The os trigonum is an inconsistently present accessory bone of the foot situated at the posterolateral aspect of the talus. It appears between the ages of eight and eleven years as a secondary center of ossification and usually fuses to the talus within one year after its appearance
1 . When this ossification center remains separate from the talus it is referred to as the os trigonum. According to Sarrafian
2 , the prevalence of this ossicle ranges from 1.7% to 7.7%. When fusion does occur and there is a large intact posterolateral talar process, it is referred to as a fused os trigonum
3 , a Stieda process
4 , or a trigonal process
2 . Since Rosenmuller first described the os trigonum in 1804
4 , there has been controversy concerning its origin. Steida
1 and Turner
5 believed it to be a secondary ossification center of the talus, whereas Shepherd
6 and Moullin
1 stated that it is an ununited fracture of the posterolateral talar process.
The os trigonum usually remains asymptomatic
4,7,8 , but an otherwise normal os trigonum can become symptomatic during or after strenuous physical activities
4,7-9 or following an acute injury to the ankle
10 . A symptomatic os trigonum, which is characterized mainly by posterior ankle pain, has been given many different names, including talar compression syndrome
4,9 , posterior ankle block
7 , posterior ankle impingement
11 , and os trigonum syndrome
12-14 .
The purpose of this study was to present the clinical results of excision of a symptomatic os trigonum through a posterolateral approach and to identify several factors affecting the clinical outcome.
In a five-year period from 1994 through 1999, during which time approximately 17,000 patients with foot and ankle problems were examined in our practice, the senior ones of us (M.N. and E.L.) operated on forty-four patients with os trigonum syndrome. In all cases, the os trigonum was excised through a posterolateral approach. Three patients were lost to follow-up. Of the remaining forty-one, twenty-three had involvement of the right ankle and eighteen had involvement of the left. There were thirty-three male patients and eight female patients, with a mean age at the operation of twenty-seven years (range, sixteen to forty-seven years). The average duration of postoperative follow-up was forty-four months (range, twenty-four to eighty-one months).
In all cases, the os trigonum syndrome was diagnosed on the basis of the history, physical examination, and radiographs (
Fig. 1 ). Thirty-five patients had a technetium bone scan (
Fig. 2 ) and thirty-eight had a computed tomography scan to confirm the diagnosis, evaluate degenerative changes in the region, and determine the precise relationship between the os trigonum and the talus (
Fig. 3 ). In all cases, the indication for surgical excision was persistent, moderate, or severe pain in the posterior part of the ankle despite a minimum of three months of nonoperative treatment. The mean time from the onset of symptoms to the surgery was seventeen months (range, four to fifty-two months).
The efficacy of the surgical treatment was assessed with the Ankle-Hindfoot Scale developed by the American Orthopaedic Foot and Ankle Society (AOFAS) in 1994
15 . The scale assigns 40 points for pain, 50 points for function, and 10 points for alignment. A perfect score of 100 points means that the patient has no pain (40 points), no limitation of either occupational or recreational activities and no need of any assistive devices for walking (10 points), the ability to walk more than six blocks (5 points) and on any walking surface (5 points), no limp (8 points), a full range of ankle and hindfoot motion (flexion plus extension, 8 points; inversion plus eversion, 6 points), no ankle or hindfoot instability (8 points), and good alignment (10 points). To calculate the postoperative AOFAS score, each patient was asked about his or her current ankle function and pain and was examined by one of us, who measured the range of motion and evaluated ankle stability and alignment. A preoperative AOFAS score was estimated retrospectively on the basis of information in the medical files and the patient's recollection. An improvement score was derived as the difference between the postoperative score and the estimated preoperative score. The period until full recovery was defined as the duration of time after the surgery until maximum relief of ankle pain and maximum ankle function were achieved.
The effect of several factors on the clinical outcome was assessed with use of a patient questionnaire that included questions regarding sports activities before the onset of symptoms, how the ankle pain started (acute trauma or chronic onset), participation in postoperative physical therapy, Workers' Compensation status, and whether the patient was satisfied with the surgical result.
