Preterm birth (before 37 completed weeks' gestation) accounts for 5–10% of births and is the major cause of perinatal mortality in North America and Europe.1 Systemic and local infections, antepartum hemorrhage, and fetal abnormalities predispose to preterm uterine contractions, but in at least half of cases, no obvious cause is found for the onset of labor.
Tocolytics like the β-adrenergic agonist ritodrine have not significantly reduced the incidence of preterm deliveries, and their main effect might be to reduce the number of deliveries within 48 hours of the onset of preterm labor.2 That allows time for administration of steroids to induce fetal lung maturation, reducing the risk of neonatal respiratory distress syndrome (RDS). Maternal and fetal side effects of tocolytics generally limit their clinical usefulness, and the consensus is that they have no effect on perinatal mortality or morbidity.2,3
Nitric oxide is an endogenously occurring biatomic molecule and can be given in the form of an NO donor. The first recorded use of an NO donor in pregnancy was reported in the British Medical Journal in 1882, when amyl nitrite was used to deliver a morbidly adherent placenta.4 Glyceryl trinitrate was used intravenously (IV) as a uterine relaxant to aid breech extraction (Greenspoon JS, Kovacic A. Breech extraction facilitated by glyceryl trinitrate spray [letter]. Lancet 1991;338: 124–5), to assist in replacing an inverted uterus,5 and in cases of retained placenta (Desimone CA, Norris MC, Leighton BL. Intravenous nitroglycerin aids manual removal of retained placenta [letter]. Anesthesia 1990; 73:787). Intravenous glyceryl trinitrate has been used as a tocolytic in term laboring monkeys,6 and in doses greater than those required for tocolysis, it did not alter hemodynamics in fetal lambs (Bootstaylor B, Roman C, Heymann MA, Parer JT. Fetal cardiorespiratory effects of nitroglycerin in the near term pregnant sheep. Am J Obstet Gynecol 1994;170[suppl]:281.
In vitro studies of uterine smooth muscle are technically difficult and the results inconsistent,7–9 but there is a consensus that nitric oxide donors cause smooth-muscle relaxation and that antagonism of nitric oxide production reverses relaxation. It was suggested that nitric oxide's relaxant effect on uterine muscle is more pronounced during pregnancy than delivery.10
In 1994, we reported the initial results of our experience with glyceryl trinitrate patches in preterm labor.11 After that study, we conducted a multicenter, randomized study comparing glyceryl trinitrate to conventional therapy (β-2 agonist ritodrine). The major outcome measure was prolongation of gestation with glyceryl trinitrate compared with ritodrine, and secondary outcomes were delivery within prespecified intervals. The study was also designed to give detailed qualitative and quantitative breakdowns of the side effect profile of glyceryl trinitrate compared with ritodrine.
Women were recruited from 14 centers in the United Kingdom and 6 other international centers from December 1, 1994, to August 31, 1996. Recruitment gestation followed national standards of practice. In the United Kingdom, it is rare to use tocolytics after 32 weeks' gestation, so women were recruited between 24 weeks and 31 weeks and 6 days. Centers outside the United Kingdom use tocolytics up to 36 weeks' gestation; in those centers, women could be entered from 24 weeks to 35 weeks and 6 days' gestation. Randomized treatments were transdermal glyceryl trinitrate or IV ritodrine; the study was not masked. The trial profile is shown in Figure 1.
Inclusion criteria were painful, regular uterine contractions palpated by the attending midwives or clinicians at more than two every 10 minutes for more than 1 hour, with or without cervical change. Exclusion criteria were hypotension (less than 80 mmHg systolic or 50 mmHg diastolic), major fetal congenital abnormalities in which tocolytics would not normally be used, unreassuring traces on fetal cardiotocography, antepartum hemorrhage or history of recurrent vaginal bleeding, placenta previa, rupture of membranes evidenced by liquor in the vagina on speculum examination, cervical suture in situ, chorioamnionitis (pyrexia greater than 37.5C with maternal tachycardia and uterine tenderness with or without offensive vaginal discharge), unexplained pyrexia, urinary tract infection, sensitivity or contraindication to nitrates or β-agonists, nonsteroidal anti-inflammatory drug therapy, and other tocolytic therapy during this pregnancy.
