Efficacy and safety of propranolol for treatment of TMD pain: a randomized, placebo-controlled clinical trial

1. Introduction

Painful temporomandibular disorder (TMD) is a common musculoskeletal condition most often caused by myalgia of masticatory muscles and/or arthralgia of the temporomandibular joints (TMJ). In 2002, symptoms affected 6.3% of females and 2.8% of males in the U.S. adult population,³⁵ and prevalence has not changed appreciably.^45,58 Chronic TMD is associated with substantial disability and suffering, and it diminishes quality of life.¹⁴ Jaw pain is the most common symptom that compels patients to seek treatment. One study reported an average TMD pain intensity rating of 4.3 on a 11-point scale, similar to the averages reported for chest pain and back pain.⁷⁵ In addition to facial pain, TMD patients frequently report various comorbid idiopathic pain conditions such as headache, low back pain, and fibromyalgia.^9,58

The current pharmacological approaches to treat painful TMD comprise nonsteroidal anti-inflammatory drugs (NSAIDs), corticosteroids, benzodiazepines, sedative hypnotics, muscle relaxants, opioids, antidepressants, and anticonvulsants.^31,32,43 However, the 2010 Cochrane review of pharmacological interventions for TMD concluded that there is insufficient evidence to support the effectiveness of the reported drugs and it emphasized a need for high-quality randomized controlled trials for this condition.⁴⁹

TMD-associated pain has a significant sympathetic component supported by the ample sympathetic innervation of the TMJ.^38,78 In rodent models of TMD, the depletion of norepinephrine or the blockade of β-adrenergic receptor reduced carrageenan-induced TMJ hyperalgesia⁶¹ and formalin-induced TMJ nociception.²⁶ A recent report found that the analgesic effect of a non-selective β-adrenergic receptor antagonist propranolol in the carrageenan-induced TMJ hyperalgesia in female rats was associated with reduction in joint inflammation.⁸⁰ In addition, intramuscular injection of propranolol in rats also reduced carrageenan-induced pain of the gastrocnemius muscle.⁴¹ While these studies used animal models of inflammatory pain, it should be noted that subclinical muscle inflammation in humans is one of the plausible causes for chronic TMD myalgia.⁴⁴

In humans, injection of propranolol into the masseter muscle of healthy volunteers decreased pain induced by local administration of serotonin,²¹ but not of hypertonic saline.⁴ Human studies exploring adrenergic blockade for treatment of TMD-related pain are scarce: one study demonstrated pain relief after a single-dose injection of propranolol⁴¹ and our pilot study found a decrease in pain rating after one-week treatment with propranolol.⁶⁹ Currently, extended-release propranolol is indicated for treatment of hypertension, angina pectoris due to coronary atherosclerosis, migraine, and hypertrophic subaortic stenosis.³⁶

To further investigate propranolol as a novel drug for management of TMD-associated pain, we conducted a phase 2b randomized controlled trial (SOPPRANO: Study of Orofacial Pain and PropRANOlol). The primary objective of the trial was to evaluate efficacy of propranolol compared to placebo in reducing a composite index of facial pain intensity and duration following 9 weeks of treatment among participants with TMD. Secondary objectives were to assess participant safety and to evaluate efficacy using secondary endpoints measuring other aspects of facial pain, physical and emotional functioning, and participant ratings of improvement.

2. Methods

2.1. Trial oversight

The trial protocol was approved by the Institutional Review Boards at each trial site. Written informed consent was obtained from each participant before any study assessments or procedures were done. Twice a year, the trial was reviewed by the independent Data Safety Monitoring Board and 100% of critical trial data was externally monitored.

2.2. Trial participants and inclusion/exclusion criteria

Participants were recruited from communities at 3 sites in the United States: the University of North Carolina at Chapel Hill (UNC), Chapel Hill, North Carolina; the University of Florida (UF), Gainesville, Florida; and the University at Buffalo (UB), Buffalo, New York. The recruitment continued from August 2015 to January 2018 and the follow-up of the last participant was finished in April 2018. The UB site started enrollment in January 2017. Eligible participants were women and men 18 to 65 years of age who had TMD myalgia (with or without arthralgia) when assessed at the baseline visit by study examiners who used the Diagnostic Criteria for Temporomandibular Disorders (DC/TMD).⁶⁴ Details of the DC/TMD examination are described below. Participants also had to report having experienced facial pain for at least 3 months, and had at least 10 days with facial pain in the 30 days prior to the examination. Another inclusion criterion was based on facial pain intensity reported on a “0” to “100” numerical rating scale in a daily symptom diary (DSD). Specifically, during the week before randomization, participants had to report a pain rating of ≥30 on at least 3 days, or their weekly average rating was ≥30. The former criterion was added to the protocol in March 2016 to complement the existing pain intensity criterion with a modification that assessed frequency of painful days.

As rescue medication, participants were allowed to take over-the-counter (OTC) nonsteroidal anti-inflammatory drugs (NSAIDs), e.g., acetaminophen or aspirin, episodically, where the episodic use was defined as no more than 3 consecutive days and no more than 18 total days of NSAIDs from baseline to treatment week 9. Concomitant use of prescription pain medications was allowed only if the medication was taken daily for at least 30 days before a baseline phase and continued to be taken daily during the trial. Orofacial appliances for facial pain could be used during the trial if their use was initiated at least 30 days prior to a screening visit and maintained for the duration of the trial.

Exclusion criteria were congestive heart failure, a clinically significant abnormal 12-lead electrocardiogram (ECG), sinus bradycardia, uncontrolled hypertension or hypotension, bronchial asthma, nonallergic bronchospasm, renal failure or dialysis, diabetes mellitus, hyperthyroidism, fibromyalgia, or uncontrolled seizures; used opioids, β-blockers or medications that could interact with propranolol; had facial trauma or orofacial surgery within 6 weeks prior to a screening visit; had major psychiatric disorders requiring hospitalization within the last 6 months prior to a screening visit; had treatment for drug or alcohol abuse within the last year; or were pregnant or nursing. The detailed list of inclusion/exclusion criteria is provided in the Supplementary Materials.