Statistical Evaluation
The mean and standard deviation were calculated for the measured quantitative variables. To examine the effect of the preoperative duration of symptoms on the clinical outcome we divided the patients into two groups: those for whom the duration of symptoms was in the upper quartile (between thirty and fifty-two months) and those for whom it was in the remaining three quartiles (between four and twenty-four months). To examine the effect of age on the clinical outcome, we divided the patients into two groups according to the median. Nonpaired t tests were used to examine the influence of the independent variables on the postoperative score and on the duration of the period until full recovery (parametric univariate analysis). Nonparametric univariate analysis was performed with use of Mann-Whitney U tests. Two separate multiple regression analyses with forward and backward selection methods were performed to identify which of the independent variables was significantly related to the postoperative score and the time to full recovery (multivariate analysis). P values of <0.05 were considered significant.
Operative Technique
With use of local anesthesia, a high ankle tourniquet, and the patient lying on the unaffected side, a posterolateral incision is made just posterior to the peroneal tendons. The skin incision starts about 4 cm proximal to the lateral malleolus and ends 1 cm distal to it. Blunt dissection is used to avoid harm to the sural nerve. Usually the sural nerve is located just posterior to the incision, and when it is identified it should be dissected carefully and retracted posteriorly. The continuation of the fascia of the deep posterior compartment is then opened, and the retrocalcaneal fat is bluntly dissected until the os trigonum can be palpated with the foot held in dorsiflexion. The posterior talofibular ligament is sharply exposed along with the posterior aspects of the ankle and subtalar joint capsules. The whole os trigonum is exposed with additional blunt dissection, and it is then removed with sharp dissection if it is a separate bone or with an osteotome if it remains partially fused to the talus. After removal of the os trigonum, the flexor hallucis longus is inspected, and it is released in the area adjacent to the excised ossicle if there is any evidence of stenosing tenosynovitis. The wound is then irrigated and is closed in layers.
Postoperatively, the leg is immobilized in a non-weight-bearing cast for two weeks. Patients then perform supervised stretching and range-of-motion exercises, and they begin full unprotected weight-bearing as tolerated. No brace or other type of support is used after the cast is removed.
Twenty-one patients had the onset of symptoms during military service. An additional five—two police officers, one laborer, one driver, and one electrician—sustained a work-related injury. There were ten professional athletes, including four soccer players, two runners, two ballet dancers, one basketball player, and one judo expert. Thirty patients reported that the symptoms had begun immediately following an acute injury to the ankle; these injuries included twenty-one ankle sprains, six ankle fractures, and three direct injuries to the hindfoot. Eleven patients reported a chronic onset of symptoms; eight of them recalled at least one previous ankle sprain. With the numbers available, no significant difference in clinical outcome could be detected between patients with an acute injury and those with a chronic mechanism of injury (see Appendix).
Imaging Findings
Preoperative lateral radiographs were evaluated preoperatively and postoperatively and new postoperative radiographs were made and evaluated at the time of the latest follow-up in all forty-one cases. Thirty-five patients had a distinctly separate os trigonum, whereas it was difficult to identify a distinctly separate ossicle in the other six patients. Of the thirty-five patients who had a bone scan, thirty-two demonstrated increased uptake in the posterior part of the talus compared with the uptake on the contralateral side (
Fig. 2 ). The preoperative computed tomography scans, performed for thirty-eight patients, demonstrated a completely separate os trigonum in twenty-eight and an ossicle with a partial connection to the talus in ten (
Fig. 4 ). The computed tomography scans demonstrated degenerative changes, almost none of which were seen on the lateral plain radiographs, in nineteen patients. These changes were seen in the talus in ten patients, in the os trigonum in nine, and in both in seven. The changes also extended into the subtalar joint in four patients.
Operative Findings
At surgery, a separate os trigonum was identified and excised in thirty-one patients. In the remaining ten patients, a partially fused ossicle was separated from the talus with an osteotome. A sufficient amount of the lateral tubercle of the posterior process of the talus was removed to eliminate any block to ankle plantar flexion. After excision of the os trigonum, the flexor hallucis longus tendon was inspected in all patients and showed signs of stenosing tenosynovitis requiring tenolysis in two. One loose body was found and was removed from the interval between the os trigonum and the talus.
