Low-Level Laser Therapy for Temporomandibular Disorders: A Systematic Review with Meta-Analysis

Objectives We systematically reviewed randomized controlled trials (RCTs) of the effect of low-level laser therapy (LLLT) versus placebo in patients with temporomandibular disorder (TMD). Methods A systematic search of multiple online sources electronic databases was undertaken. The methodological quality of each included study was assessed using the modified Jadad scale, and the quality of evidence was evaluated using the Grading of Recommendations, Assessment, Development and Evaluation (GRADE) system. Results A total of 31 RCTs were included. Total modified Jadad scale scores showed that the methodological quality was high in 30 studies and low in 1 study. Combining data from all clinically heterogeneous studies revealed positive effects of LLLT on pain relief, regardless of the visual analogue scale (VAS) score or the change of VAS score between the baseline and the final follow-up time point, while dosage analyses showed discrepant results about the effects of high or low doses for patients with TMD. Follow-up analyses showed that LLLT significantly reduced pain at the short-term follow-up. Temporomandibular joint function outcomes indicated that the overall effect favored LLLT over placebo. Conclusion This systematic review suggests that LLLT effectively relieves pain and improves functional outcomes in patients with TMD.


Introduction
Temporomandibular disorder (TMD) is a set of clinical conditions that includes disorders of the temporomandibular joint (TMJ) and/or the masticatory muscles [1]. e most common symptoms are pain, joint noises, and restricted mandibular movement [2]. A variety of other symptoms may occur, such as tinnitus, abnormal swallowing, and hyoid bone tenderness [3,4]. ese symptoms compromise quality of life (QoL) [5], sleep [6], and the psychological well-being, leading to anxiety, stress, depression, and a negative effect on social function, emotional health, and energy level [7]. e incidence of signs and symptoms of TMD varies from 21.5% to 50.5%, and they occur more frequently among women than men [8][9][10]. e etiopathogenesis of TMD remains unclear. In general, it is thought that the origin of TMD is multifactorial, including biomechanical, neuromuscular, biopsychosocial, and biological factors [11]. erefore, the mainstay of treatment for TMD is a multidisciplinary approach that includes physical therapy modalities such as manual therapy [12], electrotherapy [13], ultrasound [14], transcutaneous electrical nerve stimulation (TENS) [15], or laser therapy [16].
Among the various physical therapy modalities, lowlevel laser therapy (LLLT) has recently been put under the spotlight because of its easy application, short treatment time, and few contraindications. Many prospective clinical trials have been performed to evaluate the efficacy of LLLT. However, the results have been controversial [16][17][18][19][20][21]. Some authors have reported the superiority of LLLT over placebo [16,20,21], while others have found no significant differences between LLLT and placebo [17][18][19].
Over the past recent years, a number of systematic reviews with or without meta-analysis have analyzed the efficacy of LLLT for TMD [2,[22][23][24][25][26][27]. Based on the included studies, which were all published before 2010, four systematic reviews concluded that there was no definite evidence to support the use of LLLT in the management of TMD [22][23][24][25]. On the contrary, one meta-analysis published in 2014 concluded that applying LLLT to the masticatory muscle or joint capsule has a moderate analgesic effect on TMJ pain [2]. In 2015, another meta-analysis provided evidence that using LLLT has limited efficacy in reducing pain, but can significantly improve the functional outcomes of patients with TMD [26]. However, there is no solid evidence to support or refute LLLT for TMD.
Since the latest published meta-analysis, many new randomized controlled trials (RCTs) have been conducted, which may accumulate evidence on the use of LLLTfor TMD [4,[28][29][30][31]. erefore, in this systematic review with metaanalysis, we reevaluated the effect of LLLT versus placebo in patients with TMD. e results of this study may provide practical recommendations for clinical physicians who treat patients with TMD.

