Home Articles & Issues Published Ahead-of-Print CME Collections ABOG MOC II Podcasts Videos Journal Info
Skip Navigation LinksHome > October 2001 - Volume 98 - Issue 4 > Uterine Relaxant Effects of Cyclooxygenase‐2 Inhibitors In V...
Obstetrics & Gynecology:
Original Research

Uterine Relaxant Effects of Cyclooxygenase‐2 Inhibitors In Vitro

Slattery, Michael M.; Friel, Anne M.; Healy, David G. MB; Morrison, John J. MD, MRCOG

Free Access
Article Outline
Collapse Box

Author Information

Department of Obstetrics and Gynaecology, National University of Ireland Galway, Clinical Science Institute, University College Hospital Galway, Galway, Ireland.

Address reprint requests to: John J. Morrison, MD, Department of Obstetrics and Gynaecology, National University of Ireland Galway, Clinical Science Institute, University College Hospital Galway, Newcastle Road, Galway, Ireland; E‐mail: john.morrison@nuigalway.ie.

We thank the Health Research Board (HRB) of Ireland and the Higher Education Authority (HEA) for funding this research.

Received February 6, 2001. Received in revised form May 23, 2001. Accepted June 7, 2001.

Collapse Box


OBJECTIVE: To compare the effects of three cyclooxygenase‐2 (COX‐2) inhibitors: nimesulide, meloxicam, and celecoxib, which exhibit varying COX‐2 selectivity, on contractile activity in pregnant (before and after labor) and nonpregnant human myometrial tissue in vitro.

METHODS: Isometric tension recording was performed under physiologic conditions in isolated myometrial strips obtained from 33 women undergoing hysterectomy or either elective or emergency cesarean section. The effects of cumulative additions of nimesulide, meloxicam, and celecoxib (between 1 nmol/L and 100 μmol/L) on myometrial contractility were measured, and values for −log10 EC50 and mean maximal inhibition were compared.

RESULTS: Nimesulide, meloxicam, and celecoxib exerted significant relaxant effects on contractility in nonpregnant, pregnant nonlabor, and pregnant labor myometrial strips. Values for −log10 EC50 values (± standard error of the mean) were as follows: nimesulide (nonpregnant) 5.14 ± 0.93 (n = 6), (pregnant nonlabor) 4.91 ± 0.75 (n = 6), and (pregnant labor) 5.84 ± 0.35 (n = 6); meloxicam (nonpregnant) 6.53 ± 0.57 (n = 6), (pregnant nonlabor) 4.80 ± 0.71 (n = 6), and (pregnant labor) 5.62 ± 0.21 (n = 6); celecoxib (nonpregnant) 6.15 ± 0.99 (n = 6), (pregnant nonlabor) 7.08 ± 0.98 (n = 6), and (pregnant labor) 7.25 ± 0.99 (n = 3). Celecoxib exhibited greater potency than nimesulide or meloxicam (P < .01). The range of maximal relaxation values achieved in the three tissue types were as follows: nimesulide 68–70% (n = 18; P < .01), meloxicam 69–84% (n = 18; P < .01), and celecoxib 69–77% (n = 15; P < .01).

CONCLUSION: COX‐2 inhibitors exert significant relaxation in human myometrium with a similar potency in nonpregnant and pregnant (before and after labor onset) tissues. Celecoxib, a COX‐2 specific inhibitor, was more potent than nimesulide or meloxicam, COX‐2 preferential inhibitors.

The onset of labor is associated with an increase in uterine prostaglandin synthesis.1 Prostaglandins are formed by way of the cyclooxygenase (COX)—also known as prostaglandin–endoperoxidase H synthase (PGHS)—pathway from arachidonic acid. There are two known isoforms of the cyclooxygenase enzyme, COX‐1 and COX‐2, which are encoded by different genes.2 It is believed that COX‐1 is constitutively expressed, whereas COX‐2 is inducible and upregulated after labor onset.3 Nonsteroidal anti‐inflammatory drugs (NSAIDs), such as indomethacin, inhibit both the constitutive and inducible forms of cyclooxygenase4 and inhibit myometrial contractility.5 Such inhibitors of cyclooxygenase, which are mainly COX‐1 inhibitors, have been used for treatment of preterm labor but are associated with significant adverse fetal effects.6 COX‐2 is regarded as the inducible isoform of the enzyme with upregulation at the time of human labor,7 and hence the suggestion that it may play a central role in parturition. In addition, because of the possibility that COX‐2 inhibition may have less adverse fetal effects than COX‐1 inhibition, recent interest has focused on potential use of COX‐2 inhibitors as tocolytic agents.

