I would like to commend this article by Brown et al.; it is well written, well documented, and realistic in its analysis of extensor tendon allografts, which have become the final common pathway for extensor mechanism ruptures after total knee replacement. Although Brown et al. suggest that enthusiasm for this procedure, no matter how desperate the situation, should be modest or guarded, their data are irrefutable and consistent with my own experience of almost 250 extensor tendon allografts over the past twenty years. There is no question that the explosion of joint replacement, particularly in younger patients, will create the need for more and definitely better alternatives to this problem.
After the initial identification of an extensor mechanism rupture and the angst that every surgeon feels at that moment, it is important to try to ascertain the cause. Early ruptures are frequently due to compromised replacements that are either unstable or stiff after index total knee arthroplasty. My own experience, as well as that of the authors, is that almost 50% of these patients will need a simultaneous knee revision as well as an extensor allograft reconstruction. Many of these individuals have an infection and are undergoing the same procedure (often after an antibiotic spacer block), and they are at even greater risk for long-term infection and failure. These ruptures usually occur fairly early in the recuperative period, whereas extensor ruptures from falls or attenuation of the tendon from disease processes such as rheumatoid arthritis may present years later. The surgeon must be ready to perform a revision, probably to a constrained condylar knee or hinged device with stems to protect the subsequent and often simultaneous allograft.
Almost all surgeons faced with a primary extensor mechanism rupture will attempt a primary repair. However, for several decades, the literature has shown the results of this approach to have been largely ineffectual, often with success rates of only ≤10%. Achilles tendon allografts, iliotibial-band or semitendinosus grafts, and the newest synthetic mesh techniques all have substantial early and late failure rates. Bracing is rarely tolerated for any length of time by patients, who often think that they may be better off with a fusion or an amputation. The authors’ 4% amputation rate is certainly in keeping with that of the general experience in the orthopaedic community.
As a certain percentage of these extensor mechanism ruptures are related to or are difficult to distinguish from infected knees, this potential problem must be examined with all of the available technology: erythrocyte sedimentation rate, C-reactive protein, bone scan, and alpha-defensin levels, if they are available. To perform a primary exchange in an infected arthroplasty with the supplemental use of the nonviable allograft tissue is only to invite persistent deep infection.
A great deal of discussion goes into the fixation techniques available for the distal and proximal aspects of the graft. The choice of the patellar relationship to the joint line is usually fairly simple, as the grafts can be placed to accommodate that level by the surgeon. In my experience, the distal portion of the graft almost uniformly heals without difficulty. I have espoused and have utilized the dovetail technique of sculpting a trough in the anterior tibial tuberosity, slightly to the medial side of the knee, into which a graft can be wedged very securely. These grafts have usually been frozen, and screws run the risk of fracture of this brittle bone and are often difficult to place to avoid the tibial stem. The tibial stem should bypass the graft site if possible. Not only does it serves to eliminate the risk of fracture, but it is also very convenient to place the wires that secure the graft around the posterior part of the stem or even around the posterior part of the tibia, although the latter is done with some trepidation. If carefully fitted and appropriately secured, my experience has been approximately 98% success in healing of this distal portion of the allograft.
The technique for proximal fixation has been studied extensively, and it is quite clear that these grafts need to be inserted as tightly as possible with the limb in full extension; otherwise, they are almost uniformly doomed to failure. My approach is to use a Krackow suture through at least 10 or 15 cm of proximal tendon with a nonabsorbable suture and to have an assistant hold the tendon in full extension with considerable force. That approach then allows vertical mattress sutures to be placed between the medial and lateral margins of the residual tendon, if they exist. If the part of the suture penetrating the tendon is slightly distal to that of the host tendon, then the graft is tightened further with each stitch. Finally, if indeed there is sufficient residual tissue anteriorly, it should not be trimmed but rather cabled with nonabsorbable sutures to act as yet another anterior tendon buttress. The tension suture at the most proximal end of the incision is tightened, knotted, and secured only after all of this has been completed.
The postoperative regimen must be nonnegotiable. I prefer casts to braces to avoid accidental flexing of the knee of any degree for the first six weeks. Thereafter, a brace with 30° of increased flexion every two weeks for another six weeks seems appropriate. One has to strike a balance between a fully ingrown tendon in a stiff knee compared with a loose tendon and incomplete incorporation of the graft by the host. I believe that no weights or resistive exercises should be used in the postoperative physical therapy regimen for a minimum of six months to a year after brace removal.
I agree with the authors that the initial results are always remarkable and very gratifying but do indeed tend to deteriorate over time. In my experience, the long-term flexion contractures are usually <30°, and patients can still arise from a seated position and mount and descend stairs safely. However, for this to occur, the patient must obtain full passive extension and perhaps even a small amount of hyperextension passively in the knee to allow it to be locked out in full extension before standing and walking. Any compromise with this protocol or misunderstanding between a well-meaning physical therapist and the specific written orders is likely to lead to early failure of the allograft. Although Brown et al. are understandably disappointed in their 38% failure rate, this rate pales before the almost 90% failure rate of a direct primary repair. On average, these failures appeared at about two years, which is my experience as well. However, most of the patients, even those with a notable extensor lag, largely do not have pain because of the denervated graft and are generally more content with their result than the surgeon may be.
The statistics certainly beg the question as to whether extensor mechanism ruptures after total knee arthroplasty should be treated with primary repair techniques or whether one should go directly to an extensor allograft, with better short and long-term success rates. The latter is not currently the general practice, although it would seem reasonable to offer this option to the patient and to share the probable outcomes as well. Patients and physicians alike should be aware of the negligible long-term clinical results, regardless of the patients’ satisfaction that they are at least walking and have avoided a fusion or amputation.
Because the revision rate is now rising three times faster than the use of total knee arthroplasty in the general population, the need to manage extensor mechanism ruptures in a more efficacious and successful manner should be promoted. This article is one with which every knee arthroplasty surgeon should be acquainted.