Trazodone in Burning Mouth Pain:A Placebo-Controlled, Double-Blind Study

Tuulikki Tammials-Salonen, DDSSpecialist in Clinical DentistryInstitute of DentistryTurku University

Heli Forsseii, DDS, PhDSenior LecturerDepartment of Oral DiseasesTurku University Central Hospital

Turku, Finland

Correspondence to:Dr Heli ForsseilDepartment of Oral DiseasesTurku University Central HospitalLemminkaisenkatu 2FIN-20520 Turku, FinlandE-mail: heli.forssell@utu.fi

Aims; An 8-week parallel, placebo-controlled, double-blind trialevaluated the efficacy of the antidepressant trazodone in the treat-ment of chrome burning mouth pain. Methods: Thirty-seven care-fully selected women aged 39 to 71 (mean 58.6 years) were ran-domized to receive either 200 mg of trazodone or a placebo in asimilar manner. Pain and pain-related symptoms were evaluatedon a visual analogue scale and other measures at 0, 2, 4, and 8weeks. Results: There were no significant differences between thegroups in treatment effects for pain or pain-related symptoms.Seven patients in the trazodone group and 2 in the placeho groupfailed to complete the trial because of side effects. The most com-mon side effects were dizziness and drowsiness. Conclusion: Inthis controlled trial, trazodone failed to relieve burning moutbpain.J OROFAC PAIN 1999; 13:83-8 s.

Key words: burning moutb, oral mucosal pain, chronic pain,trazodone, antidepressants

Burning moutb syndrome (BMS) is characterized by a burningmucosal pain witbout clinically detectable lesions on tbeoral mucosa. Tbe pain is typically reported at more than one

oral site, the tongue, the palate, and the lower lip mucosa beingtbe most frequently affected sites.'-' Patients witb BMS also fre-quently complain of dry mouth and loss of or altered taste.^ Tbeprevalence of prolonged oral burning is about 8%, with less tban1% of those witb the disorder suffermg from constant burningpam symptoms."* The etiology and pathophysiology of BMS isunknown, but many causative factors of a local, systemic, or psy-cbogenic nature have been proposed, including hyposalivation,oral candidiasis, the preanemic stage of nutritional deficiencies[especially of iron, Bp vitamins, and folate), hyperglycemia, andhormonal cbanges at menopause. However, controlled studieshave not provided confirmation for the proposed etiologies,' andtreatment based on any of tbese etiologies is usually unsatisfac-tory.^'^ Burning moutb syndrome can be considered a chtonic painsyndrome, with symptoms that usually last from several months toseveral years.̂ •'̂ The pain is experienced as quantitatively similar

Journal of Orofacial Pain 8 3

Tanimiala-Salonen/Forssell

but qualitatively different from toothache pain.The verbal descriptor most commonly chosen byBMS subjects is "burning."^ Burning pain is con-sidered characteristic of chronic neuropathic painconditions in general, and some recent data seemto suggest peripberal or central nervous systeminvolvement as a possible underlying factor in

Antidepressants are frequently used for thetreatment of different chronic pain conditions,such as chronic neuropathic pains. In particular,tricyclic antidepressants have been shown to beeffective in the treatment of different pain condi-tions.'̂ "- '̂ The use of tricyclic antidepressants, eg,amitriptyline, has also been recommended in thetreatment of BMS pain,'-'-'^ but their anticholiner-gic side effects, especially dry mouth, may makethem unsuitable. The evidence of the efficacy ofnewer antidepressants in chronic pam is insuffi-cient, but their use has been recommended in casesin which tricyclics cannot be used because of theiradverse effects,'^

Trazodone is a triazolopyridine derivative that ischemically and pharmacologically unrelated toother currently available antidepressants. Its phar-macologie activity IS thought to be modulated viaantagonism at serotonin 5-HT, receptors; the pre-cise mechanism of action in humans is, however,not fully understood."* It is reported to lackalmost totally any anticholinergic side effects,viihich can be considered an important aspect intrying to find effective pain medication for a painproblem that is often associated with the feeling ofa dry moutb. Ttazodone has been compared witbamitriptyline in patients with mainly a burningtype of pain. Both drugs appeared to be equallyeffective in relieving pain, but trazodone causedsignificantly less dry mouth.'* With this back-ground, the present study was designed to find outwhether trazodone would be effective and suitablefor the treatment of chronic BMS pain.