The family of COX‐2 selective inhibitors exhibit great variation in their degree of COX‐2 selectivity. Nimesulide and meloxicam are COX‐2 preferential inhibitors4 in that they also inhibit COX‐1 activity. Nimesulide is estimated as being between five‐ and 16‐fold more selective for COX‐2.4 It decreases myometrial contractility in the nonpregnant sheep,8 delays preterm delivery in the sheep,9 and a case report has outlined that it inhibits human preterm labor.10 Meloxicam is reported to be between three‐ and 77‐fold selective for COX‐24 and is shown to have a dose‐dependent inhibitory effect on uterine contractions in both pregnant and nonpregnant rats.11 The more novel compounds described as COX‐2 specific inhibitors,4 such as celecoxib, are compounds that inhibit COX‐2 with a minimal effect on COX‐1 over the whole range of doses used. Such compounds are currently used in the treatment of osteoarthritis and rheumatoid arthritis,12 and celecoxib, for example, is reported as being up to 375‐fold selective for COX‐2.4

Few data outline the in vitro effects of COX‐2 preferential inhibitors in human myometrial tissue. Although it is known that nimesulide exerts a potent relaxant effect in tissue obtained at elective cesarean,13 we are unaware of any data comparing various COX‐2 inhibitors in different human myometrial tissue types. Systematic MEDLINE (and PubMed) search of the literature from 1990 to 2001 using the terms “cyclooxygenase‐2 inhibitors” and “uterus” did not reveal any reports outlining the comparative effects of COX‐2 specific inhibitors on different myometrial tissue types. The purpose of our study was to investigate the effects of COX‐2 inhibitors of varying selectivity (ie, COX‐2 preferential and COX‐2 specific inhibitors) on human myometrial contractility. The secondary aim was to evaluate the effects in nonpregnant tissue and in tissue obtained before and after the onset of labor, to investigate the possibility of altered sensitivity in association with pregnancy or labor.

Back to Top | Article Outline


Women attending the Department of Obstetrics and Gynecology at University College Hospital Galway, Ireland, between August 1999 and July 2000 were randomly recruited for the study. Approval was obtained from the Research Ethics Committee at University College Hospital Galway. Samples of human myometrial tissue were obtained from the fundus of hysterectomy specimens (n = 12) from premenopausal women undergoing surgery for benign conditions. The reasons for hysterectomy included menorrhagia, fibroids, and endometriosis. All operations were carried out transabdominally. Pregnant human myometrial samples were obtained at elective (n = 12) and intrapartum (n = 9) cesarean delivery from the upper midline portion of the lower uterine segment. Women who underwent induction of labor were excluded from the study. The reasons for elective cesarean delivery included breech presentation, previous cesarean delivery, and fetal growth restriction. The reasons for emergency cesarean delivery included failure to progress in labor and cardiotocograph (CTG) trace abnormalities. The criteria for inclusion in the intrapartum group were regular spontaneous uterine contractions, effacement of the cervix, and cervical dilatation larger than 3 cm before cesarean. All women were delivered at term (37–42 weeks' gestation).

Upon collection, tissue was placed immediately in Krebs‐Henseleit physiologic salt solution, composed of 4.7 mmol/L potassium chloride, 118 mmol/L sodium chloride, 1.2 mmol/L magnesium sulfate, 1.2 mmol/L calcium chloride, 1.2 mmol/L potassium phosphate, 25 mmol/L sodium bicarbonate, and 11 mmol/L glucose (Sigma‐Aldrich, Dublin, Ireland). Tissue was stored at 4 C and used within 12 hours of collection.

