We did a retrospective review of all patients who had an arthroplasty of the proximal interphalangeal joint with a pyrolytic carbon implant performed before October 2007 by a single surgeon (P.J.S.). The indication for this procedure was pain that was unresponsive to nonsurgical management and, in some cases, patients who declined alternative procedures such as arthrodesis or silicone arthroplasty. Inclusion criteria were a diagnosis of primary osteoarthritis of the proximal interphalangeal joint in the hand and a follow-up period of greater than two years. Patients with severe deformity were not considered for treatment with implantation; thus, all patients in this series had preoperative coronal angulation of <20°, adequate bone stock, and intact collateral ligaments. Also, patients with inflammatory arthropathy and posttraumatic arthritis were excluded. Pyrolytic carbon resurfacing arthroplasty had been performed in thirty-one proximal interphalangeal joints in the seventeen patients who met the inclusion criteria. The study was approved by the local institutional review board. In all but two patients, evaluations were carried out by an independent observer (T.M.S.) who was not otherwise involved in the care of these patients.
The average age at the time of the operation was sixty-four years (range, fifty-two to seventy-nine years), and seventeen of eighteen patients (thirty of thirty-one joints) were women. Implants were placed in seven index, fourteen long, eight ring, and two small fingers. Two patients, one with five and one with four implants, had arthroplasty bilaterally. Three patients received three implants unilaterally, one received two implants unilaterally, and the others received one implant.
Technique and Rehabilitation
The technique of the senior author has been described in detail previously22,23. The joint was approached through a dorsal longitudinal incision. Early in this series, the joint was exposed by creating a distally based triangular flap of the dorsal apparatus24 and later the central tendon was split longitudinally.
After surgery, the joint was splinted in extension. At seven days postoperatively, active motion of the proximal interphalangeal joint was initiated and, when the joint was not being exercised, the finger was splinted in extension. Active range of motion of the metacarpophalangeal and distal interphalangeal joints was encouraged. At week 3, active flexion of the proximal interphalangeal joint was initiated and, from weeks 3 to 6, a dynamic proximal interphalangeal joint extension splint was applied. Active range of motion of the metacarpophalangeal and distal interphalangeal joints was continued. A proximal interphalangeal joint extension splint was worn at night. Full tendon gliding and composite fist exercises were initiated at six weeks, passive range of motion was added at six to eight weeks, and light strengthening, at eight to ten weeks. At ten to twelve weeks, therapy was usually complete and the patient was discharged to a home program with an extension splint worn at night for three to six months.
Evaluation
Radiographs were made at the time of the final follow-up for sixteen of seventeen patients. One patient accepted examination but declined radiographs. This patient had received a revision silicone replacement, and prerevision radiographs were available. Preoperative radiographs were also available for all but one patient, and immediate postoperative radiographs were available for all patients.
Few standards exist for the radiographic evaluation of small joint arthroplasty implants20. We chose to measure implant subsidence on lateral radiographs, in linear terms, comparing immediate postoperative radiographs with those made at the time of the final follow-up. Distances were measured digitally, and implant subsidence was calculated as shown in Figure 1.
Each proximal interphalangeal implant has a radiopaque core coated with 0.5 mm of radiolucent carbon, so periprosthetic lucency is expected. The manufacturer states in the company literature that lucency of >0.5 mm and of regular thickness may be normal25. Thus, only implants that exhibited irregular periprosthetic lucency of >0.5 mm at the time of the final follow-up were considered loose by radiographic criterion. We created a grading system to assess dorsal, volar, and/or coronal migration (flexion, extension, and/or translation in anteroposterior or lateral radiographs) of each implant within the medullary canal, using a 4-point scale, with 0 indicating a well-aligned implant; 1 point, macroscopically evident migration; 2 points, severe migration with the implant opposing the cortex; and 3 points, a breach of the cortex by the implant (Fig. 2).
Active motion of the proximal interphalangeal joint was assessed with use of a handheld goniometer. Arc of motion at the time of the final follow-up was measured for all patients; however, for statistical analysis, the immediate prerevision range of motion was used for the five patients who underwent revision procedures. Complications were defined as dislocation, contracture (defined as any joint with an arc of motion of <35° or with a loss of motion of ≥50%)22, loosening, squeaking, intraoperative fracture, wound-healing problems, and infection.