Search Strategy and Selection Criteria.
is review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement guidelines [32,33] and Cochrane handbook for systemic reviews [34]. We systematically searched PubMed, EMBASE, CINAHL, ClinicalTrials.gov, the Cochrane Library database, AMED, Toxline, PEDro, ProQuest Digital Dissertations, PsycBite, SCOPUS, Current Contents Connect, Web of Science, and the WHO Trial Registry for RCTs comparing LLLT with a placebo intervention in patients with TMD. e following search terms were used: "temporomandibular disorder(s)" OR "temporomandibular joint disorder(s)" OR "temporomandibular joint dysfunction" OR "TMJ disorder (s)" OR "TM disorder(s)" OR "temporomandibular joint pain" OR "temporomandibular pain" OR "TM pain" OR "TMJ pain" OR "TMD" OR "temporomandibular osteoarthritis" OR "myofascial pain" OR "craniomandibular disorder(s)" OR "mandibular dysfunction" AND "laser" OR "laser therapy" OR "low level laser therapy" OR "low intensity laser therapy (LILT)" OR "low energy laser therapy (LELT)" OR "LLLT" OR "infrared (IR) laser" OR "IR laser" OR "diode laser" OR "helium-neon laser" OR "HeNe laser" OR "gallium-arsenide laser" OR "GaAs laser" OR "gallium-aluminium-arsenide laser" OR "GaAlAs laser." e last search was performed on May 16, 2017.
Inclusion criteria were as follows: (1) RCTs involving patients with TMD; (2) articles published or informally published in English or Chinese; and (3) primary studies or studies in which LLLT was compared with placebo or sham laser, with similar appearance to the active treatment but without laser irradiation, in patients with TMD; and (4) studies of LLLT for myogenous or arthrogenous temporomandibular pain, or both, regardless of age and gender. Studies including cointerventions were allowed if applied equally to both the LLLT and placebo groups.
Exclusion criteria were as follows: (1) nonrandomized or crossover studies; (2) total number of study participants (in the LLLT and placebo groups combined) less than 10; (3) meeting abstracts that did not report data for the outcomes of interest; and (4) studies involving patients with systemic diseases (i.e., fibromyalgia and rheumatoid arthritis) or pain not related to TMD (i.e., neuralgia, toothache, and psychological disturbances).

Study Selection.
Two independent reviewers (Xu and Jia) initially screened and identified relevant titles and abstracts. Full-text articles were obtained for all eligible studies, and these were assessed independently by Jin and Li against an inclusion and exclusion checklist. Disagreements were resolved by discussion until consensus was reached; if this approach failed, a third party (Cao) was consulted. e reference lists of all retrieved studies were manually examined to identify any studies missed by the electronic literature search. We also contacted all principal investigators or corresponding authors of the identified studies for additional information where necessary.

Outcome Measures.
e primary outcome of interest was pain intensity, as expressed by visual analogue scale (VAS) score (at the final follow-up time point) or the change of VAS score (between the baseline and the end of the follow-up) in the LLLT and placebo groups. e secondary outcomes included the change of TMJ function between the baseline and the end of the follow-up, oral function (masticatory performance), electromyographic (EMG) activity, adverse effects, pressure pain threshold (PPT), joint noises, tinnitus, quality of life (QoL), and psychological satisfaction in the LLLT and placebo groups.
TMJ function was assessed in terms of maximum active vertical opening (MAVO), maximum passive vertical opening (MPVO), lateral excursion (LE), and protrusion excursion (PE), expressed in millimeters.

Data Extraction.
We used data from the longest followup time point for each trial. Data were extracted and crosschecked independently by Li and Wang using a standard data extraction form that contains general information (authors, publication year), subject number, treatment-related information, and relevant clinical outcome data. Authors were contacted to clarify further information where necessary. In three RCTs that examined more than one laser dose, the placebo group was divided into two equal-sized groups to avoid "double counting" to allow inclusion of two independent comparisons within the meta-analysis [30,35,36].
Data from the included studies were pooled for further meta-analysis where appropriate. If available, means and standard deviations for outcome measures were extracted or calculated based on the published data with RevMan 5.0 software as supplied by the Cochrane Collaboration. Means and standard deviations were used to calculate mean differences (MDs) and 95% confidence intervals (CIs) in the meta-analysis.