Before experimentation, myometrial strips were dissected measuring approximately 2 × 2 × 10 mm and mounted under 2 g of tension in organ tissue baths for isometric recording as described previously.14,15 The number of strips dissected per sample varied, allowing for investigation of effects of more than one compound per patient, but each individual strip and control were used only once.14,15 The tissue baths contained 20 mL of Krebs‐Henseleit physiologic salt solution, which was maintained at 37 C, pH 7.4, and was gassed continuously with a mixture of 95% oxygen/5% carbon dioxide. Myometrial strips were allowed to equilibrate for at least 1 hour before the addition of a utero‐tonic agent. The Krebs‐Henseleit physiologic salt solution in the tissue baths was changed every 15 minutes during the equilibration period. The mechanical response of tissues was measured by calculation of the integral of selected areas for 20‐minute periods using the PowerLab hardware unit and Chart v. 3.6 software (AD Instruments Inc., Charlotte, NC). The integral of contractile activity measures all contractile activity for that time period, including frequency and amplitude.

After equilibration contractions were stimulated by bath exposure of the strips to either oxytocin (0.5 nmol/L) for myometrium obtained during pregnancy or to phenylephrine (10 μmol/L) for human myometrium obtained at hysterectomy. After this addition, the integrated tension for the next 20 minutes was calculated as a measure of contractile activity stimulated by either oxytocin or phenylephrine. Either nimesulide (Ravensberg GmBH, Konstanz, Germany), meloxicam (Inter‐Chemical Ltd., Shenzhen, China), or celecoxib (a gift from Searle, Northlake, IL) were then added to their respective tissue baths in a cumulative manner at concentrations of 1 nmol/L, 10 nmol/L, 100 nmol/L, 1 μmol/L, 10 μmol/L, and 100 μmol/L at 20‐minute intervals. Control strips were simultaneously run with bath addition of oxytocin/phenylephrine and vehicle, but without addition of a COX‐2 inhibitor. The effect of each COX‐2 inhibitor was assessed by expressing the integral calculated during the 20‐minute period after the addition of each drug concentration as a percentage of the integral obtained in the 20‐minute period before any drug addition.

A stock of nimesulide, concentration 0.1 mol/L, was made up in N,N‐dimethylformamide (Sigma‐Aldrich, Dublin, Ireland). Stock solutions of meloxicam and celecoxib in 0.1‐mol/L concentrations were prepared using dimethyl sulfoxide (Sigma‐Aldrich). A stock solution of oxytocin (1 mmol/L) (Sigma‐Aldrich) was made in ethanol. Phenylephrine (Alchem, Dublin, Ireland) was prepared as a stock solution of 1 mmol/L. Series of dilutions were made in deionized water on the day of experimentation and were maintained on ice for the duration of the experiment. Fresh Krebs‐Henseleit physiologic salt solution was made daily.

The integrals of contractile activity were measured for a 20‐minute period for each bath concentration of COX‐2 inhibitor. Hence, the effects of nimesulide, meloxicam, and celecoxib on myometrial contractility were calculated for each 20‐minute period of exposure to 1 nmol/L, 10 nmol/L, 100 nmol/L, 1 μmol/L, 10 μmol/L, and 100 μmol/L, respectively, and these values were expressed as a percentage of the integral measured for the 20‐minute period before addition of any drug. The integral of contractile activity calculated after bath addition of the final dose of COX‐2 inhibitor, subtracted from 100% (ie, before any drug addition), represented the maximal relaxant effect for the compound. The EC50 value is the concentration of drug that results in 50% of the maximal inhibitory effect and was used as a means of comparing potency. Because the EC50 values were in the micromolar range (ie, 10−6 molar), values for the −log10 molar concentration of relaxant that produces 50% inhibition (ie, −log10 EC50 values) were used for the analysis (eg, 1 μmol/L = 10−6 mol/L = −log10 EC50 of 6.0). Values for the −log10 molar concentration of relaxant that produces 50% inhibition (log10 EC50 values) were calculated by linear regression of the probit of response versus log10 molar concentration for each of the COX‐2 inhibitors studied in isolated strips of human myometrium. The −log10 EC50 values and the mean maximal inhibition values for the three compounds, in the three different tissue types, were compared using a 3 × 3 factorial analysis of variance (ANOVA) test. Post‐hoc testing, when indicated, was performed using Fishers Least Significant Difference (LSD) protected t test. A value of P < .05 was accepted as statistically significant.