Randomization lists were prepared centrally at the London School of Hygiene and Tropical Medicine for United Kingdom and international centers, and at Brescia for the hospitals at Brescia and Cremona. Randomization was by random permuted blocks, stratified by center. United Kingdom and international centers were instructed to fax subject details to King's College Hospital, London before opening a sealed randomization envelope.
At the outset of treatment, maternal and fetal measurements were made every 15 minutes. When the subjects on therapy were stable, abdominal examination to palpate contraction frequencies and strength was performed hourly together with blood pressure (BP), pulse, and temperature monitoring. A 1-hour cardiotocograph trace was done before and after glyceryl trinitrate patch or ritodrine administration, followed by intermittent monitoring.
For women randomized to glyceryl trinitrate, a Deponit 10-mg glyceryl trinitrate (Schwarz Pharma, Monheim, Germany) patch was applied directly to the skin of the abdomen. If, after 1 hour, there was no reduction in contraction frequency or strength, an additional patch was applied. No more than two patches (20 mg) could be worn simultaneously. Mild headache was treated with paracetamol (acetaminophen). Patches remained in place for full 24 hours, at the end of which they were removed and the women reassessed.
Women randomized to ritodrine were given ritodrine by IV infusions through an IVAC volumetric pump (IVAC Corp., San Diego, CA) or syringe pumps. The rate of administration was titrated to the woman's contractions and increased per local labor ward policy or infusion chart (according to Royal College of Obstetricians and Gynaecologists guidelines, normally commencing at a rate of 50 μg/minute). Maternal tachycardia was not to exceed 120 beats per minute, and if hypotension (BP below 80/50 mmHg) or unacceptable side effects (such as severe tremor or palpitations) occurred, the infusion rate was reduced or stopped. Chest pain or breathing difficulties indicated immediate cessation of ritodrine, and cardiovascular assessment.
Treatment was terminated after cessation of contractions for 24 hours, or progress of labor to delivery. If the episode of preterm labor settled, subjects were discharged with follow-up according to the clinicians' normal practice. Repeat episodes of preterm labor were treated as randomized, if possible. We recorded gestation at delivery, type of delivery, birth weight, Apgar scores at 1 and 5 minutes, and stillbirth or neonatal death.
The study was designed with a total sample of 200 women, adequate to detect a 14-day increase in mean gestation at the 5% significance level with 90% power if the standard deviation was 30 days. After inspection of the pooled outcome data for the two arms (ie, without using any data on randomized treatment), the primary outcome was defined as time from randomization to delivery, censored at 37 weeks' gestation and expressed as a percentage of time from randomization to 37 weeks. That focused on the effects of glyceryl trinitrate on preterm delivery and was approximately independent of gestational age at randomization. Predefined secondary outcomes were the proportion of women who delivered on the same day as randomization, the next day, by 7 and 14 days after randomization, and by 32, 34, and 37 weeks' gestation, and the number of weeks from randomization to delivery or 37 weeks' gestation.
Two United Kingdom hospitals (Reading and Burnley) did not use any tocolytics routinely, and in those centers, double-masked randomization to glyceryl trinitrate or placebo was done. Two patches could be applied per 24 hours, and the maximum treatment duration was 48 hours. The 12 women recruited in that way (7 glyceryl trinitrate, 5 placebo) were not included in the main analyses of glyceryl trinitrate versus ritodrine, but because ritodrine is only believed to be efficacious in prolonging gestation for 48 hours, those women were included in an alternative analysis of the proportion who delivered by 32, 34, and 37 weeks' gestation.
Continuous outcomes were compared using the t test. When distributions were non-normal, the confidence intervals (CIs) estimated by t test were checked using bootstrap methods12 and were valid in all cases. Treatment effects for different gestational ages at entry were compared by entering appropriate dummy variables and treatment group into a regression model and testing their interaction. Absolute prolongation of gestation was adjusted for entry gestation by analysis of covariance.
