatedly lifted more than 50 pounds

Athwal et al¹ reviewed the Mayo experience with fractured components. The technique of removing well-fixed stems and reinsertion into an expansion of the cement mantle was successful in about 80% of patients.

This failure mode has prompted the introduction of the stronger precoat, which, owing to the possibility of osteolysis, has been subsequently replaced in 2000 with the current plasma spray surface.

INSTABILITY

With unlinked prostheses, instability is not uncommon (see Chapter 61). Delayed or chronic subluxation is associated with excessive activity, improper initial balance, or implant wear.^7,21

If acute instability is due to ligament insufficiency or improper “balance” and does not respond to a period of immobilization, revision with a linked implant should be considered. If early dislocation is due to malpositioning of the components, immediate revision should be performed. In questionable cases, examination under fluoroscopy may show the cause of the instability and allow the management strategy to be defined (Fig. 65-3).

FIGURE 65-3 Failure of the articulation due to instability of a resurfacing implant, as seen on lateral (left) and anteroposterior (middle) radiographs. The 10-year revision is asymptomatic (right).

The selection of unlinked implant has been predicated to some extent on the assumption that a reoperation or revision would prove more reliable after this type of design. Until recently, this hypothesis has not been proven. A review of Mayo’s 30-year experience with linked and unlinked implants appears to confirm this assumption but only if the revision was from an unlinked to a linked device (Fig. 65-4).¹⁶ Specifically, reoperation of all unlinked implants resulted in a more predictable outcome than did revision after all failed linked implants. However, the linked revisions were to a great extent in patients with post-traumatic arthritis. There was no difference in success of revision when only the rheumatoid patient was considered. Furthermore, it was also demonstrated that the need to revise the unlinked implant occurred at a significantly greater frequency than did the need to reoperate on a linked device (Fig. 65-5).¹⁶

FIGURE 65-4 Kaplan-Meier curves of Mayo 30-year experience showing good survival with revision of a variety of unlinked designs to the linked Coonrad/Morrey implant (A). Note the very poor outcome when the unlinked was revised to another unlinked design (B).

(From Levy, J. C., and Morrey, B. F.: A survivorship of unlinked and linked total elbow arthroplasty for rheumatoid arthritis: A comparative study. J. Bone Joint Surg. Am. [in press].)

FIGURE 65-5 Kaplan-Meier curve showing markedly superior performance of the linked (A) compared with the unlinked device (B).

(From Levy, J. C., and Morrey, B. F.: A survivorship of unlinked and linked total elbow arthroplasty for rheumatoid arthritis: A comparative study. J. Bone Joint Surg. Am. [in press].)

LOOSENING

The early experience with total elbow arthroplasty demonstrated loosening of the humeral component about twice as often as loosening of the ulnar component.^9,12,21 This difference is not observed with the Coonrad-Morrey device because loosening currently occurs more often at the ulna than at the humerus.^4,10,16 Pain is a typical but not a universal feature. Different expressions of failure are recognized. One is a loose implant with pain but few or no x-ray changes and no bone resorption (Fig. 65-6). The other is gross instability and obvious loosening due to bone destruction, with minimal or no pain. Particulate debris results in cortical thinning or “ballooning” of the humerus or the ulna. Resorption of a significant amount of bone can pose major problems with the reconstructive procedure, especially when the bone loss is complicated by fracture (Fig. 65-7). The fracture may be “silent” due to chronic thinning or acute. The latter fracture is not caused by loosening and may heal without a need for implant revision. The topic of periprosthetic fracture is dealt with in detail in Chapter 66 because strut graft is often employed in its treatment.

FIGURE 65-6 The patient developed pain 18 years after a Dee total elbow arthroplasty. The radiograph did not suggest loosening of the bone-cement interface; however, at revision gross loosening was observed at the humeral prosthesis-cement interface

FIGURE 65-7 Because the grossly loose Pritchard-Walker II implant was not painful, the patient continued to use it, but marked osseous resorption was due to the foreign body wear