Back to Top | Article Outline


The demographic features of women who underwent elective and emergency cesarean delivery and hysterectomy are shown in Table 1. Nimesulide, meloxicam, and celecoxib all demonstrated potent, concentration‐dependent, relaxant effects on contractility in nonpregnant (n = 12) and pregnant (nonlabor [n = 12] and labor [n = 9]) myometrial strips. Figure 1 (A–C) demonstrates representative recordings of the effects of cumulative additions of celecoxib, nimesulide, and meloxicam to precontracted nonpregnant, pregnant nonlaboring, and pregnant laboring myometrial tissues, respectively. This relaxant effect was due to inhibition of both frequency and amplitude of contractions. Figure 2 (A–C) demonstrates in a graphical fashion the effects of cumulative additions of nimesulide, celecoxib, and meloxicam to nonpregnant, pregnant nonlaboring, and pregnant laboring myometrial tissues, respectively. Figures 1 and 2 both demonstrate the potent and concentration‐dependant utero‐relaxant effects produced by the COX‐2 selective inhibitors. The maximum relaxant effects of nimesulide, meloxicam, and celecoxib on each tissue type studied (± standard error of the mean [SEM]) are outlined in Table 2, along with their respective EC50 values. No significant alteration in contractility was observed in timed control strips with overall relaxation values of 6.34% ± 0.81% (P > .05) in strips run with oxytocin and vehicle only, and 5.22% ± 0.38% (P > .05) in strips run with phenylephrine and vehicle only.

Table 1
Table 1
Image Tools
Table 2
Table 2
Image Tools
Figure 1
Figure 1
Image Tools
Figure 2
Figure 2
Image Tools

The −log10 EC50 values (±SEM) for nimesulide were as follows: (nonpregnant) 5.14 ± 0.93 (n = 6); (pregnant nonlabor) 4.91 ± 0.75 (n = 6); and (pregnant labor) 5.84 ± 0.35 (n = 6). The −log10 EC50 values for meloxicam were: (nonpregnant) 6.53 ± 0.57 (n = 6); (pregnant nonlabor) 4.80 ± 0.71 (n = 6); and (pregnant labor) 5.62 ± 0.21 (n = 6). Corresponding values for celecoxib were: (nonpregnant) 6.15 ± 0.99 (n = 6); (pregnant nonlabor) 7.08 ± 0.98 (n = 6); and (pregnant labor) 7.25 ± 0.99 (n = 3). Multiple group (3 × 3 ANOVA) comparisons revealed significant differences between the −log10 EC50 values measured for the three different compounds in the three tissue types (P < .001). Post‐hoc analysis revealed that this difference was attributable to the fact that the −log10 EC50 values for celecoxib were significantly greater than those calculated for nimesulide (P < .01) and meloxicam (P < .01). This indicates that the actual EC50 value (ie, concentration resulting in 50% of maximal effect) for celecoxib is significantly less than that measured for either nimesulide or meloxicam, and hence, celecoxib is more potent.

The percent maximum reduction in contractility (ie, mean maximal inhibition values) produced by the respective compounds at bath cumulative concentration of 100 μmol/L (±SEM) for nimesulide were as follows: (nonpregnant) 69.0% ± 8.0%; (pregnant nonlabor) 68.0% ± 6.9%; and (pregnant labor) 70.6% ± 3.8% (P < .01). Mean maximal inhibition values for meloxicam were: (nonpregnant) 74.9% ± 5.8%; (pregnant nonlabor) 69.4% ± 6.5%; and (pregnant labor) 84.65% ± 3.3% (P < .01). Corresponding values for celecoxib were: (nonpregnant) 69.08% ± 7.39%; (pregnant nonlabor) 75.63% ± 6.19%; and (pregnant labor) 77.31% ± 7.25% (P < .01). There was no significant difference between the three compounds in the three different tissue types in terms of maximal inhibition achieved.

