Antenatal exposure to nonsteroidal anti‐inflammatory drugs and risk of neonatal hypertension

Abstract

Nonsteroidal anti‐inflammatory drugs (NSAIDs) are used as tocolytics, which are medications that suppress uterine contractions for preterm birth prevention. Their effect on cerebral/systemic vascular beds poses the question of whether antenatal NSAID exposure is associated with neonatal hypertension. We performed a retrospective case‐control study in a tertiary neonatal intensive care unit, including 40 hypertension cases (hospitalized neonates ≥ 35 weeks with systolic BP > 100 mm Hg on three consecutive days) compared to 134 controls matched by gestational age at delivery, plurality, and delivery date. Cases and controls were compared by antenatal NSAID exposure, other common tocolytics, and maternal/neonatal characteristics and complications. Multivariable logistic regression was used to estimate the odds of hypertension among those with prenatal exposure to NSAIDs versus those without exposure. Newborns with hypertension had a lower gestational age at delivery and increased incidence of neonatal complications, including respiratory distress syndrome, bronchopulmonary dysplasia, surfactant administration, longer duration of ventilation, and history of umbilical artery catheterization. Days of indomethacin exposure were positively associated with greater odds of neonatal hypertension (OR 1.17 [1.00 to 1.38], P = 0.055), even after adjustment for other factors associated with neonatal hypertension. Newborns with hypertension were less likely to have been exposed to calcium channel blockers as a tocolytic. The results of our study suggest an association between prenatal exposure to nonsteroidal anti‐inflammatory drugs and neonatal hypertension. Furthermore, our data suggest that prenatal calcium channel blocker exposure may protect against the development of neonatal hypertension. Future multicenter studies are needed to understand the risks of tocolytics and subsequent consequences in preterm infants.

1. INTRODUCTION

Prompt diagnosis of preterm labor is important for decision‐making regarding both non‐medical and medical interventions to reduce the morbidity and mortality associated with birth prior to 37 weeks gestation. Tocolytic agents are medications used to suppress premature labor by reducing uterine contractions and lengthening the time to delivery to allow for administration of antenatal corticosteroids to promote fetal maturity. Tocolytic agents commonly used include magnesium sulfate, nonsteroidal anti‐inflammatory drugs (NSAIDs; eg, indomethacin and sulindac), calcium channel blockers (CCB), and terbutaline. Despite their frequent use, there is no uniform recommendation regarding the preferred tocolytic agent for prevention of preterm birth.1

NSAIDs are potent vasoactive drugs that inhibit the secretion of inflammatory prostaglandins and cytokines.2 These anti‐inflammatory properties are thought to be the mechanism by which they diminish uterine contractions. For this purpose, indomethacin has been used since the 1970s and in a recent study it was found to be the tocolytic with the highest probability of delaying delivery by 48 hours.3 NSAIDs also cause direct and indirect vasoconstriction on vascular smooth muscle, with potential adverse neonatal outcomes of early closure of the ductus arteriosus, intracranial hemorrhage, renal dysfunction, periventricular leukomalacia, and necrotizing enterocolitis (NEC).4, 5 Sulindac, another NSAID used for tocolysis, has been associated with adverse neonatal outcomes of NEC and early constriction of the ductus arteriosus, but to a less severe degree and in a more transient manner.6, 7

Given the effect of NSAIDs on cerebral and systemic vascular beds, we hypothesized that vasoconstriction of smooth muscle in systemic vascular beds might predispose infants exposed to antenatal NSAIDs to neonatal hypertension. Therefore, we sought to determine if there is an association between antenatal NSAID exposure and the subsequent development of hypertension (HTN) in the neonate, a phenomenon that has not been extensively studied. The other objective of this study was to determine if any other pre‐ or postnatal factors might be associated with neonatal HTN, including exposure to other tocolytic medications.

2. MATERIALS AND METHODS

We conducted an IRB‐approved retrospective case/control study to evaluate the association between antenatal exposure to NSAIDs and neonatal HTN in premature infants. This was a secondary analysis of an original study to evaluate the association between antenatal NSAID exposure and neonatal necrotizing enterocolitis.6 We compared neonatal HTN cases to normotensive controls born in a single Level III NICU from January 1, 2007 to February 20, 2012 at Good Samaritan Hospital in Cincinnati, Ohio. A waiver of informed consent was obtained due to the retrospective nature of the study. Cases were defined as a hospitalized neonate ≥ 35 weeks corrected gestational age (CGA) with hypertension listed in the problem list and systolic BP greater than 100 mm Hg on three consecutive days. This definition was chosen to be most conservative in our definition of hypertension and avoid misclassifying infants with normotension as a case; for reference, the 99th percentile systolic BP for neonates 34 weeks CGA is 90 mm Hg. Furthermore, guidelines recommend consideration of antihypertensive therapy in neonates when the blood pressure is consistently greater than the 99th percentile.8 Exclusion criteria included major congenital anomalies, monochorionic twins, and triplet and higher order gestations. With dichorionic twin gestations, one twin was randomly selected for study inclusion.