Material and Methods

The material of the study consisted of 37 women(aged 39 to 71 years, mean 58,6 years) selectedover a 5-year period (from 1992 to 1996) from146 consecutive female patients who were referredto the Department of Oral Diseases at TurkuUniversity Central Hospital because of oralmucosal burning pain. One of the investigatorsperformed a screening evaluation of all subjects.The patients underwent a thorough clinical exami-nation, including measurement of whole salivary

flow, blood samples (blood count and levels ofglucose, Bij vitamins, and folate), and diagnosis ofcandidiasis. The investigators asked patients aboutpain intensity and duration, overall health, andmedications.

Altogether, 35 patients whose pain could possi-bly be related to some physical finding wereexcluded to obtain a strictly diagnosed gtoup ofpatients with burning mouth pain (16 patients hadoral mucosal lesions, 14 patients had oral candidi-asis, 3 patients had hyposahvation, 1 had hyper-glycemia, and 1 had low serum folate). In addi-tion, 13 patients wbo used centrally acting painmedications or anxiolytics, as well as 2 patientswho wete considered to suffer from dementia,were excluded.

Criteria for inclusion were daily, or almostdaily, oral hurning pain that had lasted 6 monthsor longer and had a moderate to severe intensity(at least 30 on a visual analogue scale [VAS] of 0to 10Ü mm). Forty-one of the remaining patientswere not eligible, 30 because rhe pain did notoccur daily or almost daily, 8 because of mildpain intensity, and 3 because of recent onset ofthe pain problem. Eligible patients received infor-mation about tbe study, and those willing to par-ticipate were included after granting informed,written consent. Eighteen patients refused to par-ticipate.

The design of the trial was a double-blind, ran-domized, parallel-group, placebo-controlled study.Identical capsules of trazodone and of passiveplacebo were packed in the same way. Randomi-zation was performed in hlocks of 6 by the manu-facturer of tbe drug (Orion). The trial protocolwas approved by the Joint Commission on Ethicsof Turku University and Turku University CentralHospital,

Patients began taking one 100-mg capsule oftrazodone in the evening for the first 4 days, andafter this, the medication was increased to two100-mg capsules in the evening. Control patientstook the placebo in a similar manner. The triallasted for S weeks. Patients were evaluated at 0, 2,4, and 8 weeks by tbe other investigator.

Visual analogue scales were used at each timepoint to measure the intensity of BMS pain; theinfluence of the pain on eating, speaking, andsleeping; and the suffering caused by the pain. Atthe beginning and end of the trial (0 and 8 weeks),the intensity and character of the pain were fur-ther defined by the use of tbe Einnish version^*' ofthe McGili Pain Questionnaire.^' Possible depres-sion was measured with the Beck DepressionInventory. The pain threshold of the subjects was

84 Volume 13. Number2, 1999

Tammiala-Salonen/Forssell

assessed at the beginning of the trial hy askingthem to estimate the intensity of the pain thevexperienced when taking a regular blood test. Thepatients' assessment of theit total nnprovementand benefits of the treatment m relation to sideeffects were obtained separately at the end of thetrial.

During the trial, subjects kept diaries on possibleside effects and on other pain medication they hadtaken. Patient comphance was estimated by count-ing the capsules left in the hottle at the final visit.The randomization code was not opened duringthe trial. During the triai, nothing suggested thatthe blind had not remained intact for the patients.The examiner could not guess the treatment of thesubjects.

Statistical Analysis

Analysis of variance of repeated measurementswas used to analyze the treatment and time effects,as well as the interaction of treatment and time fornormally distributed responses. If necessary, thelogarithm or square-root transformation wasapplied. The analysis of variance was carried outwith a mixed procedure on SAS software (SASInstitute, Garry, NC), which allows missing datain follow-up and includes the possibility of model-ing the covariance structure of measurements. Forcategorical responses, the statistical comparisonhetween treatment and placebo was carried outwith the Fisher exact test separately at each timepoint. For ordinal scaled outcome variable, theKruskal-Wallis test was applied. P values of lessthan 0,0,5 were interpreted as significant, and thelevel in confidence intervals was 95%. Statisticalcalculations were performed with SAS System forWindows, release 6.12 (1996).