Clinical Outcome
The postoperative AOFAS scores of the forty-one patients ranged from 52 to 100 points, with an average (and standard deviation) of 87.6 ± 11.4 points. The patients lost points mainly because of mild ankle pain and decreased recreational function. The average score was 31.5 ± 8.5 of 40 available points for pain, 25.1 ± 3.7 of 28 available points for subjective ankle-hindfoot function, and 31.1 ± 2.2 of 32 available points for the range of motion of the ankle and hindfoot and the stability and alignment of the ankle as evaluated with objective physical examination. The estimated preoperative AOFAS scores ranged from 28 to 77 points and averaged 51.7 ± 15 points, which resulted in an estimated average improvement of 35.9 ± 18.5 points. This improvement was mainly due to increased ratings for subjective ankle pain and function; the range of motion of the ankle and hindfoot and the stability and alignment of the ankle did not seem to be changed much by the surgery. Compared with the preoperative rating, thirty-seven patients had an increase in the postoperative rating, of 15 to 64 points; three patients had equal preoperative and postoperative scores; and one patient had a 9-point decrease in the postoperative score. The last four patients were the only ones who stated that they were dissatisfied with the result of the operation. The duration until full recovery averaged 5 ± 3 months (range, one to twelve months).
The thirty-three patients with symptoms for two years or less prior to the surgery had an average postoperative score of 90 ± 10 points compared with 78 ± 12.4 points for the eight patients who had had preoperative symptoms for more than two years (p = 0.006 according to the t test, and p = 0.011 according to the Mann-Whitney U test). Multiple regression analysis showed that this variable influenced the postoperative score independently (p = 0.011) (see Appendix).
Before the surgery, the patients participated in sports activities (mostly soccer, running, basketball, and dancing) for an average of 10.7 ± 9.2 hours per week. The ten professional athletes and the eight amateur athletes, who were training for an average of two hours or more per day, had an average postoperative score of 91.6 ± 9.1 points compared with 84.5 ± 12.2 points for the remaining patients (p = 0.047 according to the t test, and p = 0.072 according to the Mann-Whitney U test). The twenty patients who were more than twenty-three years old at the time of the operation had an average postoperative score of 84.2 ± 11.4 points compared with 90.9 ± 10.7 points for the younger patients (p = 0.059 according to the t test, and p = 0.029 according to the Mann-Whitney U test). However, the number of patients was not large enough for us to detect significant differences between these groups with use of multiple regression analysis.
The time until full recovery averaged 5.7 ± 3.2 months for the patients who were involved in a Workers' Compensation claim compared with 3.8 ± 2.7 months for those who were not (p = 0.053 according to the t test, and p = 0.03 according to the Mann-Whitney U test). Multiple regression analysis revealed that the existence of a Workers' Compensation claim influenced the duration until full recovery independently (p = 0.048). The number of patients in the study was not large enough for us to detect a significant relationship between the clinical outcome and participation in postoperative physical therapy or the findings on the preoperative computed tomography scans.
Complications
In two patients, the sural nerve was transected intraoperatively, resulting in a permanent sensory neurapraxia. There were seven early postoperative complications, including four cases of transient sural nerve neuropraxia that resolved spontaneously within six months, two additional cases of permanent sural nerve neuropraxia, and one superficial wound infection that required local treatment and resolved within one week. The one late complication was a case of reflex sympathetic dystrophy, with partial resolution of symptoms by one year after the operation and then no further improvement by twenty-seven months.
When conservative treatment fails to decrease the symptoms of os trigonum syndrome, surgical excision of the ossicle is the treatment of choice
1,12,13,16 . Previous studies of the results of excision of the os trigonum have not included the use of standardized rating systems to determine the outcome of this operation
4,8,10,11,14,17,18 . Moreover, to our knowledge, no study has addressed any factors that might influence the clinical outcome. In the present study, we used the AOFAS Ankle-Hindfoot Scale to evaluate the clinical results of excision through a posterolateral approach in forty-one patients and we examined the effect of several factors on the outcome of this procedure.