Back to Top | Article Outline


The results from this study clearly demonstrate that COX‐2 inhibitors (nimesulide, meloxicam, and celecoxib) are potent relaxant agents in human myometrial smooth muscle, including myometrium obtained during pregnancy (before and after labor onset) and tissue obtained from premenopausal hysterectomy specimens. Celecoxib, the most selective of all three COX‐2 inhibitors used, was significantly more potent than the two COX‐2 preferential compounds (nimesulide and meloxicam). We observed no alteration in sensitivity, in association with pregnancy or labor, when compared with nonpregnant premenopausal myometrial tissue. The measured potency of these COX‐2 inhibitors is comparable to that of other known utero‐relaxant compounds such as calcium channel blockers,16 nitric oxide donor compounds,17 and beta‐adrenergic agonists.16 The clinical relevance of our findings is that COX‐2 inhibitors may be of value therapeutically as tocolytic agents, and that compounds exhibiting greater COX‐2 selectivity may be more potent in inhibiting uterine contractions at lower drug concentrations.

Nearly all existing NSAIDs are predominantly COX‐1 selective.4 In recent years, it has become clear that COX‐2 preferential inhibitors (eg, nimesulide and meloxicam) exhibit significant, yet varying, COX‐1 inhibitory properties.4 Another group of compounds, the COX‐2 specific inhibitors, are agents that inhibit COX‐2 but have little or no effect on COX‐1 over the whole range of doses used (eg, celecoxib).4 The potential for adverse fetal effects from therapeutic use of COX‐1 inhibition for preterm labor are well recognized.6 The possibility that highly selective COX‐2 inhibition might have the efficacy of compounds such as indomethacin, but without the adverse side effects, has been raised as a novel option for preterm labor treatment.10 For these reasons, a further aim of our study was to investigate the effects of COX‐2 preferential inhibitors of varying selectivity (nimesulide and meloxicam), and of a COX‐2 specific inhibitor (celecoxib), on isolated human myometrial contractility. Our results show that a compound with high levels of COX‐2 selectivity (ie, celecoxib) demonstrates greater potency in human myometrial tissue than those compounds with lower levels of COX‐2 selectivity. Because of these findings and earlier reports of fetal adverse effects with nimesulide,18 it may be appropriate that further clinical studies investigating the potential therapeutic role of COX‐2 inhibitors for pre‐term labor should be restricted to COX‐2 specific inhibitors (ie, celecoxib‐type compounds).

Prostaglandins are potent utero‐tonic agents, but it is possible that COX‐2 inhibitors exert their utero‐relaxant effect by mechanisms other than inhibition of prostaglandin synthesis. This has not been addressed in our study, but previous reports suggest that nimesulide may have another mechanism of action.13 Sawdy et al13 have demonstrated that both nimesulide and indomethacin reduced the calcium channel current in human myometrial myocytes at concentrations that inhibited contractility. Because they were less effective in inhibiting the calcium channel current than in inhibiting contractility, it was suggested that calcium antagonism may be one of several mechanisms of action for COX‐1 and COX‐2 inhibitors, including the inhibition of prostaglandin synthesis. The exact mechanisms of relaxation of COX‐2 inhibitors will therefore be the subject of further research. There are further limitations to our study. Although our studies were all performed in human tissue, all tissue was obtained at term gestation. The reasons for this include the practical and ethical constraints that pertain to fresh collection of myometrial specimens from emergency cesareans performed before labor, or after the onset of labor, at preterm periods of gestation. Finally, differences in quoted selectivity for various COX‐2 inhibitors can vary more than tenfold, and no method of determining precise selectivity commands universal support.4

In conclusion, COX‐2 preferential inhibitors, and COX‐2 specific inhibitors, are potent utero‐relaxant agents in human myometrium. The COX‐2 specific inhibitor celecoxib exhibits greater potency for myome‐trial relaxation than COX‐2 preferential inhibitors, and therefore, COX‐2 specific inhibitors may represent the safest group of compounds for future clinical studies as therapeutic agents for preterm labor.