Results

Of 37 patients, 18 received trazodone and 19received a placebo. The mean duration of pain inthe trazodone group was 3,0 years (6 months to17 years) and in the placebo group it was 2,8 years(6 months to 20 years). The mean age of thepatients in the trazodone group (61,1 years) wasslightly higher than in the placebo group (56.2years). The groups had a baseline difference inpain intensity at the start of rhe medication, themean VAS in the trazodone group being 12,6 mmhigher than in the placebo group (P < 0.05).Patients randomized in the trazodone group con-sidered taking a blood test somewhat more painful

(mean VAS 7,6) than patients in the placebo group(mean VAS 5.4), but the difference was not statis-tically significant. There were no other baselinedifferences between the groups. No other painmedications for BMS pain were used by thepatients, but 2 patients in the trazodone group and2 in the placebo group used anti-inflammatoryanalgesics for other pain conditions regularly oralmost regularly.

Seven subjects in the trazodone group and 2 inthe placebo group failed to finish the trial becauseof side effects, mainly because of dizziness. In thetrazodone group, 4 subjects withdrew soon afterthe start of the trial and 3 subjects withdrew after2 to 4 weeks. In the placebo group, 1 subject with-drew during the first week and another withdrewafter 6 weeks. Based on a count of the pills, com-pliance was excellent among the patients whocompleted the trial.

The difference between the groups in treatmenteffect or treatment by time interaction for painintensity was not significant (Fig 1), The meandifference in pain reduction between the tra-zodone and placebo groups at 8 weeks, as shownhy the VAS, was ^ . 8 (95%, confidence interval,range; -20.3 to 10.7), However, there was a gen-eral reduction in the pain ratings. The mean painVAS in the trazodone group at the beginning ofthe study was 59.2, and at 8 weeks it was 45.3; inthe placebo group VAS values were 46.6 and34,3, respectively. The decrease in pain intensitywas significant (P < 0.01) in both groups. Therewere no significant differences between thegroups in treatment effects or treatment by timeinteractions for the influence of the pain on eat-ing, speaking, and sleeping, or for the sufferingcaused by the pain.

At the baseline, 17 subjects were identified bythe Beck Depression Inventory as depressed, usinga cut-off score of 13, There was no significanttreatment effect or treatment by time interactionon the depression score, but the scores decreasedsignificantly (P < 0.01) in both groups during thetrial. The subgroup of depressed patients did notdiffer from the wbole group with respect to anypain-related measurements, but the number ofdrop-outs was slightly higher (6 of 9) in this sub-group.

The mean score for the number of words chosenin the McGill Pain Questionnaire, which isthought to reflect the intensity of the pain experi-enced,'' was 8.2 in the trazodone group and 7.5 inthe placebo group at the baseline. There was nosignificant difference between the groups or withinthe groups at different time points.

Journal of Orofacial Pain 8 5

Tammiala-Salonen/Forssell

100

f 80E,£• 60

tu•E 40

£ 20

Trazodone Placebo

0 2 4

Time (weeks)

Fig 1 Graph showing changes in pain intensity. The mean values andstandard deviations of pain intensity on a 100-mm visuai analogue scaleIn trazodone and placeho groups are shown at different follow-up points.

Table 1 Side Effects Reported by Patients

Side effect

DizzinessDrowsinessAbdominal painsHeadachePalpitationsTremorDry mouthNightmaresUrinary incontinence

Treatment

Trazodone(n = 18)

1195322301

Placeho(n -19 )

124221120

Patients in the trazodone group reported signifi-cantly more dizziness [P < 0-001) and drowsiness[P < 0.05) than patients in the placebo group(Table 1)- Two patients in the trazodone groupand 8 in the placebo group reported no sideeffects. There were no significant differencesbetween groups in the patients' global assessmentof improvement or benefits of the treatment(Tables 2 and 3).