2.3. Trial design and intervention

This multisite, randomized, double-blind, placebo-controlled, parallel-group, phase 2b trial consisted of a telephone or visit pre-screening (conducted up to 28 days before a screening visit), a baseline period (1 to 3 weeks), a treatment period (10 weeks including a 1-week titration, an 8-week maintenance period, and a 1-week taper), and a follow-up period (1 week). A diagram of the trial is presented in Figure 1. After signing informed consent at the screening visit, participants were evaluated for eligibility during a subsequent baseline period. Eligible participants were randomized to receive propranolol hydrochloride extended release (ER) 60 mg twice daily (BID) or placebo. In both treatment arms, the study drug was titrated starting with one capsule per day (QD) in the evening and increasing after one week to two capsules per day (in the morning and in the evening). After reaching the target dose, participants stayed on this dose for 8 weeks and then were tapered down to one capsule per day (in the evening) for 1 week before terminating the use of the study drug. Dose adjustment and drug discontinuation were allowed in case of adverse events. For protocol-specified evaluations, participants completed 6 scheduled visits: screening, randomization (propranolol 60 mg QD), week 1 (propranolol 60 mg BID), week 5 (propranolol 60 mg BID), week 9 (propranolol 60 mg QD), and week 11 (follow-up).

Figure 1.

Trial design.

2.4. Randomization and masking

Randomization was initiated by study coordinators via an electronic web response system. The randomization list was generated centrally by randomization staff at the Data Coordinating Center (DCC: Rho Inc, Durham, NC), who remained independent from other DCC staff throughout the trial. The randomization sequence was created with Proc Plan in SAS statistical software (SAS v9.2, SAS Institute Inc, Cary, NC) and was stratified by study site with a 1:1 allocation using fixed permutated blocks of 4. The DCC sent randomization assignment directly to site pharmacies that prepared bottles with investigational product for participants.

An independent pharmacy (Triangle Compounding Pharmacy, Cary, NC) provided 60-mg extended release propranolol over-encapsulated in a blue, oblong capsule and matching placebo. The placebo was matched to the study drug for color, size, and weight, and contained microcrystalline cellulose. The matched drug and placebo were shipped to site pharmacies for distribution to study coordinators. Site staff, investigators, participants, monitors, and a statistician analyzing a primary endpoint were blinded to the allocation. The blinding was tested by asking participants to report their perceived group allocation at weeks 5 and 9 of treatment.

2.5. Trial endpoints and procedures

We followed recommendations of the Initiative on Methods, Measurement, and Pain Assessment in Clinical Trials (IMMPACT)¹⁷ by measuring 6 core domains: (1) pain, (2) physical functioning, (3) emotional functioning, (4) participant ratings of improvement, (5) symptoms and adverse events, and (6) participant disposition. As further recommended by IMMPACT,¹⁹ we conducted thermal and mechanical quantitative sensory testing (QST). The endpoints were collected at baseline and weeks 5 and 9 of treatment, if not otherwise specified.

2.5.1. Primary endpoint

As specified a priori in the study protocol, the primary endpoint was change in a weekly mean facial pain index after 9 weeks of treatment. The pain index was selected as the primary endpoint based on findings from the pilot study,⁶⁹ where the composite index was sufficient in capturing changes seen separately using endpoints of pain intensity and pain duration, each of which represents an imporant component of the pain experience. The weekly mean facial pain index was computed as the arithmetic mean of a daily facial pain index calculated during the week before randomization and for the week prior to each study visit. The daily facial pain index was computed as daily facial pain intensity multiplied by daily facial pain duration and divided by 100. The daily facial pain intensity was rated on a standard numerical rating scale from 0 = ”no pain” to 100 = “worst pain imaginable,” while the daily facial pain duration was measured as percentage of waking day when participant had pain. The weekly mean facial pain index was calculated if a participant completed at least 4 of the 7 daily reports of pain intensity and duration per week; otherwise, it was considered missing. The component scores of pain intensity and pain duration were analyzed as secondary endpoints. To prevent potential spill-over of headache pain into ratings of facial pain, separate DSD items asked about intensity and duration of tension-type and migraine headache. All targeted types of pain for the DSD were anchored through explicit instructions to each participant. For example, facial pain was described to participants as “pain in your jaws, temples, ears, or in front of your ears.”

2.5.2. DC/TMD examination and pain-related secondary endpoints

Chronic TMD myalgia and arthralgia were classified according to the DC/TMD specifications.⁶⁴ To summarize: examiners assessed masticatory muscles and the temporomandibular joints, noting any sites at which pain was elicited either by palpation or upon jaw maneuver (jaw opening or lateral excursion). When pain was elicited, examiners asked participants to report if the pain was familiar to any facial pain symptoms experienced in the preceding 30 days. At the end of the examination, participants were asked if jaw function during the previous 30 days had altered (made better or made worse) any of the familiar evoked pain from the examination. TMD myalgia and arthralgia were classified separately, when there were: a) evoked, familiar pain in either the temporalis or masseter muscles (myalgia) or in the TM joints (arthralgia), and b) a positive history of the evoked familiar pain having been modified by jaw function. An examiner calibration session conducted prior to the enrollment demonstrated excellent inter-examiner agreement (kappa for pairwise comparisons ranged from 0.82 to 1.00).

In addition to TMD classification, three examination findings are presented in this paper as secondary endpoints: pain-free, maximum unassisted, and maximum assisted jaw openings.

TMD-related disability and interference in functioning were assessed using the Graded Chronic Pain Scale (GCPS).⁷⁵ The GCPS grade is derived from several variables: (1) the characteristic pain intensity computed as the mean, multiplied by 10, of the 3 pain items; (2) the pain interference score, computed as the mean, multiplied by 10, of 3 pain interference items; and (3) pain disability days. Based on these 3 variables, participants were classified into 6 chronic pain grades: 0 = no pain, I = low pain intensity and low pain-related disability, IIa = high pain intensity and low pain-related disability, IIb = high pain intensity and high activity interference, III = moderate pain-related disability, IV = severe pain-related disability. For analyses, the GCPS endpoint was dichotomized into a low-grade category including grades from 0 to IIa and a high-grade category including grades from IIb to IV.

Limitations jaw function were assessed with the Jaw Functional Limitation Scale (JFLS),^52,53 a validated questionnaire that measures limitations on 3 subscales: mastication (6 items), vertical jaw mobility (4 items), and verbal and emotional expression (8 items). Each item is rated on an 11-point scale from 0 = “no limitation” to 10 = “severe limitation”. The subscales are computed as a mean response for all items in the subscale, while the total score is the mean of all 18 items.

