Abstract
Objective. Intraarticular corticosteroid (IAC) injections are often used to treat temporomandibular joint (TMJ) arthritis associated with juvenile idiopathic arthritis (JIA). One potential complication of IA therapy is heterotopic bone formation (HBF). The purpose of our study was to evaluate risk factors for HBF development in children with JIA who received IA therapy for TMJ arthritis.
Methods. This was a retrospective study of children with JIA who had received ≥ 1 IAC injection into the TMJ. Survival regression analysis was performed to identify risk factors for the development of HBF.
Results. There were 238 children included, of whom 33 (14%) developed HBF. No cases of HBF were diagnosed prior to the initial injection. Univariate analysis revealed that the risk factors for development of HBF were the total number of injections received into the TMJ and age at diagnosis of JIA, while the length of time from diagnosis of JIA to the first injection was inversely associated with the risk of HBF formation. The total number of injections was no longer significant following adjusted survival models. Children with HBF had increased physical examination evidence of acute or chronic changes, namely decreased maximal incisal opening and increased likelihood of jaw deviation.
Conclusion. HBF within the TMJ is relatively common in patients with JIA receiving IAC injections for TMJ arthritis. Future prospective studies are required to delineate the risks posed by the injections themselves as opposed to the underlying disease activity, as well as to evaluate alternative forms of local therapy to the TMJ.
- JUVENILE ARTHRITIS
- TEMPOROMANDIBULAR JOINT
- INTRAARTICULAR INJECTIONS
- HETEROTOPIC OSSIFICATION
Juvenile idiopathic arthritis (JIA) is a chronic inflammatory disease of the joint lining tissue in childhood, with a prevalence in the United States of about 1 in 1000 children1. Based on magnetic resonance imaging (MRI), temporomandibular joint (TMJ) involvement occurs in 40–100% of cases2,3,4,5,6,7. Undiagnosed and untreated, the expected morbidity of TMJ arthritis in children with JIA includes micrognathia, malocclusion, facial dysmorphism, and chronic pain8,9. Because TMJ arthritis can be active and destructive despite aggressive systemic therapy (conventional and biologic disease-modifying antirheumatic drug therapy; DMARD) and otherwise quiescent arthritis6,10,11, intraarticular (IA) corticosteroid (IACS) therapy has been used to control TMJ arthritis10. Several studies have documented the benefits of this approach7,12,13,14, although it is clearly not curative13,14,15. Although these studies have not reported significant adverse events beyond transient local Cushingoid reactions or subcutaneous atrophy and hypopigmentation10,16,17, longterm safety concerns have been raised18,19. In particular, 1 concern is the risk of heterotopic bone formation (HBF), defined as the presence of bone in soft tissue, such as within or immediately outside the TMJ capsule20. This can occur even in peripheral joints, where periarticular or IA calcifications have been reported following IACS injections in 0.2–6% of injected joints21,22,23. The risk may be higher in the TMJ because of unique anatomical features, including its small space and IA ossification site24. A previous report documented the presence of HBF in 12 children with JIA who had undergone multiple injections with IACS, although it was unclear whether the primary risk factor in these subjects was the injections or the severity of the underlying arthritis that prompted local therapy20. Similarly, a second study reported IA ossifications in 21% of patients with JIA who had received IACS therapy to the TMJ25.
To evaluate this question further, clinical and imaging data were reviewed on a much larger cohort of 238 children with JIA who had undergone at least 1 round of IACS, with or without addition of IA infliximab (IAI)26. The subjects who developed HBF were compared to those who did not. This will help identify individuals at risk for HBF at the TMJ, as well as better define the pathophysiology of HBF in children with JIA.
MATERIALS AND METHODS
Subjects
This was a retrospective study of all children with JIA between the years 2009–2015 who were referred to a single oromaxillofacial (OMF) surgeon (PDW) for evaluation and management of TMJ arthritis, limited to those who underwent IACS injection of at least a single TMJ and had at least 1 followup MRI following the injection. All patients had been evaluated by a pediatric rheumatologist, and diagnosis of JIA was based upon the International League of Associations for Rheumatology criteria27. Demographic, clinical, and imaging data were obtained through review of the electronic medical records from Children’s of Alabama (CoA) and the University of Alabama at Birmingham (UAB). Maximal incisal opening (MIO) was routinely measured using the Therabite Measuring Scale (Atos Medical); there was no standardized protocol for its measurement, nor was there adjustment for the incisal vertical overbite. Presence versus absence of lateral deviation of the jaw at rest or with opening was documented at each visit; this was based solely on physical examination findings and did not take into account other potential causative factors (e.g., posterior crossbite).