All patients completed subjective measures documenting pain, satisfaction, and function at the time of the final follow-up. These measures included the Michigan Hand Outcomes Questionnaire (MHQ), a validated outcomes questionnaire developed specifically for the hand26. Hands were scored in six different domains: total outcome score, activities of daily living, pain, work performance, aesthetics, and satisfaction. Visual analog scale (VAS) scores for pain (ranging from 0, indicating no pain, to 10, indicating intolerable pain) were assessed for the twelve patients who had not undergone revision procedures. Satisfaction with the result was assessed for all patients with a Likert scale (ranging from 1 point, indicating very dissatisfied, to 5 points, indicating very satisfied). Patients were asked if they would have the procedure again, were satisfied with the appearance, or had experienced complications. All subjective evaluations were performed at the time of the final follow-up. No preoperative subjective evaluations were obtained.
The results were calculated with and without the digits that had an arthrodesis or revision (thirty-one digits in total and twenty-six when digits that had an arthrodesis or revision were not included). Inferential analysis was performed with use of paired Student t tests, which have been used in most studies of pyrolytic carbon resurfacing arthroplasty of the proximal interphalangeal joint15,17,19-21. A p value of 0.05 was used in each case.
Source of Funding
There was no external funding for this investigation.
The average follow-up period was fifty-five months (median, sixty-one months; range, twenty-five to eighty-five months). Seventeen (55%) of thirty-one joints were followed more than five years and seven joints (23%), more than six years. No patient was lost to follow-up.
The average range of motion for the entire sample was 57° (range, 15° to 95°) preoperatively, 67° (range, 30° to 90°) at the best measurement postoperatively, and 31° (range, 0° to 100°) at the time of the final follow-up. The postoperative best range of motion for the entire sample occurred at a mean of 11.6 months postoperatively (range, one to forty-seven months). The average range of motion was significantly improved at the best postoperative measurement compared with the preoperative measurement (p < 0.05), but it worsened significantly between the preoperative and final follow-up evaluations (p < 0.05).
When the five digits that required revision surgery were eliminated from the cohort that underwent statistical analysis, the average arc of motion in the remaining twenty-six joints was 56° (range, 20° to 90°) preoperatively, 66° (range, 40° to 90°) at the best postoperative measurement, and 34° (range, 0° to 100°) at the time of the final follow-up. The postoperative best range of motion (analyzed after eliminating the five digits that had arthrodesis or revision) occurred at a mean of 11.7 months (range, one to forty-seven months). The decline from the preoperative measurement to the final follow-up evaluation was significant (p < 0.05).
The VAS measurement of pain at rest averaged 3 (range, 0 to 7) of 10 at the time of the final follow-up for the entire cohort.
Implant subsidence averaged 1.9 mm (range, 0 to 7.6 mm) in the proximal phalanx and 0.9 mm (range, 0 to 3.4 mm) in the middle phalanx. Subsidence was minor in many patients and could be attributed to radiographic projection, but was severe in other patients. Twelve of thirty-one implants in the proximal phalanx subsided >2 mm, and five of thirty-one implants in the middle phalanx subsided >2 mm. One patient who was previously reported to have 2 mm of subsidence of the implant in the proximal phalanx of the index finger and 5 mm of subsidence in the implant in the proximal phalanx of the small finger23 had subsidence of 8 mm and 7 mm, respectively, at the time of the final follow-up.
Migration was defined as flexion, extension, or translation deformity in the anteroposterior or lateral plane. Twenty proximal and fourteen distal implants migrated perceptibly within the bone at the time of the final follow-up evaluation in the coronal or the sagittal plane, or both (Fig. 2, Table I). Migration was considered severe (grade 2 or greater) in seven proximal implants and three distal implants, and some joints had grade-2 migration in both the proximal and distal implant and/or in multiple planes. One implant was classified as grade 3; the stem had breached the dorsal cortex at the time of the final follow-up (Fig. 3).
Subjective Evaluation
Patient satisfaction for the entire sample averaged 3.4 (range, 1 to 5) on a 5-point Likert scale (ranging from 5, indicating very satisfied, to 1, indicating very dissatisfied). Five patients were very satisfied, five were satisfied, three were indifferent, and four were very dissatisfied. Patient ratings of the appearance (with 0 indicating poor and 10, excellent) averaged 7 (range, 1 to 10), and eleven of seventeen patients stated that the appearance was improved. Twelve of seventeen patients stated that they would have the operation again.
All patients completed the Michigan Hand Outcomes Questionnaire (MHQ) at the time of the final follow-up. The hand treated operatively was compared with the noninvolved hand for the fifteen patients who received arthroplasty unilaterally. The data showed significantly poorer status for the operatively treated hand in all categories except aesthetics (Table II).