2.5.3. Physical functioning

Short Form-12 Health Survey version 2 (SF-12 v2)⁷⁶ was used to assess overall physical functioning. The SF-12 v2 is a briefer measure of health-related quality of life derived by using 12 items from the original Short Form 36 instrument. This questionnaire produces 2 scores: a mental health composite score and a physical health composite score.

Sleep was assessed with Pittsburgh Sleep Quality Index (PSQI).⁸ This 19-item instrument assesses sleep quality during the previous month across several domains: subjective sleep quality, sleep latency, sleep duration, habitual sleep efficiency, sleep disturbances, use of sleep medication, and daytime dysfunction.

As propranolol is used for migraine prophylaxis, participant migraine status was assessed using a structured headache interview,⁷⁰ based on the International Classification of Headache Disorders, 3^rd edition beta version (ICHD-3 beta).¹ In addition, to evaluate headache-related disability, we used the Headache Impact Test (HIT-6) that consists of 6 items: pain, social functioning, role functioning, vitality, cognitive functioning, and psychological distress.³⁹ The participant answered each question using responses ranging from “never” to “always.” These responses were summed to produce a total HIT-6 score, with higher scores indicating a greater degree of headache-related burden.

Alcohol consumption was assessed at screening via the Alcohol Use Disorders Identification Test (AUDIT)⁶³ and was monitored throughout the duration of the trial. Smoking habits were evaluated at screening and at the end of the trial through a questionnaire, based on National Health and Nutrition Examination Survey (NHANES) smoking-related items.

2.5.4. Emotional functioning

The SF-12 v2 mental health composite score⁷⁶ was used to assess overall emotional functioning. The Hospital Anxiety and Depression Scale (HADS) evaluated anxiety and depression.⁸² Symptom Checklist 90-Revised (SCL-90R) somatization subscale assessed a degree of somatic symptoms experienced by participants.⁴⁰ The Perceived Stress Scale (PSS) evaluated 14 sources of stress to produce an overall perceived stress rating.¹¹ The Coping Strategies Questionnaire Revised (CSQ-R)⁶² was used at screening to measure the frequency with which participants engaged in specific coping activities when experiencing pain, using a 7-point numerical scale ranging from 0 = “never do that” to 6 = “always do that”. The data for the catastrophizing subscale are presented.

2.5.5. Participant ratings of improvement

Participants’ ratings of global improvement were evaluated after 5 and 9 weeks by asking “Since beginning treatment at this study, how would you describe the change (if any) in activity limitations, symptoms, emotions, and overall quality of life related to your painful condition?” Response options were from the 7-point Patient Global Impression of Change (PGIC) scale³³ (1 = “no change or worse”, 2 = “almost the same”, 3 = “a little better”, 4 = “somewhat better”, 5 = “moderately better”, 6 = “better, and a definite improvement”, and 7 = “a great deal better”). For analyses, this endpoint was dichotomized by combining scores from 1 to 4 in one category of “No improvement” and scores from 5 to 7 in another category of “Moderate or greater improvement”.

2.5.6. Safety, concomitant medication coding, and compliance

Safety endpoints were collected at each visit and included adverse events (AEs), vital signs (systolic and diastolic blood pressure and heart rate), and concomitant/rescue medication use. The AEs were surveyed through a semi-structured review of organ systems. In addition, eight AEs, commonly associated with the use of propranolol, were systematically assessed through a structured questionnaire. These AEs included dizziness, unusual fatigue, nausea, diarrhea, constipation, hands numb or tingling, sleep problems, and depression. The Common Terminology Criteria for Adverse Events (CTCAE) version 4.0 were used for grading of AEs and the Medical Dictionary for Regulatory Activities (MedDRA) version 18.0 was used for coding of AEs.

At screening, a 12-lead ECG was reviewed by site cardiologists for exclusion of clinically significant cardiovascular abnormality. Vital signs were measured with the Accutorr Plus (Datascope Corp., NJ) and computed at each visit as a mean of 3 measurements taken at 2-minute intervals after a participant rested in a seated position for 10 minutes.

At each visit, participants reported their use of all medications, including the study drug. Non-study medications were coded using the World Health Organization Drug Dictionary (WHODrug) version 2015.01. Study drug compliance was primarily assessed through counts of the returned capsules at each visit and through participant daily entries of drug usage in the DSDs as a backup.

2.5.7. Quantitative sensory testing (QST)

Heat pain threshold and tolerance were assessed on the ventral forearm using commercially available thermal stimulators (Pathway or TSA-II; Medoc; Ramat Yishai, Israel) accessorized with a 16×16 mm thermode. Heat pain threshold was defined as the temperature at which the participant first perceived heat pain, whereas heat pain tolerance was defined as the temperature at which the participant could no longer tolerate the pain. The temperature increased from a baseline of 32°C with a 0.5°C/s rate of rise until the participant responded by pressing the button. The cutoff temperature for both measurements was 50°C. Average thermal threshold and tolerance were calculated from 4 assessments conducted with a 5-second interstimulus interval at different sites of the ventral forearm. For all heat tests, participants were first given practice runs on a site distant from subsequent testing to verify the participant’s understanding of the procedure.

Pressure Pain Thresholds (PPTs) were assessed bilaterally over temporalis, masseter, and trapezius muscles, TMJs, and lateral epicondyles with a pressure algometer (FDX-10, Wagner Instruments, CT). The PPTs were defined as the amount of pressure at which the participant first perceived the stimulus to be painful. One pre-trial assessment was performed at each site, followed by additional assessments until 2 measures differing by less than 0.2 kg were obtained, or 5 assessments were administered. In either case, the mean of the 2 closest values was recorded as the threshold estimate. Pressure stimuli were delivered at an approximate rate of 1 kg/s. The cutoff pressure for all body sites was 5 kg. The values from the right and left sides were averaged to obtain a single PPT value per anatomical site.

2.6. Statistical analysis

The main goal of the statistical analysis was to estimate change in study endpoints and evaluate efficacy of propranolol by testing a priori hypotheses concerning treatment group differences. Descriptive statistics for safety measures were also generated. Statistical analysis was performed using SAS v9.4 (SAS Institute Inc, Cary, NC).

2.6.1. Intention-to-treat, per-protocol, and safety samples

The primary and secondary analyses applied the intention-to-treat (ITT) principle by analyzing all available data from all randomized participants according to their treatment allocation. Specifically, the ITT sample comprised data from all post-randomization visits through week 9, in which there was a valid measurement of the primary endpoint, namely, the facial pain index. By necessity, the ITT sample excluded any post-randomization visits in which there was no valid value for the primary endpoint. Because mixed models do not require imputation for missing data, individuals with some missing endpoints were retained in the analyses.