Screening for TMJ arthritis
During the time of the study, most patients evaluated by pediatric rheumatology at CoA and diagnosed with JIA underwent routine screening for TMJ arthritis by MRI. The MRI protocol is as follows: prior to the administration of contrast, coronal T1-weighted (T1W) and fat-saturated (FS) T2W images are obtained, followed by FS sagittal proton density, FS T2W, and T1W images of the left and the right TMJ. Following the intravenous administration of Magnevist 0.1 mmol/kg, coronal T1W images of the bilateral TMJ were obtained, followed by sagittal T1W images of the left and right TMJ. Children with abnormal findings on TMJ MRI were typically referred to a single OMF surgeon at UAB for evaluation and consideration for IA therapy for TMJ arthritis.
IA therapy for TMJ arthritis
All TMJ injections were performed by the same experienced OMF surgeon (PDW) at UAB. If patients failed systemic medical therapy and had 3 of 5 clinical criteria (pain, decreased maximal incisal opening, growth disturbance, open bite, MRI synovitis), they were considered for IA injections. The corticosteroid preparation used was either triamcinolone hexacetonide (TH) or triamcinolone acetonide (TA), depending on the availability of the former. The medicine was injected at 10 mg/ml (about 1 ml, sometimes less, depending on the joint space volume) up to twice per year13. Since late 2011, IAI was also offered to children with TMJ arthritis refractory to ≥ 1 IACS injections26. This was injected at a concentration of 5 mg/ml (mixed 1:1 with 1% xylocaine), maximal volume possible, typically twice, occurring 6 weeks apart28. An OMF surgeon (PDW) with > 25 years of experience performed the joint injections in the standard fashion of temporomandibular arthroscopic technique, which demonstrated a forward jaw thrust.
Screening for HBF
Children with substantial destructive changes on MRI (Figure 1), as well as concerning features by history or examination (e.g., abnormal motion and grinding), typically undergo traditional or cone beam computed tomography (CBCT) to evaluate for HBF. For our study, all CBCT scans were reviewed for evidence of HBF, namely lamellar bone within the soft tissue of the joint where bone does not normally exist. All suspected cases of HBF were confirmed by a pediatric radiologist (DWY). An illustration of an MRI suggestive of HBF is shown in Figure 1; a confirmatory CBCT scan is shown in Figure 2. Pathologic specimens obtained at time of joint replacement surgery in 2 patients are shown in Figure 3.
Statistical analysis
Data were summarized using means ± SD for continuous variables and n (%) for categorical data. The outcome of interest was the time to development of HBF following the JIA diagnosis. Those who did not develop HBF were classified as censored using their last MRI visit to compute the time to censor. Predictors and risk factors of interest were race, sex, JIA type, age at JIA diagnosis, time to first injection following JIA diagnosis, number of corticosteroid injections, number of infliximab injections, and number of total injections. To investigate the association of these risk factors with the time to development of HBF, survival regression model, in particular Cox proportional hazard model, was fitted. The first step was to investigate the individual risk factors by fitting the Cox model separately for each factor. The second step was to fit the Cox model with all risk factors found to be significant in the first step to determine how that association changes after adjusting for other significant risk factors. Estimates of the HR and their CI were calculated; variables with p values < 0.05 were deemed significant and included in the final adjusted model. Analyses were done using SAS version 9.4 (SAS Inc.). Comparisons of MIO and jaw deviation among children with and without HBF were performed with the Student t test and chi-squared test, respectively.
Ethical statement
Approval for this study was obtained from the Institutional Review Board at UAB. Because of the retrospective features of this study, informed consent was granted an exemption in writing (IRB approval no. X111104004).
RESULTS
Patient population
There were 238 subjects included in the study; their clinical and demographic features are summarized in Table 1. As per the inclusion criteria, all the subjects had undergone at least 1 IACS injection; 55 (23%) had also received IAI. In 233 (98%), all the injections were bilateral. The durations of time from diagnosis of JIA to their final MRI, as well as from their first IACS injection to their final MRI, were 4.0 ± 3.1 and 2.1 ± 1.3 years, respectively (Table 1). Most of the children received traditional DMARD in conjunction with systemic biologic therapy.