Assessment of Methodological Quality.
All included studies were assessed for methodological quality using the modified Jadad scale [37]. Two reviewers (Xu and Jia) performed the assessment independently, and discrepancies between reviewers were resolved by consensus. Studies achieving four or more points (from a maximum of eight) were considered to be of high quality, while studies scoring below four were considered to be of low quality.
2.6. Statistical Analysis. Dichotomous outcomes were expressed as relative risks (RRs) and continuous outcomes were expressed as the weighted mean differences (WMDs), both were presented with 95% CIs. Pooled effect sizes were based on the results of pain intensity (assessed by VAS) as well as MAVO, MPVO, LE, and PE values in millimeters. Revman 5.0 Software was used to summarize the effects and to construct the forest plots for all comparisons. Heterogeneity was examined according to the I 2 statistic alongside the chi-squared test; if I 2 was greater than 50%, the randomeffects model was applied [38]. Qualitative analysis was performed if studies failed to provide data to be pooled for analysis. Publication bias was assessed by examination of funnel plots for primary outcomes. A symmetric funnel plot represented lower risk of bias and vice versa. Because interstudy heterogeneity precluded a meta-analysis in some outcomes, narrative synthesis of related studies was employed.

Subgroup and Sensitivity
Analysis. Subgroup and sensitivity analysis were planned in the presence of heterogeneity. Subgroup analysis was performed to evaluate the effect of the intervention at different laser dosages and follow-up periods (short-term and long-term effects). Sensitivity analysis was performed for testing the robustness of the pooled effect size where appropriate. Effects were examined according to methodological quality, to ensure that the analysis was not biased by a low-quality study or a study with a large population.

Evaluation of Quality of Evidence.
e quality of evidence was evaluated using the Grading of Recommendations, Assessment, Development and Evaluation (GRADE) system [39], which is based on five domains (limitations of the study design, inconsistency, indirectness, imprecision, and publication bias).

Search Results.
e study selection process is shown in Figure 1. A total of 1537 records were identified from searches. No unpublished manuscripts were identified. After excluding 1506 records, a total of 31 articles that met the inclusion criteria were included in the present systematic review.

Quality Assessment.
A summary of the quality assessment using the modified Jadad scale scores is shown in Table 1. Total scores showed that the quality of 30 studies was high, with a minimum of 4 points and a maximum of 8 points. One study had low quality (3 points) [50] (Supplementary Material Appendix A).

Effects of Laser erapy.
e 31 RCTs showed mixed results, as reported by the authors, with two-thirds reporting positive effects favoring LLLT and one-third reporting inconclusive results or no effect. Twenty-two studies provided sufficient data to calculate effect sizes for key outcome measures and were included in the meta-analysis. Subgroup analysis was performed for laser dose and follow-up period using the random-effects model. Studies were subcategorized into low dosage (≤50 J/cm 2 ) versus high dosage (>50 J/cm 2 ) and into short-term follow-up (≤2 weeks) versus long-term follow-up (>2 weeks). For all studies, we only collected data from the final follow-up time point.

Primary
Outcomes. All 22 studies, except three [4,40,47], used VAS to assess pain as one of the primary outcome measures. However, as a result of data detectability, only 19 studies were subjected to meta-analysis.

Mean Difference of VAS Score.
e mean difference of VAS score (change scores from baseline) between the baseline and the final follow-up time point was used [17,42,49,53]. When the data were missing, we calculated them using the published relevant data with RevMan 5.0 software [1, 16, 18-21, 28, 30, 31, 35, 41, 43, 50, 51, 54]. Analysis of data from 679 subjects (19 studies) revealed a significant difference between the LLLT and placebo   Pain Research and Management

Sensitivity Analysis.
e pooled results were in favor of LLLT based on the above-mentioned main findings (see Supplementary Material Appendix C), after excluding lowquality [50] and/or extreme value and maximal weight value (large population) studies [20].
Most of these outcomes showed significant differences between the LLLT and placebo groups, except for MAVO and PE at low dosage and PE at unknown dosage. In addition, there were no MPVO data in the high-dosage subgroup. All data of TMJ function in follow-up subgroups showed significant differences between the LLLT and placebo groups, except for MAVO at the short-term follow-up (Supplementary Material Appendix D).