Primary efficacy analysis was repeated in the per-protocol sample, defined as the subset of the ITT sample after excluding: a) assessments from any visits made at or after a major protocol deviation, and b) participants whose compliance with study medication was less than 60% over all study visits.

The safety analysis used data from all randomized participants who received at least one dose of study medication, and the treatment group was based on the study drug used, regardless of the allocated treatment.

2.6.2. Evaluation of efficacy hypotheses

The null hypothesis for the primary test of efficacy was that the mean change in the pain index at week 9 did not differ between treatment groups. Because there was only one test, the threshold to judge statistical significance was set at P<0.05 using a 2-tailed test. It was evaluated with a mixed model for repeated measures (MMRM) using the MIXED procedure in SAS. The dependent variable was the change in the mean pain index, and independent variables were fixed effects for study site (two dummy variables to account for the three study sites), sex (binary variable), race (dichotomized as White or non-White), the mean pain index at randomization (a continuous covariate), visit (the two dummy variables to account for the repeated visits at weeks 1, 5, and 9), allocated treatment group (binary variable), and a visit × treatment interaction. Participants were included as a random effect in the model, which used an unstructured variance–covariance pattern, and the denominator degrees of freedom for any sample imbalances were adjusted using Kenward Roger’s method. An LSMESTIMATE statement calculated the adjusted mean, standard error, and 95% confidence limits (95% CLs) for the difference between treatment groups at week 9, and the t statistic was calculated for inferences.

Additional analysis of the pain index dichotomized the endpoint using established definitions of a clinically meaningful pain reduction,^18,25 namely, a reduction in the mean pain index of ≥30% or ≥50% relative to the participant’s mean pain index at randomization. Each binary endpoint was modeled with a log-binomial model that used the generalized estimating equations (GEE) approach in SAS’s GENMOD procedure. This is analogous to the mixed model used for the continuous variable, except that GEE fits a marginal model to longitudinal data, and hence regression parameters are population-averages of the treatment effect. For the GEE model, participants were the repeated effect, and covariates were the same terms as used in mixed models, including a visit × treatment interaction. The identity link for the binomial distribution was specified, meaning that parameter estimates represent the adjusted net difference in probability of responding. Estimates, standard errors, and 95% CLs were calculated and the F-statistic was used to calculate P-value. The number needed to treat (NNT) was also calculated as the inverse of each parameter estimate and its 95% CLs. For comparison with other studies, odds ratios (ORs) and 95% CLs were also calculated by specifying the logit distribution for the binary distribution for the same GEE model. Consistent with the a priori statistical analysis plan, no critical threshold was nominated for P-values, and no adjustment was made for multiple statistical tests. Secondary endpoints were modeled according to their scale of measurement: either MMRM for continuous variables (e.g., pain intensity) or GEE models for binary variables (e.g., PGIC and all other dichotomized response analyses).

2.6.3. Safety evaluation

For the safety analysis, adverse events were aggregated across visits and the data were analyzed per participant. The proportion of participants with each category of adverse event was calculated and P-values for treatment group differences were computed using exact tests for binary logistic regression in SAS’s LOGISTIC procedure.

2.6.4. Sample size justification

The target sample size of 200 randomized participants was selected after calculating that it would provide 90% power to detect a 27% difference between treatment groups in the primary endpoint (i.e., mean reduction in pain index of 8.1 in the propranolol group versus 2.7 in the placebo group), assuming standard deviation of 11.9. Those estimates were deemed to be clinically meaningful and credible, based on results from our pilot study that used a randomized cross-over design to test efficacy of propranolol versus placebo.⁶⁹

3. Results

3.1. Study participants

A total of 200 participants were randomized in a 1:1 ratio to the propranolol (n=100) and placebo (n=100) treatment groups (Fig. 2). Overall, 174 participants (87%) completed the study, with equal percentage of completers in each group (87%). Main reasons for the trial discontinuation were withdrawal of consent (n=14) and loss to follow-up (n=6). One participant did not provide any post-baseline data and was excluded from the ITT population. The per-protocol sample comprised 143 participants (71.5%). The most frequent protocol deviations leading to participant exclusion from the per-protocol analyses were: (1) use of OTC analgesics that exceeded the protocol definition of episodic use (n=15 and n=21 in the propranolol and placebo groups, respectively); (2) non-compliance with the study drug (n=11 and n=10 in the propranolol and placebo groups, respectively); and (3) inability to reach maintenance dose (n=9 and n=5 in the propranolol and placebo groups, respectively). The safety sample comprised all 200 randomized participants who received at least one dose of the study drug.

Figure 2.

Flow diagram of participants in the trial.

The participant baseline demographic and clinical characteristics in the ITT sample were similar among treatment groups (Table 1 and Table S1 in the Supplementary Materials). The majority of participants were middle age, white females. The mean time since onset of the TMD pain was 11 years and almost all participants suffered from both myalgia and arthralgia (93.0%). The mean weekly facial pain index was 30.6 (on the 0–100 scale) with the mean weekly facial pain intensity of 47.6 (on the 0–100 scale) and the mean weekly facial pain duration of 58.3% of waking day. On average, the participants reported 24 days with facial pain in the last 30 days and 41.7% of them had TMD GCPS grades from IIb to IV, signifying high pain intensity and moderate/severe disability. Based on an established definition of jaw opening lesser than 40 mm as limited,⁶⁴ the participants’ mean pain-free jaw opening was restricted (29.8 mm), while no limitation was found in their mean maximum unassisted or assisted opening (44.2 and 48.2 mm, respectively), indicating no structural TMJ limitation for opening in majority of the cases. In contrast, the reported mean limitation in jaw function measured by the JFLS was 2.6 (on the 0–10 scale), which is typical for a chronic painful TMD sample.⁵¹

Table 1.