Predictors of HBF
There were 33 subjects (13.9%) who developed HBF (Figure 3); this was present unilaterally on the right in 16 (48%), unilaterally on the left in 5 (15%), and bilaterally in 12 (36%). Table 2 shows predictors of HBF by Cox model, which is a form of survival analysis. TH and TA were grouped together under CS, because of the rarity of the latter during the time period of the study. In the unadjusted modeling of time to HBF following JIA diagnosis, the only factors found to be associated with increased risk of HBF were age at diagnosis of JIA and the total number of injections. In this case, the hazard of developing HBF from JIA diagnosis increased by 28% for every year increase in the age at diagnosis (HR 1.279, 95% CI 1.169–1.398, p < 0.0001) and 25% for every additional injection (HR 1.254, 95% CI 1.04–1.512, p = 0.0184). In general, children with more severe arthritis were more likely to receive repeated IA TMJ injections, and it is our practice to screen for TMJ arthritis early in the disease course and to refer for injections if necessary. However, 43 of the patients reported herein had delays of upwards of 3 years from diagnosis of JIA to initial diagnosis of TMJ arthritis. In 36 of these patients (84%), this delay was because the children were either diagnosed with JIA prior to the establishment of a pediatric rheumatology program at CoA in 2007, or because the children transferred their care from other providers. Although there is no way to know the actual duration of TMJ arthritis in these patients prior to the initial assessment, since TMJ arthritis is frequently present at disease onset2, it is likely that many of them had longstanding TMJ arthritis. To the extent that active TMJ arthritis is a risk factor for HBF, it would follow that these subjects would be at greatest risk. Therefore, a variable representing the time from diagnosis of JIA to the initial intra-TMJ CS injection was created. Of interest, the time to first injection showed evidence of a protective effect. In particular, for every additional year added to the time to first injection after JIA diagnosis, the hazard of HBF was estimated to decrease by 56% (HR 0.44, 95% CI 0.296–0.655). Sex, race, and JIA category had no evident association with the development of HBF from JIA diagnosis. In the adjusted model with age of JIA diagnosis, total number of injections, and time to first injection from JIA diagnosis (Table 2), the total number of injections was no longer a significant predictor; however, age at JIA diagnosis and time to first injection from diagnosis were still significant risk factors. HR were similar for these risk factors in both the unadjusted and adjusted models.
As a sensitivity analysis, the analysis was repeated excluding 18 subjects who were initially treated at an outlying hospital, for whom there might be less precision regarding the date of diagnosis of JIA. This did not dramatically change the final model, which again showed that time from diagnosis to the first injection was protective with HR of 0.55 (95% CI 0.36–0.82, p = 0.0075), and the age at JIA diagnosis was associated with increased risk, with HR of 1.2 (95% CI 1.09–1.32, p = 0.0002).
Regarding potential differences between local therapies, only 20 patients received therapy with TA, of whom 13 additionally received TH, so we were likely to be insufficiently powered to compare these 2 CS preparations. Regarding IAI, every patient who received this therapy had also received 1 or more corticosteroid injections, so we could not compare IAI with IACS. After adjusting for the total number of injections, there did not appear to be a difference in type (i.e., CS vs infliximab), but we were underpowered to detect such an effect.
Outcome of HBF
Information on MIO and presence versus absence of jaw deviation at the time of the final visit were available on 223 and all 238 patients, respectively (Table 3). Despite younger age, children without HBF had larger MIO (4.5 ± 0.73 vs 3.9 ± 1.0, p = 0.004) and were significantly less likely to have presence of jaw deviation on opening (28% vs 47%, p = 0.034). Finally, 3 of the 33 subjects who developed HBF required joint replacement surgery.