Back to Top | Article Outline


1. Challis JR, Lye SJ, Gibb W. Prostaglandins and parturition. Ann NY Acad Sci 1997;828:254–67.

2. Kosaka T, Miyata A, Ihara H, Hara S, Sugimoto T, Takeda O, et al. Characterisation of the human gene (PTGS2) encoding prostaglandin-endoperoxide synthase 2. Euro J Biochem 1994;221(3):889–97.

3. Slater D, Dennes W, Sawdy R, Allport V, Bennett P. Expression of cyclooxygenase types-1 and -2 in human fetal membranes throughout pregnancy. J Mol Endocrinol 1999;22(2):25–130.

4. Hawkey CJ. COX-2 inhibitors. Lancet 1999;353:307–14.

5. Quass L, Goppinger A, Zahradnik HP. The effect of acetylsalicylic acid and indomethacin on the catecholamine- and oxytocin-induced contractility and prostaglandin (6-keto-PGF1 alpha, PGF2 alpha)-production of human pregnant myometrial strips. Prostaglandins 1987; 34(2):257–69.

6. Norton ME, Merrill J, Cooper BA. Neonatal complications after administration of indomethacin for preterm labor. N Engl J Med 1993;329:1602–7.

7. Slater DM, Dennes WJB, Campa JS, Poston L, Bennett PR. Expression of cyclooxygenase types-1 and -2 in human myometrium throughout pregnancy. Mol Hum Reprod 1999;5:880–4.

8. Baguma-Nibasheka M, Nathanielsz PW. In vivo administration of nimesulide, a selective PGHS-2 inhibitor, increases in vitro myometrial sensitivity to prostaglandins while lowering sensitivity to oxytocin. J Soc Gynecol Invest 1998;5(6):296–9.

9. Poore KR, Young IR, Hirst JJ. Efficacy of the selective prostaglandin synthase type 2 inhibitor nimesulide in blocking basal prostaglandin production and delaying glucocorticoid-induced premature labor in sheep. Am J Obstet Gynecol 1999;180:1244–53.

10. Sawdy R, Slater D, Fisk N, Edmonds DK, Bennett P. Use of a cyclooxygenase type-2-selective non-steroidal anti-inflammatory agent to prevent preterm labour. Lancet 1997;350:265–7.

11. Yousif MH, Thulesius O. Tocolytic effect of the cyclooxygenase-2 inhibitor, meloxicam: Studies on uterine contractions in the rat. J Pharm Pharmacol 1998;50:681–5.

12. Goldenberg MM. Celecoxib, a selective cyclooxygenase-2 inhibitor for the treatment of rheumatoid arthritis and osteoarthritis. Clin Ther 1999;21(9):1497–513.

13. Sawdy R, Knock GA, Bennett PR, Poston L, Aaronson PI. Effect of nimesulide and indomethacin on contractility and the Ca2+ channel current in myometrial smooth muscle from pregnant women. Br J Pharmacol 1998;125(6):1212–7.

14. Morrison JJ, Ashford MLJ, Khan RN, Smith SK. The effects of potassium channel openers on isolated human myometrium before and after the onset of labor. Am J Obstet Gynecol 1993;169:1277–85.

15. Slattery MM, O'Leary MJ, Morrison JJ. Effect of parathyroid hormone-related peptide (PTHrP) on human and rat myometrial contractility in vitro. Am J Obstet Gynecol 2001;184:625–9.

16. Saade GR, Taskin O, Belfort MA, Erturan B, Moise KJ Jr. In vitro comparison of four tocolytic agents, alone and in combination. Obstet Gynecol 1994;84(3):374–8.

17. Morrison JJ, Perera D, O'Brien PJ, Marshall I, Rodeck CH. Effects of nitric oxide (NO) substrate, NO donor compounds and NO synthase inhibitors on contractions of isolated human pregnant myometrium. Br J Obstet Gynaecol 1996;103:483–4.

18. Peruzzi L, Gianoglio B, Porcellini MG, Coppo R. Neonatal end-stage renal failure associated with maternal ingestion of cyclooxygenase-type-1 selective inhibitor nimesulide as tocolytic. Lancet 1999;354(9190):1615.

© 2001 The American College of Obstetricians and Gynecologists


Article Tools