Baseline demographic and clinical characteristics of participants (ITT population)^*

Characteristics	Placebo (n=99)	Propranolol (n=100)
Age, years	34.2 (13.29)	33.9 (12.19)
Sex, female	78 (78.8%)	77 (77.0%)
Race
White	71 (71.7%)	79 (79.0%)
Black	18 (18.2%)	12 (12.0%)
Asian	6 (6.1%)	5 (5.0%)
Other	4 (4.0%)	4 (4.0%)
Study site
University of North Carolina	45 (45.5%)	45 (45.0%)
University of Florida	30 (30.3%)	31 (31.0%)
University at Buffalo	24 (24.2%)	24 (24.0%)
BMI, kg/m2	29.5 (11.00)	29.1 (14.40)
FACIAL PAIN
Time since onset, years	10.5 (9.43)	11.3 (9.85)
TMD myalgia and arthralgia	92 (92.9%)	93 (93.0%)
Weekly pain index, 0–100 scale	31.2 (18.94)	30.0 (19.96)
Weekly pain intensity, 0–100 scale	47.4 (15.58)	47.9 (15.23)
Weekly pain duration, 0–100 scale	60.1 (24.61)	56.6 (26.57)
Painful days in the last 30 days, n	24.0 (6.64)	23.7 (7.21)
Pain-free jaw opening, mm	29.6 (11.58)	30.0 (10.79)
Maximum unassisted jaw opening, mm	44.5 (10.08)	43.9 (8.44)
GCPS grade, IIb-IV	41 (41.4%)	42 (42.0%)
PHYSICAL FUNCTIONING
Migraine	53 (53.5%)	51 (51.0%)
HIT-6, 36–78 scale	55.2 (8.70)	54.4 (9.10)
AUDIT, 0–40 scale	2.3 (1.73)	2.5 (2.08)
Current smokers	18 (18.2%)	14 (14.0%)
EMOTIONAL FUNCTIONING
HADS anxiety, 0–21 scale	7.5 (4.59)	6.9 (4.19)
HADS depression. 0–21 scale	3.3 (3.14)	3.7 (3.54)
CONCOMITANT THERAPY
Oral appliances	19 (19.2%)	15 (15.0%)
TMD-specific NSAIDs	6 (6.1%)	9 (9.0%)
Non-TMD specific NSAIDs	12 (12.1%)	13 (13.0%)
Antidepressants	17 (17.2%)	23 (23.0%)
Benzodiazepines	4 (4.0%)	9 (9.0%)
Muscle relaxants	2 (2.0%)	3 (3.0%)

^*Data are means (SD) or numbers (%).

Abbreviations: AUDIT, Alcohol Use Disorders Identification Test; BMI, body mass index; GCPS, Graded Chronic Pain Scale; HADS, Hospital Anxiety and Depression Scale; HIT-6, Headache Impact Test-6; ITT, intention to treat; NSAIDs, nonsteroidal anti-inflammatory drugs; SD, standard deviation; TMD, temporomandibular disorder.

At baseline, 104 participants (52.3%) met criteria for definite or probable migraine and reported a mean HIT-6 score of 54.8, consistent with substantial headache-related disability. The mean baseline values for physical/emotional functioning and vital signs were within normal limits. For TMD treatment, orofacial appliances were used at night by 17.1% of participants and NSAIDs by 7.5%. In addition, 12.6% of participants used NSAIDs for other pain conditions. A substantial proportion of participants was taking antidepressant medications (20.1%).

3.2. Efficacy

The reduction in a mean weekly facial pain index was slightly greater in the propranolol group compared with the placebo group at week 9, although the difference was marginal and not statistically significant (difference vs placebo: −1.8, 95% CL: −6.2, 2.6) (Table 2 and Figure 3A). The primary endpoint in the per-protocol sample also did not differ between treatment groups (difference vs placebo: −0.5; 95% CL: −5.1, 4.1). However, the ≥30% responder rate for the propranolol group at week 9 was greater compared with placebo (69.0% vs 52.6%; OR = 2.1, 95% CL: 1.1, 3.9) with a NNT = 6.1 (95% CL: 3.2, 55.3) (Table 2 and Figure 3B). Likewise, the ≥50% responder rate at week 9 was greater in the propranolol than the placebo group (55.5% vs. 39.2%; OR = 2.0, 95% CL: 1.1, 3.7) with a NNT = 6.1 (95% CL: 3.2, 69.7) (Table 2 and Figure 3C). Because the responder rate and NNTs can vary substantially depending on the response cutoff point used, we present a cumulative distribution function curve and NNTs across the entire range of the response, as was recommended by Farrar and colleagues.^23,24 Shown in Figure S1A in the Supplementary Materials, the proportion of responders for pain index is consistently higher in the propranolol group compared with placebo across all cutoffs and the area under the curve (AUC) for propranolol is 18.3% greater than for placebo over a therapeutically important range of response, such as 20% to 70%.

Table 2.

Summary of primary and secondary endpoints (ITT population)

Endpoints	Placebo (n=99)	Propranolol (n=100)	P value
PRIMARY ENDPOINT
Mean weekly pain index, baseline to week 9
Change from baseline, mean (95% CL)a	−12.1 (−15.5, −8.7)	−13.9 (−17.4, −10.5)
Difference vs placebo, mean (95% CL)		−1.8 (−6.2, 2.6)	0.414
≥30% Reduction in mean weekly pain index, baseline to week 9
Subjects achieving response, % (95% CL)b	52.6 (41.4, 63.8)	69.0 (57.8, 80.2)
Odds ratio (95% CL)c		2.1 (1.1, 3.9)	0.028
Number needed to treat (95% CL)b		6.1 (3.2, 55.3)	0.028
≥50% Reduction in mean weekly pain index, baseline to week 9
Subjects achieving response, % (95% CL)b	39.2 (28.5, 49.8)	55.5 (43.9, 67.1)
Odds ratio (95% CL)c		2.0 (1.1, 3.7)	0.032
Number needed to treat (95% CL)b		6.1 (3.2, 69.7)	0.031
SECONDARY ENDPOINTS Mean weekly pain intensity, baseline to week 9
Change from baseline, mean (95% CL)a	−13.6 (−17.6, −9.7)	−17.1 (−21.1, −13.1)
Difference vs placebo, mean (95% CL)		−3.5 (−8.6, 1.7)	0.187
Mean weekly pain duration, baseline to week 9
Change from baseline, mean (95% CL)a	−16.6 (−21.3, −11.8)	−17.9 (−22.8, −13.0)
Difference vs placebo, mean (95% CL)		−1.4 (−7.6, 4.9)	0.665
PGIC score dichotomized, baseline to week 9
Subjects achieving response, % (95% CL)b	27.7 (16.8, 38.6)	43.4 (31.5, 55.3)
Odds ratio (95% CL)c		2.0 (1.1, 3.7)	0.032
Number needed to treat (95% CL)b		6.4 (3.3, 68.1)	0.031
Pain-free jaw opening, baseline to week 9
Change from baseline, mean (95% CL), mma	2.3 (0.6, 4.1)	5.5 (3.7, 7.4)
Difference vs placebo, mean (95% CL), mm		3.2 (1.0, 5.4)	0.005
Maximum unassisted jaw opening, baseline to week 9
Change from baseline, mean (95% CL), mma	0.3 (−1.0, 1.7)	0.8 (−0.6, 2.2)
Difference vs placebo, mean (95% CL), mm		0.4 (−1.2, 2.1)	0.584
HIT-6 score, baseline to week 9
Change from baseline, mean (95% CL)a	−3.1 (−4.6, −1.7)	−5.1 (−6.6, −3.6)
Difference vs placebo, mean (95% CL)		−1.9 (−3.7, −0.2)	0.034