Complications
There was a total of sixty complications in twenty-eight joints (Table III); only three joints were free of complications. The most striking complication was fracture of an implant in the proximal phalanx in one patient (Figs. 4-A and 4-B). This patient returned at our request more than six years after her most recent follow-up evaluation and did not know that implant fracture had occurred. Five implants (16%) had dislocated. At twenty-seven months postoperatively, one patient had dislocation of the implant and underwent arthrodesis of the proximal interphalangeal joint. Another patient had dislocations in two different joints; the first occurred eight days postoperatively in the index finger and the second, in the middle finger four years later. After reduction of the dislocation, the middle finger was stable, but the index finger later underwent an arthrodesis. As a result of severe implant subsidence and loosening, a third patient (Figs. 3, 5-A, and 5-B) had two proximal interphalangeal joint dislocations. These dislocations were chronic and resulted from progressively severe subluxation over time.
Four implants (13%) had progressively subluxated but had not dislocated at the time of the final follow-up. Contracture occurred in twenty of thirty-one joints. All twenty had an arc of motion <35°, and seventeen had a decrease of ≥50% from the preoperative range of motion. One patient had a fixed contracture of the proximal interphalangeal joint in the ring finger at 90°.
Fifteen (48%) of thirty-one implants were loose by radiographic criteria, and all fifteen exhibited at least 0.9 mm of radiolucency around the implant. All five patients who underwent arthrodesis or arthroplasty revision had grossly loose implants at the time of this surgery. Many implants developed progressively severe loosening over time.
Patients reported squeaking in eleven joints (35%) at some point postoperatively, and four joints exhibited persistent audible squeaking at the time of the final follow-up. Minor wound-healing problems occurred in two patients (two joints), and intraoperative phalangeal fracture occurred during implantation in two patients (two joints).
Six joints (19%; six patients) required reoperation (four had an arthrodesis, one had revision arthroplasty, and one had excision exostosis). One patient with a painful fixed contracture of the proximal interphalangeal joint underwent successful arthrodesis. Another patient requested arthrodesis for stiffness. In two patients, the joints dislocated repeatedly and underwent arthrodesis. Another patient, with a 40° flexion contracture, had a revision arthroplasty with a silicone implant. One patient had excision of exostosis, but the implant was left intact.
Arthroplasty of the proximal interphalangeal joint may preserve motion. Resurfacing arthroplasty with a pyrolytic carbon implant has the theoretical advantage of preserving stability because the collateral ligaments are preserved and the implant has a tongue-and-groove design27. Most series in the literature have described good results with this implant15-19; however, follow-up has been less than two years in all but one study15. Our study followed patients for an average of fifty-five months.
Previous investigations of arthroplasty of the proximal interphalangeal joint with pyrolytic carbon implants have not noted a significant or clinically important change in interphalangeal joint motion at the time of the final follow-up at an average of nineteen months15-23,28. In our series, the average arc of motion increased from the preoperative measurement to the best postoperative measurement (at an average of one year postoperatively), but as time progressed, motion and function declined. This was evident by a significant (p < 0.05) decrease in the average range of motion for all thirty-one joints from the preoperative evaluation (57°) to the time of the final follow-up (31°), and even for the twenty-six joints remaining after those needing revision were excluded from the analysis (average, 56° to 34°, respectively). Our finding of a significant decrease in motion at the time of the final follow-up contrasts with the reports in the literature that found no significant change in the range of motion at the time of the final follow-up. We believe that deterioration in the range of motion occurs after a longer follow-up period, and therefore decreases were not found in previous reports (see Appendix).
The clinical importance of the range of motion of the proximal interphalangeal joint is not well understood. One recent study found that patients with simulated arthrodesis of the proximal interphalangeal joint lacked deficits on Jebsen hand function tests but had difficulty with precision handling tasks29. In our study, many patients also had limited range of motion of the distal interphalangeal joint, which likely contributed to the clinical importance of the 26° loss of proximal interphalangeal motion.
We recorded numerous complications including implant fracture. Skie et al., in 2007, reported an intraoperative fracture of an implant30, and Daecke et al., in 2006, reported implant fracture in a rabbit12, but we found no report of a fracture of a pyrolytic carbon proximal interphalangeal implant in a human following surgery. This complication does not necessarily call into question the strength and excellent wear characteristics of pyrolytic carbon, which have been demonstrated in humans and animals13,31. It does, however, underscore the serious problems of loosening and migration that have occurred with pyrolytic carbon implants in the proximal interphalangeal joint.
Chung et al. cited dislocation as a concern, and three of twenty-one joints in their series had a dislocation17. Meier et al. reported two conversions to arthrodesis because of dislocation18. Bravo et al. reported one joint that was revised to amputation after developing so-called instability, but they did not report frank dislocation15. Five joints dislocated in three patients in our series: three acutely and two after implant subsidence and migration.