Discussion

Burning mouth syndrome remains an ill-definedcondition that presents both diagnostic and treat-ment problems.^'"^ We tried to select our patientmaterial very carefully according to the latestunderstanding of the nature of the problem andonly included patients for whom no obvious causefor the burning mucosal pain was evident in orderto get material with "true" burning mouth symp-toms-^^ For this reason, and because only patientswith chronic, frequent, and moderate to severe

86 Volume 13, Number 2, 1999

Table 2 Parients' Global Assessmenr ofimprovement at the Final Visit

Tammiala-Salonen/Forssell

Table 3 Patient Assessment of Benefits inRelation to Side Effects

Assessment

ImprovedNo effectWorse

Treatment

Trazodone( n = l l )

8 (73%)2(18%)1 (9%)

Placebo(11.17)

13 (76%)4 (24%)0 C0%)

Assessment

EfficientNeutralInconvenient

Treatment

Trazo do ne( n = l l )

6 (55%)4 (36%)1 (9%)

Placebo(n=16)

13(81%)3(19%)0 (0%)

intensity pain were included, rhe recrniting of thepatient material took a long time. Also, the greatnumber of patients who were unwilling ro partici-pate in the trial added to the problem. The reasonsfor refusal were various and included long-disrancetravel, negarive attitudes in general toward medi-carions, and ahility to cope with the pain. The painof those who refused was of a slightly shorterduration but had the same mean intensity as thatof the participants.

According ro the results of this trial, rrazodonefailed to relieve burning pain or pain related symp-toms. There were no significanr differences intreatment effects with trazodone compared ro theplacebo. Considering the range of the confidencelimits of the main outcome of this stndy, the con-clusion seems warranred since the reduction inpain VAS compared to placeho was +10,7 at itshest and -20,3 at its worst.

Our trial lasted 8 weeks, which should be longenough to detect treatment effects of antidepres-sants in pain treatment, Trazodone dosages used inrhe trearmenr of depression can vary: oiderpatients are reporred to rolerate dosages of up to300 or 400 mg per day.'^ We used a lower dosage,200 mg, which was comparable to that used inother studies on trazodone m pain patients andpossihly explains the lack of efficacy on depressionin this study. Ventafridda er al'^ reporred receivingcomparable treatment effects on pain with 75 to225 mg of trazodone and 25 to 75 mg ofamitriptylme. In the only trial on pain patients inwhich trazodone was titrated as high as side effecrsallowed, rhe mean dosage remained at 201 mg,-The rarher high occurrence of side effects reportedby our patients might have rendered the use of anyhigher dosages difficult. The side effect profile cor-responded to that reported on trazodone in gen-eral.^^ Some patients in our marerial, as well as inother studies,^ '̂̂ '* have also reported dry monrh.

In contrast to the positive findings presented byVentafridda et al , ' ' the present study suggestedthat trazodone lacks analgesic efficacy in chronicpain. This is in line with some previous placebo-controlied studies on trazodone,-^'-'' The results onBMS patients were nor encouraging: rhe efficacywas comparable ro that of the placebo, and sideeffects were common.

In general, there is growing evidence to supportthe effectiveness of antidepressants in the treat-ment of chronic pain,"-''' but their mode of actionis poorly understood.^''''^'" Tricyclic antidepres-sants with mixed serotonergic and noradrenergicbiochemical profiles have rhe strongest evidence tosupport their efficacy in the treatment of differentchronic pain conditions, but there are no con-trolled trials on their use in BMS pain.

The only earlier controlled trial of pain manage-ment in BMS has been the work hy Loldrup etal.'*' In their placebo-controlled srudy, whichincluded patienrs with different idiopathic pains,they treated a group of 77 burning mouth painpatients with clomipramine, an antidepressantwith serotonergic biochemical profile; mianserin, anoradrenergic antidepressanr; or a placebo. Nostatistically significant difference between the 3treatmenrs was found in patients with BMS,

Considering the fact that BMS pain can continuefor years and can, at worst, he experienced assevere, daily, continuous pain, ir is understandablethat there is a real need ro try to find effectivetreatment for the symptoms. In the presentplacebo-controlled trial, trazodone failed to relieveburning mouth pain. Obviously, the search for aneffective treatment for burning mouth pain mustcontinue.

Journal of Orofacial Pain 87

Tammiala Salonen/Forssell

Acknowledgments

The authors are grateful to Katja Salokangas, MSc, for statisri-cal analysis of the data. The study was supported by the FinnisliDental Society.

References

1. Grushka M. Clinical features of burning mouth syndrome.Oral Surg Oral Med Oral Pachol 1987;63i3O-36,

2. Lamey P-J, Lamb A. Prospective study of aeriological fac-tors in burning mouth syndrome. Br Med J 1988:296:1,243-1,246.

3. Sbip J, Grubhka M, Lipton A, Mott A, Sessle BJ, DionneR. Burning muuth syndrome: An update. J Am Dent Assoc1995;126;íí42-853.