^aEstimates from a mixed-model repeated-measures (MMRM) analysis adjusted for treatment, visit, treatment-by-visit interaction, study site, sex and race as fixed effects, with a baseline value of the endpoint as a covariate and unstructured covariance matrix. Participant was a random effect. No adjustment for multiple comparisons was made.

^bEstimates from generalized estimating equations for log-binomial models adjusted for treatment, visit, treatment-by-visit interaction, study site, sex and race as fixed effects, with a baseline value of the endpoint as a covariate.

^cEstimates from generalized estimating equations for logistic models adjusted for treatment, visit, treatment-by-visit interaction, study site, sex and race as fixed effects, with a baseline value of the endpoint as a covariate.

Abbreviations: 95% CL, 95% confidence limits; HIT-6, Headache Impact Test-6; ITT, intention to treat; PGIC, Patient Global Impression of Change.

Figure 3.

Regression-model estimates of four endpoints: A. Mean pain index; B. Percentage of participants with at least 30% reduction in pain index; C. Percentage of participants with at least 50% reduction in pain index; D. Percentage of participants reporting overall improvement in their pain condition. The pain index represents the product of pain intensity (0–100 numerical rating scale) multiplied by pain duration (0–100% of day), divided by 100. It was recorded using daily symptom diaries completed prior to study visits at baseline (week 0) and up to three follow-up visits occurring 1, 5 and 9 weeks after initiating treatment with either propranolol 60 mg, BID () or placebo (). Percentage reductions in pain index were calculated relative to baseline and dichotomized to signify the percentage of participants with ≥30% reduction and ≥50% reduction. Global impression of change was reported on a 7-point scale at visits 3 and 4, and dichotomized to signify the percentage of participants with either “moderate”, “definite” or “a great deal” of improvement. Adjusted means were estimated using mixed model for repeated measures with covariates baseline pain index, study visit, treatment group, visit x treatment group interaction, study site, gender, and race/ethnicity. Adjusted percentages and their 95% confidence limits were estimated with a log-binomial regression model using the generalized estimating equation method allowing for repeated visits by study participants. Covariates were as described for the mixed model. For all models, treatment-group difference in means or percentage at week 9 were computed with custom estimates of predicted population margins based on parameters in the model. Differences in percentages were used to compute a number-needed-to-treat (NNT) and its 95% confidence limits. P-values tested the null hypothesis that the means or percentages at week 9 are equivalent in the two treatment groups.

The effect of propranolol on the mean weekly pain intensity (difference vs placebo: −3.5; 95% CL: −8.6, 1.7) and the mean weekly pain duration (difference vs placebo: −1.4; 95% CL: −7.6, 4.9) was not significant (Table 2). The correlation between pain intensity and pain duration was moderate to large, with Pearson correlation coefficients ranging from 0.49 at baseline to 0.71 at week 9. For the purpose of comparison with other studies that have used pain intensity as a primary endpoint, we also present results from a responder analysis and a cumulative distribution function curve using the pain intensity endpoint (Table S2 and Figure S1B in the Supplementary Materials). For a given threshold within a given treatment group, the pain intensity endpoint yielded a lower percentage of responders than the pain index. However, differences between treatment groups (i.e., the measures of treatment efficacy in a randomized clinical trial) were fairly consistent for the two endpoints, as were treatment-group differences in the AUCs. Across the clinically important threshold range from 20% to 70%, the 17.7% treatment-group difference in AUC values for pain intensity was comparable to the 18.3% difference for pain index. Compared with pain index, pain intensity displayed a greater degree of variation in threshold-specific NNTs. The NNT of 20.1 at the 50% threshold appeared to be an outlier, due primarily to a threshold-specific spike in the proportion of placebo responders.

The analysis of the dichotomized PGIC at week 9 showed that a greater proportion of participants reported a moderate or greater global improvement in the propranolol group compared with placebo (43.4% vs. 27.7%; OR = 2.0, 95% CL: 1.1, 3.7) with a NNT = 6.4 (95% CL: 3.3, 68.1) (Table 2 and Figure 3D). The percentages of patients endorsing any of 7 response options in each treatment group are presented in Table S3 in the Supplementary Materials. The PGIC ratings at week 9 were significantly greater in the propranolol group (difference vs placebo: 0.82, 95% CL: 0.28, 1.37, P = 0.003), indicating a nearly 1-point improvement on the 7-point PGIC ordinal scale.

The results of the analyses for other secondary endpoints at week 9 are shown in Table 2 and Table S2 in the Supplementary Materials. Among other facial pain-related secondary endpoints, there was a significant improvement in pain-free jaw opening (difference vs placebo: 3.2 mm; 95% CL: 1.0, 5.4) although no significant difference in maximum unassisted or assisted jaw openings or in a JFLS score emerged. Among endpoints associated with physical or emotional functioning, headache-related disability as measured by the HIT-6 questionnaire was the only endpoint with a significant reduction after propranolol treatment (mean group difference vs placebo: −1.9; 95% CL: −3.7, −0.2). Among QST endpoints, a significant difference was observed only for the heat threshold (difference vs placebo: 0.7; 95% CL: 0.0, 1.5).

3.3. Adverse events, compliance, and masking

At least one adverse event was reported for 86% participants receiving propranolol and 75% of those receiving placebo (Table 3). Serious adverse events (SAEs), severe adverse events, and adverse events leading to discontinuation were infrequent. SAEs occurred in 1% of participants given placebo and in 4% of those given propranolol, including 1% of participants in the propranolol group who reported an SAE prior to the treatment period (Table S4 in the Supplemental Materials). Each specific SAE occurred in no more than one participant. One death due to metastatic breast cancer occurred in the propranolol group. Except for the events in a participant with fatal breast cancer, all SAEs were resolved by the end of the trial.