PPT. Seven studies investigated pain by measuring
PPT [19,29,30,35,48,51,55], expressed as mm, kpa, or kg/cm 2 . It is impossible to estimate the overall effect size across the different scales. Four studies showed a significant change [29,30,35,51], while two reported no change of PPT in the LLLT group compared to the placebo group [19,48]. In another study [55], a significant improvement was observed in the LLLT group only at a dosage of 4 J/cm 2 (P � 0.0156), but not at 8 J/cm 2 (P � 0.4688).

EMG Activity.
ree studies measured EMG activity before and after treatment [35,45,55]. LLLT did not promote any changes in EMG activity [45,55]. However, in another study [35], EMG records in the maximum voluntary clenching (cEMG) of the MLILT (a modified high-energy LILT) group were significantly higher after the final treatment than those of the placebo group (P � 0.022, 95% CI � 5.96-68.66 microV), but there was no significant difference of cEMG between the CLILT (a conventional lowenergy LILT) and placebo groups.

Oral Function Outcome
Measures. Masticatory efficiency was evaluated in two studies [1,29]. However, the used methods were different, for the results were considered inadequate for meta-analysis. Although both studies indicated that masticatory performance might be better after LLLT treatment, only one study [29] showed a significant improvement in masticatory performance in the LLLTgroup at the end of treatment compared with baseline values (P < 0.01).
3.6.6. Tinnitus. One study evaluated subjective tinnitus [4]. e study applied two types of LLLT in bilateral subjective tinnitus with TMD, namely, LLLT with Nd:YAG (1064 nm) and LLLT with a diode laser (810 nm). Both Nd:YAG and diode laser were effective for the treatment of subjective tinnitus related to TMD.
3.6.7. QoL and Psychological Satisfaction. None of the included studies reported QoL or psychological satisfaction as an outcome measure.

Adverse Effects.
Nine of the 31 included studies reported no adverse effects related to laser application during or after the treatment period [4,16,28,30,47,[51][52][53][54]. e other studies lacked information regarding the adverse effects of laser exposure.

Publication Bias.
Considering the heterogeneity of the studies, funnel plots were drawn according to different outcome measures. Visual assessment of funnel plots did not show considerable asymmetry in pain (VAS score), LE, and PE, indicating that the publication-related bias was low for these outcomes. However, there was asymmetry in MAVO and MPVO outcomes, indicating that the publication bias was high for these two outcomes (Supplementary Material Appendix E).

Quality of Evidence.
e present meta-analysis investigated a total of six types of outcomes (including 29 subgroup analyses stratified by laser dosage and follow-up period) about pain intensity and mandibular function. e GRADE assessment of the level of evidence for these outcomes is shown in Supplementary Material Appendix F. e quality of the evidence was judged to range from very low to moderate. All domains affected low grades except indirectness.