DISCUSSION
Findings of HBF following TMJ IA therapy raise the possibility that they represent sequelae of the injection itself. Indeed, findings such as this have prompted some to limit the use of IACS into the TMJ18,19. However, a study from Seattle of subjects with HBF noted that of the 2 subjects who underwent biopsy, both had active new bone formation suggestive of sequelae of persistent arthritis, rather than crystalline deposition as would be expected if the etiology were the CS injections themselves20. The data presented herein provides mixed conclusions as to the etiology of HBF. Clearly, the number of injections was associated with risk of HBF. However, the challenge of studying risk factors for HBF is that for obvious ethical reasons, healthy children or even JIA patients without known TMJ arthritis do not undergo IA therapy. Thus, study of the total number of injections is confounded by the disease severity that prompted these injections. An important finding in our work is that children who had a long lag between diagnosis of JIA and introduction of IA therapy appeared to be protected against the development of HBF. If unopposed arthritis in the joint were the main factor, the exact opposite finding would have been anticipated; thus, our data suggest that the injections themselves may be an important contributing factor to the development of HBF, conclusions also reached by Lochbuhler, et al25. If this in indeed the case, it is unclear whether this increased risk of HBF associated with IA therapy of the TMJ is limited to CS or associated with any therapy; in our study, there was no obvious protective benefit associated with infliximab, although we may have lacked sufficient power to detect such an effect. Nevertheless, this finding, in combination with our previous data showing that there is no evident effectiveness of IAI26, fails to provide any compelling rationale to switch from IACS to IAI. Further, to the extent that any IA therapy may predispose to HBF, it follows that simple lavage may also predispose patients to this complication, although none of the patients were treated in that manner during the study period.
Patients with HBF had a smaller MIO despite increased age and increased likelihood of jaw deviation, both of which findings are associated with increased severity of TMJ arthritis29,30. This may simply reflect complications of the worsening arthritis that prompted repeat IA therapy. It is also possible that HBF itself may worsen the severity of the arthritis. Indeed, the bony fragments are visible within the joint space (Figure 2 and Figure 3) and thus may irritate the synovial lining. In support of this possibility, data published in 1984 suggested that periarticular ossifications following IA therapy are composed of hydroxyapatite crystals that can produce inflammatory arthritis31, in which case the therapy could indirectly cause worsening of the underlying disease. As noted above, 3 children underwent joint replacement therapy of the TMJ, indicative of a severely damaged joint.
An unexpected finding was that HBF was more likely in older children. Specifically, each additional year at diagnosis of JIA was associated with a 38% increase in the hazard of HBF (Table 2). This does not appear to reflect differences within the risk of JIA categories or by sex, none of which were statistically significant. Instead, it suggests that older children may have greater intrinsic risk, perhaps because injections were being given during the growth spurt, and thus aggressive systemic therapy may be preferable over local therapy in this population.
Our study has limitations. This was a retrospective study, without any specific protocols for IA therapy, repeat MRI, or CBCT scans. Because of the features of the findings, CBCT scan is clearly far more sensitive for the detection of HBF as compared to MRI, so it is possible that cases were missed. However, all the subjects included in the study did have at least 1 followup MRI following administration of IA therapy, and it is our practice to obtain CBCT scans on those with particularly destructive changes. The MIO may be challenging to measure, particularly in uncooperative children or in patients with overbite or who are missing their incisors. Despite these limitations, however, it is widely used as an outcome measure in the assessment of TMJ arthritis2,10,17.
Finally, it bears emphasis that HBF may not by itself affect function of the TMJ. In contrast, there is an extensive body of literature indicating that active TMJ arthritis can result in altered shape and function, complicating activities of daily life including mastication and talking7. Further, it is clear that systemic therapies do not always treat the underlying arthritis6. Therefore, while we agree with previous calls for caution in the use of IACS of the TMJ18,19, these risks must be weighed against the risks of unopposed arthritis involving this joint, particularly in patients with substantial arthritic changes in whom systemic therapy has already optimized. Additionally, future studies should prospectively evaluate the safety and effectiveness of IACS, as compared to systemic therapy alone, IAI26, arthrocentesis with lavage only32, and possibly even iontophoresis33.
Acknowledgment
The authors acknowledge Drs. Thomas Winokur and David Kelly for assistance with creating the pathology figures.
- Accepted for publication February 12, 2018.
REFERENCES
- 1.
- 2.
- 3.
- 4.
- 5.
- 6.
- 7.
- 8.
- 9.
- 10.
- 11.
- 12.
- 13.
- 14.
- 15.
- 16.
- 17.
- 18.
- 19.
- 20.
- 21.
- 22.
- 23.
- 24.
- 25.
- 26.
- 27.
- 28.
- 29.
- 30.
- 31.
- 32.
- 33.