The average postoperative AOFAS score in the present study was 87.6 points. This outcome is comparable with the average postoperative score of 86.4 points for eleven patients treated with arthroscopic excision in the clinical series reported by Marumoto and Ferkel
19 . We believe that these clinical outcomes are highly satisfactory and support a recommendation for surgical excision of a symptomatic os trigonum. The time to full recovery averaged five months and ranged from one to twelve months in our study. This finding is compatible with those in other clinical series in which open techniques were used
4,8,11,17 .
In the past, os trigonum syndrome was described mainly in professional ballet dancers, soccer players, and runners
4,7-9,11,14,17 . Therefore, it is not surprising that the mechanism of injury in these studies was usually chronic and thought to be the result of repetitive impingement on the ossicle. Most of the participants in our study were not professional athletes or dancers, and 73% reported an acute injury preceding the onset of symptoms. This finding suggests that, in the general population, the predominant mechanism of injury is a fracture of a previously intact trigonal process or a disruption of a cartilaginous synchondrosis between the talus and the os trigonum
1,8,17 . An avulsion injury might also initiate or aggravate this condition
13,16,18 . Our findings are supported by those of Veazey et al.
10 , who reported an acute mechanism of injury in all of nine patients in a clinical series that included only two athletes.
We found that the postoperative score was significantly lower when the duration of symptoms prior to the surgery had been more than two years. Using multiple regression analysis, we found this to be the only variable that influenced the postoperative score independently. We believe that this finding supports the theory that hypertrophic capsulitis and accumulation of inflammatory tissue play a part in the development and persistence of os trigonum syndrome, as has been suggested previously
7,11,20 . Unlike the ossicle itself, which can be excised with relative ease, the inflammatory tissue extends locally and cannot always be completely débrided, thus contributing to postoperative symptoms in some patients.
Several series, including ours, have shown a relatively long mean duration of symptoms prior to surgery
11-13,18,19 . Thirteen of our patients were referred to us after having been treated for a different diagnosis for a long time. We agree with Lombardi et al.
21 and Veazey et al.
10 that misdiagnosis is common among these patients. We believe that a high level of suspicion and early diagnosis and treatment of os trigonum syndrome are important not only to relieve current pain but also to improve the prognosis.
In our series, the clinical outcome was poor in four patients, all of whom had had the onset of the original symptoms immediately following an ankle sprain. In one of these patients, reflex sympathetic dystrophy developed postoperatively, resulting in a postoperative AOFAS score of 61 points compared with an estimated preoperative score of 70 points. A second patient, in whom the posterior ankle pain and tenderness had disappeared completely after the surgery, had a final postoperative score of 52 points due to a second area of pain in the sinus tarsi. The AOFAS Ankle-Hindfoot Scale does not discriminate among specific areas of pain in the foot and ankle, and this prevents accurate evaluation of patients with multiple painful lesions or injuries in the foot and ankle.
There was no apparent explanation for the persistent symptoms in the remaining two patients who had a poor outcome. Both patients had had a very long duration of symptoms prior to the surgery (fifty-two and forty-four months), were relatively older (thirty-eight and forty-two years old), and did not participate in vigorous sports activity. All three of these factors were found to contribute to a poorer outcome in our study; therefore, in retrospect, we think that we should not have operated on these two patients, especially in light of their relatively high estimated preoperative scores (71 and 75 points). We will recommend a longer term of conservative treatment as an alternative to surgery for similar patients in the future.
The main limitation of this study is the retrospective determination of the preoperative scores. Toolan et al.
22 found a mean difference of 5.3 points between preoperative AOFAS scores determined prospectively before elective foot and ankle surgery and preoperative AOFAS scores determined retrospectively after the surgery. They concluded that patients tend to recall a worse preoperative status than they originally reported. Dawson et al.