Table 3.

Adverse events in the safety population^a

Adverse Events	Number (%) of Participants
	Placebo (n = 100)	Propranolol (n = 100)	P valueb
All events
Adverse events	75 (75.0%)	86 (86.0%)
Treatment-emergent adverse eventsc	70 (70.0%)	84 (84.0%)
Serious adverse events	1 (1.0%)	4 (4.0%)
Adverse events leading to study discontinuation	0 (0.0%)	1 (1.0%)
Severe treatment-emergent adverse events	4 (4.0%)	6 (6.0%)
Death	0 (0.0%)	1 (1.0%)
Treatment-emergent adverse events by typed
Gastrointestinal disorders
Diarrheae	16 (16.0%)	18 (18.0%)	0.851
Nauseae	6 (6.0%)	14 (14.0%)	0.097
Constipatione	9 (9.0%)	10 (10.0%)	1.000
General disorders
Fatiguee	16 (16.0%)	29 (29.0%)	0.041
Infections and infestations
Nasopharyngitis	10 (10.0%)	12 (12.0%)	0.822
Upper respiratory tract infection	4 (4.0%)	7 (7.0%)	0.537
Nervous system disorders
Dizzinesse	6 (6.0%)	21 (21.0%)	0.003
Paraesthesiae	9 (9.0%)	5 (5.0%)	0.407
Hypoesthesiae	5 (5.0%)	4 (4.0%)	1.000
Headache	1 (1.0%)	5 (5.0%)	0.212
Psychiatric disorders
Insomniae	10 (10.0%)	13 (13.0%)	0.658
Depressione	9 (9.0%)	7 (7.0%)	0.795
Sleep disordere	0 (0.0%)	6 (6.0%)	0.029
Musculoskeletal and connective tissue disorders
Back pain	5 (5.0%)	2 (2.0%)	0.445
Ear and labyrinth disorders
Vertigo	7 (7.0%)	5 (5.0%)	0.767

^aThe safety population included all participants who were randomized and received at least 1 dose of the study drug. If a participant had multiple types of adverse events, he/she was counted once for each type. If a participant had multiple events of the same type, he/she was counted once for that type.

^bP values were computed using exact logistic regression with adverse event incidence as the dependent variable and treatment group as the independent variable.

^cTreatment-emergent adverse events were defined as adverse events occurring during a treatment period and included adverse events considered to be related as well as those not considered to be related to the study treatment.

^dData are presented for treatment-emergent adverse events reported by ≥5% of participants in any treatment group and coded by System Organ Class and Preferred Term from the MedDRA v18.0.

^eExpected adverse events systematically surveyed through the symptom questionnaire.

Most of the treatment-emergent adverse events were mild to moderate in severity and had similar incidence in both groups, with exception of fatigue, dizziness, and sleep disorder that were more frequent in the propranolol group (Table 3). As expected, propranolol reduced the values of vital signs but the absolute reductions were small (Table S2) and well tolerated.

Participant compliance with the study drug was very good: only 11% of participants receiving propranolol and 10% of those receiving placebo took < 60% of the study drug, specified in the study protocol as non-compliance. On average, participants in the propranolol group used 88% of the study drug, while participants in the placebo group used 89%. Rescue medication (OTC analgesics) was taken by 24% of participants receiving propranolol and 19% of those receiving placebo. Only 8% of participants exceeded the protocol definition of episodic use of rescue medication in each group. During the treatment period, the mean number of days with the rescue medication per participant was low: 3.7 days (5.6% of the treatment days) and 1.1 day (1.9% of the treatment days) in the propranolol and placebo groups, respectively.

At week 9, 168 (84.4%) participants provided information about their perceived treatment allocation: 48 (55.8%) and 54 (65.9%) participants in the placebo and propranolol groups, respectively, correctly identified their treatment. The participant agreement in guessing their allocation was poor (kappa = 0.22, 95% CL: 0.07, 0.36), indicating successful participant masking.

4. Discussion

Among participants with painful TMD, there was no signficant treatment group difference in the primary endpoint of facial pain index after 9 weeks of treatment. Yet, when the same endpoint was dichotomized, proportions of responders were greater in the propranolol group than the placebo group across the entire clinically important range of 20% - 70% response. Cumulative response curves for the more traditional endpoint, pain intensity, yielded a similar treatment group difference favoring propranolol, although in the propranolol group there was a higher probability of response for the pain index than for pain intensity. For example, among participants receiving propranolol, 59% had a ≥50% reduction in pain index but only 36% reported a ≥50% reduction in pain intensity. The difference between these endpoints is important to understand for interpretation of the trial results. Additionally, the proportion of participants who reported a moderate or greater global improvement in their TMD was also greater for propranolol. Hence, the overall findings are consistent with a clinically meaningful benefit of propranolol in reducing facial pain. Likewise, nominally significant results (i.e., P < 0.05) were obtained in secondary analyses comparing the treatment groups with respect to pain-free jaw opening, headache-related disability, and sensitivity to heat stimulation.

The apparent paradox in treatment effect estimates of pain index (or intensity) analyzed as a continuous variable (i.e., analysis of a mean change) vs a dichotomized variable (i.e., responder analysis) stems from a group difference in distributions of a change (Figure S2 in the Supplementary Materials). In both treatment groups, a critical mass of responses was concentrated around thresholds of 30% and 50% reductions. However, noticably more participants in the placebo group compared to the propranolol group fell short of these thresholds. In contrast to the usual situation where responder analysis is associated with less statistical power, the distribution led to more statistical power for the responder analysis.⁶⁶ FDA and IMMPACT guidelines encourage responder analysis because it estimates a clinically meaningful change in the endpoint at the individual level, which is not the case for analysis of group means.^18,47

A summary statistic for evaluating the difference between groups and comparing it between studies is the NNT.¹² The observed NNTs for standard 30%/50% thresholds of pain index, 30% threshold of pain intensity, and PGIC were approximately 6, implying that six patients need to be treated with propranolol in order for one of them to benefit because of the drug. Since the NNT is calculated relative to the control group, it relates specifically to improvement because of the drug and not because of other reasons, such as placebo response or spontaneous remission. Regrettably, NNTs have not been widely reported for studies of TMD pharmacotherapy: the Cochrane review summarized efficacy only in terms of average pain reductions.⁴⁹ However, in clinical trials of patients with another muscle pain disorder, fibromyalgia, the average NNT for a ≥30% pain reduction was 10 for serotonin-norepinephrine reuptake inhibitors (SNRIs) duloxetine and milnacipran, 7.2 for pregabalin, and 4.9 for tricyclic antidepressants, while the NNTs for a ≥50% reduction were even higher.^15,30,77 The average NNT for a moderate benefit in the PGIC ranged from 5 for SNRIs to 11 for pregabalin.^15,77 Therefore, the NNT estimates in this study represent a clinically important outcome.