Discussion
is systematic review and meta-analysis summarized RCTs that compared the effect of LLLT with placebo for the treatment of TMD. e results of the studies indicated that LLLT was effective in reducing TMD pain compared to placebo. In addition, LLLT could improve functional outcomes. Combining data from all clinically heterogeneous studies demonstrated positive effects of laser on pain relief, regardless of VAS score or the change of VAS score between the baseline and the final follow-up time point, while dosage subgroup analyses showed discrepant results about high or low dosage for TMD patients. e follow-up subgroup analysis showed more consistent results, suggesting LLLT significantly reduced pain at the short-term follow-up both in the VAS score and the change of VAS score. However, there was no significant difference at the long-term followup between LLLT and placebo. TMJ function outcomes, assessed in terms of MAVO, MPVO, LE, and PE (the changes between baseline and the end of follow-up) indicated that the overall effect favored LLLT over placebo. Most (five out of 7) studies indicated PPT improved by LLLT, but most (two out of 3) studies showed no change in EMG activity. e use of LLLT has been seen as a complementary option for the treatment of TMD [16,20,21,26] due to its analgesic, anti-inflammatory, and regenerative effects with no reported adverse effects and good acceptance by patients [16,28,30,51,53,54,57,58]. In view of the lack of robust evidence about the effects of LLLT on TMD, recent systematic reviews did not reach a consensus [2,[22][23][24][25][26][27].
Here, we update the clinical evidence for the effects of LLLT on TMD.
Our results regarding the analgesic effect of LLLT are consistent with findings of Chang et al. [2], in contrast to those of Chen et al. [26], but the results regarding functional outcomes (motion) were in accordance with those of Chen et al. [26]. e strengths of our systematic review are the larger number of studies and the inclusion of the most recent publications since the last review in the subject.
As pain is the principal complaint of patients with TMD, pain is the most common reason why patients with TMD seek medical help. Pain occurs at any stage of TMD and pain reduction contributes to ameliorating jaw motion [26,59], chewing [60], and masticatory performance [29]. erefore, our primary outcome measure was pain intensity. Given the heterogeneity of the included studies, meta-analyses were performed using subgroups of studies according to the dosage and follow-up time. Although the overall effects of LLLT on pain were positive in both the VAS score and the change of VAS score, subgroup analysis reached contrary conclusions for high dosage versus low dosage in these two parameters. It is difficult to draw precise conclusions regarding an effective dosage window from these studies due to the contrary conclusions from subgroup analysis (analyzed by actual VAS score or the change of VAS score) and the wide dosage range employed. Four of the included studies [35,36,43,45] compared high dosage with low dosage, but only one study [35] showed the superiority of high dosage, while the others showed no differences between the two dosages. e mechanism underlying the therapeutic effects of LLLT is under debate [61,62]. e magnitude of the laser effect seems to also depend on the dosage of laser [63]. Bjordal et al. [64] believed that the controversy on the efficacy of LLLT on TMD laid in the disagreement on the dosage of laser. Laser acupuncture has been suggested to be a dosage-dependent modality [65,66], suggesting that the energy delivered to the target point by laser acupuncture has to reach a threshold in order to produce a desired effect.
us, the dosages used in the included studies may explain the observed differences in outcomes.
With regard to the relationship between laser effectiveness and follow-up period, Law et al. found that long-term follow-up effects increased in three types of musculoskeletal disorders (myofascial pain/musculoskeletal trigger points, lateral epicondylitis, and temporomandibular joint pain) [67]. Pooled effect sizes were doubled during the follow-up Pain Research and Management period compared to those at the end of intervention, suggesting that laser may have delayed or long-lasting effects. However, our follow-up subgroup analysis showed a more consistent result, contrary to Law et al., both of pain evaluation methods showing significant differences only in short-term follow up between the laser and placebo groups. In one trial conducted by Carli et al. [28], laser, piroxicam, and placebo significantly improved the VAS score. An evaluation at 30 days after the end of the treatment showed that the laser did not have a residual effect, and piroxicam was more effective than the laser to reduce the level of muscular pain in patients with TMJ arthralgia. ese results illustrate that laser may have a short-lived effect, which is consistent with our follow-up subgroup analysis data. However, because of the high degree of heterogeneity, this finding needs to be explored in further research. Nonetheless, the present systematic review and others support the continued use of laser for treating TMD/musculoskeletal pain.