4. Tammiala-Salonen T, Hiidenkari T, Parvinen T. BurningiTiuuth in a Finnish adult population. Community DentOral Epidemiol 1993;2I:67-7I.

5. Gorsky M, Silverman S, Chinn H. Chnical characteristicsand management outcome in rhe burning moutb syn-drome. An open study of 130 patients. Oral Surg OralMed Oral Pathol I991;72:192-195.

6. Tourne LP, Fricton JR. Burning mourh syndrome. Criticalreview and proposed chnical management. Oral Surg OralMed Oral Pathol 1992;74:1JS-167.

7. Van der Waal I. The Burning Mouth Syndrome.Copenhagen: Munksgaard, 1990.

R. Grinspan D, Blanco GF, Allevato MA, Stengel FM.Burning moutli syndrome. Int J Dermatol 1995^34:483-487.

9. Grushka M, Sessle BJ, Miller R. Pain and personality pro-fileb in burning mouth syndrome. Pain 1987;28:155-167.

10. Grubbka M, Sessle BJ, Howley TP. Psych o physical assess-ment of tactile, pain and thermal sensory functions inburning mouth syndrome. Pain 1987;28:1Ö9-184.

11. Svensson P, Bjerring P, Arendt-Nielsen L, Kaaber S.Sensory and pain thresholds Io orofacial argon laser stim-ulation in patients wirh chronic burning moutb syndrome.Glin J Pain 1993;9:207-21S.

12. Jääskeläinen S, Forsseil H, Tenovuo O. Abnormalities ofthe blink reflex in burning moutb syndrome. Pain1997;73:455^60.

13. Magni G. The use of antidepressants in rhe trearment ofchronic pain. Drugs 199l;42i730-748.

14. Onghena P. van Houdenbove B. Antidepressant-inducedanalgesia in chronic non-malignant pain: A meta-analysisof 39 placebo-controlled studies. Pain 1992;49:20S-219.

15. McQuay HJ, Tramer M, Nye BA, Carroll D, Witfen PJ,Moore RA. A systematic review of antidepressants in neu-ropatbic pain. Pain 1996;68^217-227,

16. Kreisberg M. Tricyclic antidepressants: Analgesic effectand indications in orofacial pain. J Craniomandib DisordFacial Oral Pain 1988;2:171-177.

17. McQuay HJ, Moore RA. Antidepressants and chronicpain. BMJ 1997;314:763-764.

18. Haria M, Fitton A, McTavish D, Traîodone: A review ofIts pbarmacology, therapeutic use in depression and thera-peuric potential in other disorders. Drugs Aging 1994;4:331-355.

19. Ventafridda V, Caraceni A, Saita L, ßonezzi C, De ConnoF, Cuarise G, er al. Trazodone for deafferentarion pain.Comparison wirh amitriptyline, Psychopharmacology1988;95:44-49,

20. Ketovuori FI, Pöntinen PJ. A pain vocabulary in Finnish:The Finnish pain questionnaire. Pain l981;ll:247-253.

21. Melzack R. The McGill Pain Questionnaire: Major prop-cities and scoring methods. Pain 1975;l:277-299.

22. Zakrzewska J. The burning mouch syndrome remains anenigma |editorial|. Pain l995;62:253-257.

li. Goodkin K, Gullion CM, Agras WS. A randomized, dou-ble-blind, placebo-controlled trial of trazodone hydrochlo-ride in chronic low back pain syndrome. J ClinPsychopbarmacol 1990;l0:269-278.

24. Davidoff G, Guarracini M, Roth E, Sliwa J, Yarkony G.Traiodone hydrocbloride in tbe treatnient of dysestheticpain in traumatii: myelopathy: A randomized, double-blind, placebo-controlled study. Pain Í987;29:151-161.

25.Vrerhem M, Boivie J, Arnqvist H, Holmgren H,Lindström T, Thoreil L-H. A comparison of amitriptylineand maprotiline in che treatment of painful polyneuropa-thy in diabetes and nondiabetes. Clin J Pain 1997;13:313-323.

26. Loldrup D, Langemark M, Hansen HJ, Olesen J, Bech P.Clomipramine and mianserin in cbronic idiopatbic painsyndrome. Psychopharmacology 1989;99:l-7.

88 Volume 13,Number2, 1999