Many patients in our study had loosening at the time of the final follow-up. An examination of the available literature revealed that osseous "ongrowth" may provide only weak fixation, appositional fixation may not always occur, and loosening may be common with pyrolytic carbon implants. Pyrolytic carbon is known to have relatively poor attachment strength to bone11,32-35, and one early report emphasized that "appositional attachment" is one order of magnitude less than attachment by bone ingrowth11. The intimate contact necessary for appositional fixation also does not occur in all pyrolytic carbon implants. In a series studying four noncemented metacarpophalangeal implants in baboons, consistent direct appositional fixation (osseous ongrowth) was found in only one of four implants; in the other three metacarpal implants and all four phalangeal implants, there was evidence of a fibrous layer between bone and implant36, making direct appositional fixation impossible. Another investigation found an interposed fibrous layer in three of six prostheses implanted in baboon mandibles9. The fibrous layer was also found when pyrolytic carbon implants were studied in rabbit knees12 and beagle femora34.
In an in vivo rabbit model, all eight pyrolytic carbon proximal interphalangeal joint implants subsided and were grossly loose with poor implant-bone contact in the pyrolytic carbon group on histologic examination12. Our experience with loosening is not surprising given the poor attachment strength, inconsistent fixation, and loosening found in the literature. Migration within the bone has been emphasized by the authors of two studies. Herren et al. reported migration in nine of seventeen implants20, and Meier et al. reported migration in nine of twenty-four implants18. Bravo et al. noted "radiographic change in position" in twenty of fifty implants, but stated that none of these implants appeared to be loose and that this is a normal result of the interaction between the implant and bone15. The authors stated that osseous ongrowth causes movement of implants early after surgery, but movement decreases over time. Our findings do not confirm this observation as many of our implants did not settle into a stable position over time (see Appendix).
Excellent pain relief has been previously reported after arthroplasty with pyrolytic carbon implants. Bravo et al. reported that VAS scores improved, on the average, from 6 to 1 (on a scale of 0 [no pain] to 10 [intolerable pain])15. Wijk et al. stated that thirty-six of forty patients were pain-free at rest and eighteen were free of pain with activity19. Meier et al. reported average VAS scores of 0.9 at rest and 2.8 with weight-bearing (on a scale of 0 to 10)18, and Stütz et al. reported total pain relief at rest and with motion in 80% of the patients16.
In our series, average VAS scores were relatively low (3 of 10); however, the MHQ scores reflected more pain in the operatively treated hand than in the noninvolved hand. While most (ten) of the seventeen patients were satisfied, the MHQ satisfaction scores were lower for the operatively treated hand. Compared with the MHQ scores reported by Chung et al. (for a one-year follow-up period)17, our patients had lower scores in every MHQ category, but because there are no preoperative comparisons for pain relief and satisfaction, broad inferences cannot be drawn from the subjective data.
This study has limitations. First, it is a retrospective investigation and preoperative VAS scores, subjective questionnaires, and MHQ data were not available. Another limitation is the small number of proximal interphalangeal arthroplasties (thirty-one joints), although this study is the third largest series to date15,19. Additionally, the data for range of motion may not follow a normal distribution, and the Student t test may not have been the optimum choice to test significance. The t test, however, was used because it was the primary statistical instrument used in most other studies of pyrolytic carbon arthroplasty of the proximal interphalangeal joint15,19-21,26. The finding of lower subjective scores for the operatively treated hands may simply reflect the fact that those hands had more advanced osteoarthritis, and poorer subjective scores may be a result of coexisting arthritis in the ipsilateral hand. This comparison should not be relied on to make judgments about the effectiveness of the implant as a pain-relieving device. Finally, asking patients if they would repeat a procedure may be a biased question that does not necessarily reflect the effectiveness of the surgery. Patients may be hesitant to state they made the wrong decision or may be concerned about disappointing their caregiver.
A strength of this study is that no patient was lost to follow-up. An additional strength is that all patients had proximal interphalangeal joint osteoarthritis. Finally, to our knowledge, the average length of follow-up is the longest of any peer-reviewed study of this implant.
This series is a reminder that proximal interphalangeal joint arthritis is still an "unsolved problem."37 Although the literature includes positive evaluations of these implants15,16, others are less enthusiastic20,21.
Our results are also problematic. On the basis of our findings, with substantially decreased motion of the proximal interphalangeal joint, gross implant migration, implant loosening, and numerous complications, we no longer use this implant in our practice. Future versions of this implant may be able to address its current deficiencies.