Besides the subjective pain assessments, the objective clinical outcomes include TMJ function, PPT, EMG, and masticatory performance. e overall effects in MAVO, MPVO, LE, and PE favored LLLT over placebo. In most studies, LLLT increased PPT, but did not affect EMG activity. Only one study [29] showed that LLLT improved masticatory performance at the end of treatment compared with baseline values, but other study [1] showed no significant effects.
Skin surface application of laser (trigger points/tender points) was used in most of the included studies, external auditory meatus was used in some studies [4,41,52], and only one study added remote application sites at acupuncture points [16]. We are unable to perform analysis on different sites for the variety and the complexity of these application sites.
It is worth to mention that using the combination of two wavelengths [21,53] yields positive result. Shirani et al. [53] combined GaAs and InGaAlP lasers, usually applied for deep-lying disorders and superficial disorders, respectively. Another study [21] applying GaAlAs at 650 nm and 830 nm also obtained good effects, implying that the combination of two laser wavelengths may be beneficial to patients with TMD. During TMD treatment with LLLT, the variability of laser type, frequency, dosage, exposure time, application area, number of laser sessions, and therapy duration may increase heterogeneity in effects.
us, the findings of clinical studies must be interpreted against the background. Using proper laser parameters is important to obtain better effects, as suggested by Law et al. [67] Additionally, due to the multifactorial etiology of TMD, including biopsychosocial and biological factors [11], about one-third of the patients report eating problems and feelings of depression or dissatisfaction with life [68]. However, none of the included studies focused on psychological assessment. TMD diagnosis has been standardized based on research diagnostic criteria for temporomandibular disorders (RDC/TMD) that constitute a multidimensional diagnostic research tool adopted worldwide [69,70]. is standardization has improved reproducibility among clinicians and has facilitated the comparison of results among researchers [16,28,44]. It is important to establish standardized therapy regimens about TMD through evaluating relevant behavioral, psychological, and psychosocial factors (e.g., pain status variables, depression, nonspecific physical symptoms, and disability levels) [69,70]. In addition, the effect of laser can be further evaluated by adding QoL and patient satisfaction as outcome measures.
Nine of the 31 included studies explicitly stated that no adverse effects were observed [4,16,28,30,47,[51][52][53][54]. e other studies lacked information regarding the adverse effects of laser exposure. Nevertheless, it is thought that LLLT is safe. LLLT is noninvasive and has few or none adverse effects, which may contribute to increased patient comfort.
is systematic review has some limitations. First, the methodological quality varied among the included studies. Second, there was a high degree of heterogeneity because of differences in TMD diagnosis, laser style, laser parameters, treatment regimens, outcome measurements, and follow-up time, which hindered some comparisons between studies. Furthermore, discrepancies also existed in the inclusion and exclusion criteria. ird, although we tried to obtain full data, some data were missing because some of the studies reported continuous variables such as MAVO or MPVO without SD, while others used box plots or histograms to represent the data. Moreover, most of the included studies had small sample sizes (≤60 subjects), which limited the generalizability of the conclusions. Fourth, although a systematic search of multiple databases was undertaken, some unpublished grey literature might have been missed. In addition, some of the included studies were not used in the meta-analysis. us, potential publication bias and selection bias could not be eliminated. We only included English and Chinese language articles, which could induce a language bias. Pooled analysis of all kinds of TMD (muscular origin, articular origin, or a combination of both) degraded the level of conclusion because of the poor description of TMD type in some studies. Given the above reasons, the findings from this study should be interpreted cautiously.

Conclusion
e results of this systematic review and meta-analysis are encouraging. Despite the above-mentioned limitations, the overall effect illustrated that LLLT effectively relieves pain in the treatment of TMD. LLLT may induce a short-term effect only, but the existing evidence does not allow us to determine an effective dosage window. Moreover, LLLT also improves the functional outcomes in TMD. In view of the high discrepancy among the included studies, this systematic review highlights the need for more well-designed RCTs with larger sample sizes to evaluate the efficacy of LLLT. Future research should carefully define the study population and provide the rationale for the parameters chosen. is would facilitate not only replication in the clinical setting, but also improve trial homogeneity and allow data to be pooled for meta-analysis. Furthermore, it is necessary to examine different laser parameters, treatment regimens, evaluation times, and outcome measures because it is noninvasive, safe, easy-to-use, and cheap.

Conflicts of Interest
e authors declare that they have no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.