Outcome data were missing for 12 (4.8%) of 245 randomized subjects. Baseline characteristics of the two groups were comparable (Table 1). Most women were treated as randomized, 73% with ritodrine, 78% with glyceryl trinitrate. The number of treatment crossovers and those not receiving treatment was similar in both groups (Table 2). Twenty-six women discontinued ritodrine (seven to start glyceryl trinitrate), and 17 women discontinued glyceryl trinitrate (seven to start ritodrine). The number of emergency cesareans was similar in each group, 25 (21%) for ritodrine and 20 (18%) for glyceryl trinitrate.
There was no significant difference in the primary outcome measure, percentage prolongation of gestation to 37 weeks (Table 3). Glyceryl trinitrate and ritodrine prolonged gestation by 74%. The actual, uncensored prolongation of ritodrine and glyceryl trinitrate was also similar (Table 4). There was no evidence that treatment effect differed with gestation at entry (Table 3; P = .16).
Glyceryl trinitrate and ritodrine were not significantly different in reducing delivery within any of the time limits in Table 4; however, the lower number of deliveries before 37 weeks in the glyceryl trinitrate group (37% compared with 48%) was near statistical significance (P = .08). The number of deliveries on the day of randomization in the glyceryl trinitrate and ritodrine groups was not significantly different (2% compared with 5%, respectively, P = .13).
Among 12 women randomized to glyceryl trinitrate versus placebo, three of five delivered before 37 weeks in the placebo group, compared with one of seven in the glyceryl trinitrate group. When those data were combined with the main data, the proportions delivering within 32, 34, and 37 weeks were reduced in the glyceryl trinitrate group by 1%, 2%, and 13% (36% versus 49%), respectively, compared with ritodrine or placebo. Those results are only slightly more favorable to glyceryl trinitrate than those for glyceryl trinitrate compared with ritodrine, but the last difference just reaches statistical significance (P = .04; 95% CI 1%, 25%).
Table 5 shows the adverse events associated with glyceryl trinitrate and ritodrine. Complete data were collected for 97 women (78%) who received ritodrine and 94 (78%) who received glyceryl trinitrate. The major side effect from glyceryl trinitrate was headache, which occurred in 30% of women compared with 1% in the ritodrine group. All other maternal adverse events were more frequent for ritodrine, in particular tachycardia (58%), palpitations (13%), and nausea (10%), all of which were infrequent or did not occur at all in the glyceryl trinitrate group. The number of combined neonatal deaths and stillbirths in both groups was the same (2%). The most serious maternal complication of tocolytic usage, pulmonary edema, occurred in one woman in the ritodrine group.
The results of this study were consistent with similar tocolytic effects for ritodrine and glyceryl trinitrate. There were no significant differences in the primary outcome; in both groups mean prolongation of gestation was 74% of time from randomization to 37 weeks, but the CI suggested that the true difference could have been as large as 10% in favor of glyceryl trinitrate or 10% in favor of ritodrine. These translate to an average difference of 9 days in women randomized at 24 weeks and 3.5 days in women randomized at 32 weeks. Figure 2 shows a near-significant reduction in preterm delivery rate for glyceryl trinitrate (37%) compared with ritodrine (48%). The small group of women randomized to glyceryl trinitrate compared with placebo also showed a lower preterm delivery rate for glyceryl trinitrate, and when all the data were analyzed together, the risk of preterm delivery was reduced from 49% with ritodrine or placebo to 36% with glyceryl trinitrate (P = .04).
These results should be interpreted with caution because they are secondary outcomes, but if they are true, it would be important to distinguish whether glyceryl trinitrate was more efficacious than ritodrine at times removed from the first treatment (ie, acute treatment having a delayed effect later in gestation). Ritodrine is effective in reducing the number of deliveries within 48 hours of commencing treatment,2 which is the main rationale for its use. The results of our study suggest a trend toward fewer deliveries in the ritodrine group compared with glyceryl trinitrate, on the first day of randomization, but the differences are small.