One downside of an NNT is its reliance on specific thresholds of a continuous measure and sensitivity to arbitrary deviations at these thresholds. As seen in the cumulative response curves for pain intensity, the 50% threshold yielded a higher NNT than other NNTs across nearby thresholds of 40–60% because of the threshold-specific surge in the number of placebo responders. To monitor for such aberrations, it is important to review NNTs across an entire range of clinically relevant thresholds.

As with previous trials of propranolol, the most common adverse events were diarrhea, fatigue, dizziness, and insomnia.⁶⁷ Most were mild to moderate in severity and their rates were similar between treatment groups, except for fatigue, dizziness, and sleep disorder that were more frequent in the propranolol group. The higher rates of adverse events, compared with previous trials,⁴² can be attributed to this study’s active surveillance of harms which yields more adverse events than passive surveillance.^5,34

Propranolol is a non-selective β-adrenergic receptor antagonist with similar affinity for β₁ and β₂-adrenoreceptors and lower affinity for β₃-adrenoreceptors.² It is used for treatment of hypertension, ischemic heart disease, arrhythmias, migraine, and anxiety.^3,36 Analgesic effect of propranolol in musculoskeletal pain is plausible, given evidence for an enhanced adrenergic tone in chronic pain states. For example, increased levels of epinephrine and norepinephrine are found in patients with TMD and fibromyalgia^22,72 and sympathetic dominance (parasympathetic suppression) is reported in patients with TMD, fibromyalgia, and arthritis.^46,56,73 Stimulation of the cervical sympathetic nerve leads to development of tension in jaw muscles in animal models, also implicating sympathetic input in development of TMD myalgia.⁵⁴ Therefore, β-adrenergic blockade could exert an analgesic effect by attenuating this adrenergic dysregulation. Indeed, in various animal models, propranolol produced anti-nociception, although different studies identified various β-adrenoreceptor subtypes responsible for the effect.^26,37,50,61 Potential mechanisms may include a direct action on peripheral nociceptors^10,37,61 and interactions with immune^{20,29,68,71,80,81} and cardiovascular systems.^4,7,41,57

Because propranolol crosses the blood-brain barrier, it can influence pain processing centrally. In migraine animal models, propranolol inhibits trigeminal nociception through antagonism of β₁-adrenergic receptors on thalamocortical neurons⁶⁵ and blocks chronic sensitization of descending pain pathways from higher brain regions to trigeminocervical complex.⁶

Propranolol can also alter affective modulation of pain by reducing anxiety.^59,60 In rodents, β-adrenergic blockade prevents anxiety-like behavior and reduces negative affective components of pain.^16,79 In humans, propranolol is used to treat anxiety,^36,48,55 and anxiety is associated with increased odds of TMD.²⁷ While baseline levels of anxiety in this study were quite low, there was a trend toward greater reduction in anxiety for the propranolol group compared with the placebo group (P = 0.111).

To our knowledge, only three studies have assessed propranolol’s effect in TMD patients. Consistent with our findings, Light et. al.⁴¹ noted lower ratings of TMD pain after intravenous administration of a single low dose of propranolol, compared with placebo. In our pilot trial⁶⁹ with a crossover design, we found a significant reduction in pain index after one week of oral administration of 40 mg of propranolol, although the magnitude of effect was modest. In a placebo-controlled trial with a factorial design, Gonçalves et. al.²⁸ used 90 mg of short-acting propranolol with and without an oral appliance for 3-month treatment of migraine patients with concomitant TMD. It is difficult to compare our primary result with the analyses of TMD endpoints from that trial because Gonçalves et. al.²⁸ focused on migraine-related endpoints and their exploratory TMD-related endpoints did not include weekly TMD pain ratings. The TMD endpoints, comparable to our trial, were pressure pain thresholds and maximum unassisted opening, for which neither trial found significant group differences.

The strengths of the study are: (1) use of validated DC/TMD examination for selecting participants with TMD myalgia and arthralgia and for monitoring diagnostic status; (2) use of daily diaries for facial pain assessment; (3) relatively low dropout rate; (4) thorough assessments of rescue and concomitant medications; (5) active surveillance of harms; and (6) participants’ excellent compliance with the daily diaries and study drug.

This study has several limitations. First, this phase 2b trial was not powered for assessments of all secondary endpoints. To better characterize our patient population and understand treatment effects, we collected many secondary endpoints encompassing several categories for exploratory analysis, as described by D’Agostino¹³ and endorsed in IMMPACT recommendations.⁷⁴ Specifically, the secondary endpoints (1) provided background information, (2) served as components of the composite primary endpoint, (3) could aid in understanding the mechanisms of action of the treatment, (4) were related to secondary hypotheses that were not major objectives of the treatment, and (5) were intended for exploratory analyses. Given the variety of secondary endpoints and a limited study size, no adjustment for multiplicity was performed. Therefore, results for these secondary endpoints should be interpreted with caution. Second, this trial was limited to evaluation of endpoints at a follow-up of 9 weeks after randomization and a longer follow-up could be beneficial. Third, this study did not include certain patient populations, such as patients with comorbid hypertension, fibromyalgia, hyperthyroidism, or opioid medication use. Further studies are needed to inform the use of propranolol in these populations.

To achieve results generalizable to clinical care settings, we enrolled participants of any sex, race, and ethnicity and they were allowed to maintain preexisting treatments during the trial. The exclusion criteria were limited to propranolol’s contraindications and health conditions that could have affected pain ratings, for example comorbid hypertension, fibromyalgia, or hyperthyroidism.

In conclusion, among patients with chronic painful TMD, propranolol had a statistically non-significant effect, compared with placebo, on mean change in facial pain index but propranolol was efficacious in achieving ≥30% and ≥50% pain index reductions and in improving participants’ ratings of global change after 9 weeks of treatment. The drug was safe and well tolerated. These results provide sufficient evidence to justify further investigation of propranolol for TMD management.