23 also found an overestimation of treatment effect in patients with low-back pain when a retrospective method was used to evaluate the clinical outcome. Thus, recall bias limits the value of the estimated preoperative and improvement scores in our study, which is why we included only postoperative scores in the statistical analysis. However, we do not believe that the large gap between the average preoperative and postoperative scores in our study (36 points) can be attributed only to a recall bias; we think that it indicates a substantial clinical benefit from the surgery itself. A prospective study could show this benefit more definitively.
The only factor that was found to influence the time to full recovery in our study was involvement in a Workers' Compensation claim. Compensation issues have been shown to influence both the outcome and the recovery period after other orthopaedic procedures
24-26 . The treating physician should be aware of this issue and expect a longer recovery period for patients involved in such claims.
Bone-scanning can effectively localize osseous injuries in and around the talus
10,27,28 . A bone scan was performed on thirty-five patients in our study and revealed increased uptake in the posterior talar region in thirty-two. The remaining three patients had a characteristic history and findings on physical examination and radiographs consistent with a symptomatic os trigonum. The postoperative scores for these three patients ranged from 90 to 100 points, contradicting previous assumptions that a normal bone scan virtually eliminates the diagnosis of os trigonum syndrome
18,29 .
We recommend that a computed tomography scan be performed after the diagnosis of os trigonum syndrome has been established, conservative treatments have failed, and surgical excision is considered to be indicated. A computed tomography scan enables the surgeon to determine the exact location, size, and shape of the ossicle and thus to adjust the extent of the surgical incision. In addition, the os trigonum was noted to be partially connected to the posterior process of the talus in ten of our thirty-eight patients evaluated with computed tomography (
Fig. 4 ). Even with reinspection, it is very difficult to detect this feature on a lateral radiograph
1 . The computed tomography scan also revealed degenerative changes, mainly in the talus, in half of our patients (
Fig. 3 ); this is another feature that is very difficult to visualize on lateral radiographs
19 .
Stenosing tenosynovitis of the flexor hallucis longus tendon may coexist with os trigonum syndrome, usually when the condition is chronic, as in ballet dancers
8,12,14 . Intraoperative signs of this condition were found in two patients in our study, and tenolysis was performed in both without difficulty. While this finding was expected in one patient, a ballet dancer, the other patient was a twenty-one-year-old woman with no history of participation in sports activity in whom the symptoms had started immediately after an ankle sprain. In contrast to other reports, we did not have any difficulty performing the tenolysis through the posterolateral approach
8,11,14 . Therefore, we think that it is appropriate to examine the flexor hallucis longus intraoperatively and to decompress it if there are signs of tenosynovitis.
The surgical approach used to excise the ossicle is controversial. Like others, we believe that the posterolateral approach is easier and safer than the medial approach
4,9 . There were no cases of peroneal tendinitis or injury of the posterior tibial nerve, as has been described in the past
7,17 , in our series. In our opinion, the only disadvantage of the posterolateral approach is the relatively high number of sural nerve complications reported here and by others
10 . The eight cases of sensory neuropraxia of the sural nerve (four of which were permanent) in our patients resulted in a sensory deficit in a small region of the lateral superior part of the ankle. We make the surgical incision close to the peroneal tendons so that it lies anterior to the sural nerve
30 , and this minimizes the risk of injury to the branches that run posteriorly to supply the skin. We recommend that a distinct effort be made to identify, retract, and protect the sural nerve when the posterolateral approach is used.
A subtalar arthroscopic technique for excision of the os trigonum has also been described
19,21 . The main advantage of that approach is a shorter recovery period. The disadvantages are that the procedure takes longer, requires a high degree of dexterity, and is not recommended for surgeons who perform arthroscopy only occasionally
19 . Our study demonstrated that the clinical outcome is similar between open and arthroscopic approaches.
In conclusion, an os trigonum that has remained symptomatic after a minimum three-month trial of conservative treatment can be treated effectively by surgical excision through a posterolateral approach. The outcome is favorable, although sural nerve injury can occur.
A table showing the parameters affecting the clinical outcome is available with the electronic versions of this article, on our web site at www.jbjs.org (go to the article citation and click on "Supplementary Material") and on our quarterly CD-ROM (call our subscription department, at 781-449-9780, to order the CD-ROM).