Ritodrine is in widespread use as oral maintenance following episodes of acute preterm labor and has reduced the likelihood of further acute tocolytic administration, or preterm delivery during treatment.13 The narrow dose window between therapeutic efficacy and unacceptable or dangerous side effects (tremor, tachycardia, palpitations, hypotension, or pulmonary edema) limits the dose and duration of acute and maintenance ritodrine. That might be the advantage of other tocolytics such as glyceryl trinitrate: its main side effect was a headache (in 30% of women), which can usually be dealt with using simple analgesics. The effect of glyceryl trinitrate on maternal and fetal cardiovascular characteristics in this study was minimal compared with ritodrine and is described in detail separately.14
β agonists are susceptible to tachyphylaxis and reduction in tocolytic efficacy with time. After the first few days' treatment, there is demonstrable down-regulation of myometrial β receptors.15 Glyceryl trinitrate's possible effect in reducing preterm delivery might hold the prospect of using it for maintenance or prophylaxis of repeat preterm labors. The advantage of glyceryl trinitrate would be once-daily, transdermal application at a dose unlikely to cause either headaches, cardiovascular disturbance, or other side effects particularly related to β agonists.
1. Berkowitz GS, Papiernik E. Epidemiology of preterm birth. Epidemiol Rev 1993;15:414–43.
2. Canadian Preterm Labour Investigation Group. Treatment of preterm labor with the beta adrenergic agonist ritodrine. N Engl J Med 1992;327:308–12.
3. King JF, Grant A, Keirse MJ, Chalmers I. Beta mimetics in preterm labour: An overview of the randomized controlled trials. Br J Obstet Gynaecol 1988;95:211–22.
4. Barnes F. Hour-glass contraction of the uterus treated with nitrite of amyl. BMJ 18 March 1882:377.
5. Altabef K, Spencer JT, Zinberg S. Intravenous nitroglycerin for uterine relaxation of an inverted uterus. Am J Obstet Gynecol 1992;166:1237–8.
6. Jennings RW, MacGillivray, Harrison MR. Nitric oxide inhibits preterm labour in the rhesus monkey. J Mat Fet Med 1993;2:170–5.
7. Lees JH, Change KC. Different sensitivity to nitric oxide of human pregnant and nonpregnant myometrial contractility. Pharmacol Comm 1995;5:147–54.
8. Buhimschi I, Yallampalli C, Dong YL, Garfield RE. Involvement of a nitric oxide-cyclic guanosine monophosphate pathway in control of human uterine contractility during pregnancy. Am J Obstet Gynecol 1995;172:1577–84.
9. Jones GD, Poston L. The role of endogenous nitric oxide synthesis in contractility of term or preterm human mymetrium. Br J Obstet Gynaecol 1997;104:241–5.
10. Bansal RK, Goldsmith PC, He Y, Zaloudek CJ, Ecker JL, Riemer, RK. A decline in myometrial nitric oxide synthase expression is associated with labor and delivery. J Clin Invest 1997;99:2502–8.
11. Lees C, Campbell S, Jauniaux E, Brown R, Ramsey B, Moncada S, Martin JF. Arrest of preterm labour and prolongation of gestation with glyceryl trinitrate, a nitric oxide donor. Lancet 1994;343:1325–6.
12. Efron B, Tibshirani RJ. An introduction to the bootstrap. New York: Chapman and Hall, 1993.
13. Holleboom CA, Merkus JM, van Elferen LW, Keirse MJ. Double-blind evaluation of ritodrine sustained release for oral maintenance of tocolysis after active preterm labour. Br J Obstet Gynaecol 1996;103:702–5.
14. Black RS, Lees C, Thompson C, Pickles A, Campbell S. The maternal and fetal cardiovascular effect of transdermal glyceryl trinitrate and intravenous ritodrine. Obstet Gynecol 1999;94:(in press).
15. Englehardt S, Zieger W, Kassubek J, Michael MC, Lohse MJ, Brodde OE. Tocolytic therapy with fenoterol induces selective down-regulation of beta-adrenergic receptors in the human myometrium. J Clin Endocrinol Metab 1997;82:1235–42.