Neonatal intensive care unit nurses measure blood pressure in the neonatal patients at least twice per day; this may increase to every four hours if the baby is critically ill (eg, on a ventilator). The blood pressure is taken when the infant is quiet and with no other clothing in place. The infants in this study qualified as cases through blood pressure measured on the arm or leg using the Critikon Neonatal Blood Pressure Cuff. After measuring the circumference of the limb and then following the manufacturer's recommendation for cuff size, the appropriately sized cuff was used.

Controls were neonates ≥ 35 weeks CGA randomly chosen from a list of births with similar gestational age (GA), plurality, and delivery dates. They were initially matched 1:1 by gestational age, indication for delivery by three broad categories (preterm labor, premature rupture of membranes, and indicated delivery to include intrauterine growth restriction, hypertensive disorders, and non‐reassuring fetal status), and who first delivered after the case‐patient within the same years as the study. However, due to the low prevalence of neonatal hypertension, all defined cases and matched controls during the study period were included with a ratio of at least 1:3.

All charts for women and their infants who met inclusion criteria were reviewed to abstract data, including information regarding antenatal NSAID exposure. Data accuracy was confirmed by a second investigator review of 10% of the charts, revealing discrepancies in less than 5% of the collected data variables. Maternal demographic variables included maternal age, race, ethnicity, history of prior preterm birth, and history of cesarean delivery. Pregnancy complication variables included maternal diabetes (pre‐existing and gestational), hypertensive disease (chronic, gestational, pre‐eclampsia), tobacco use during pregnancy, twin gestation, oligohydramnios, premature rupture of membranes (defined as less than 34 weeks), preterm labor (defined as labor before 37 weeks), and exposure to antenatal corticosteroids (betamethasone). Details of NSAID use for tocolysis were collected, including gestational age at initial exposure, duration of treatment, total milligrams administered, and latency period between treatment and delivery.

Neonatal characteristic data variables were gestational age at delivery, weight at birth, birth weight percentile, and sex. Data were collected on neonatal outcome variables, including surfactant administration, intraventricular hemorrhage, need for inotropes, bronchopulmonary dysplasia, respiratory distress syndrome, hours of respiratory support, prophylactic postnatal indomethacin use, patent ductus arteriosus, umbilical artery catheterization, pulmonary hypertension, and neonatal death.

With our current sample size and case/control ratio (40 cases and 134 controls), we had 80% power with an alpha error of 5% to detect 2.8 fold increase odds of NSAID exposure among HTN cases. Student's t‐test and Fisher's exact tests were used to compare categorical and continuous variables between cases and controls. Multivariable logistic regression was performed to estimate the odds of HTN among those with prenatal exposure to NSAIDs for tocolysis versus those without NSAID exposure, adjusting for GA at birth, infant sex, infant race, sulindac duration, indomethacin duration, doses of steroids, and calcium channel blocker exposure (yes/no). P values of less than 0.05 and 95% confidence intervals, not including 1, were considered statistically significant. A test of normality was performed for all data variables, and the appropriate statistical test was used for analysis accordingly.

3. RESULTS

During the study period of 2007 through early 2012, there were 40 infants with neonatal HTN who met inclusion/exclusion criteria (cases) and 134 infants without hypertension (controls). There was no significant difference in baseline maternal demographics and clinical characteristics between neonatal HTN cases and controls (Table 1).

Table 1.

Comparison of maternal demographic and clinical characteristics between cases of neonatal hypertension and controls

Characteristic	Infants with HTN (cases), N = 40	Infants without HTN (controls), N = 134	P value
Maternal age, years	27.35 ± 5.26	27.19 ± 5.86	0.874
Maternal race (Caucasian)	26 (65.0%)	90 (67.2%)	0.799
Maternal ethnicity (non‐Hispanic)	39 (97.5%)	132 (99.2%)	0.410
Maternal diabetes	2 (5.0%)	19 (14.2%)	0.167
Maternal hypertensive disease	10 (25.0%)	32 (23.9%)	0.885
Maternal smoking	7 (17.5%)	24 (18.0%)	0.937
Prior preterm birth	10 (25.0%)	32 (23.9%)	0.885
Twin gestation	17 (42.5%)	44 (32.8%)	0.261
Fetal growth restriction	3 (7.7%)	17 (12.7%)	0.571
Oligohydramnios	5 (12.5%)	19 (14.3%)	0.775
Premature rupture of membranes	12 (30.0%)	40 (29.9%)	0.986
Antenatal steroids	33 (82.5%)	108 (81.8%)	0.922
Doses of antenatal steroids	1.78 ± 1.27	1.48 ± 0.89	0.097
Cesarean delivery	28 (70.0%)	82 (61.2%)	0.311

Overall, 77.5% of case infants and 75.4% of control infants had a history of antenatal tocolytic exposure. The most common tocolytic class used for both cases and controls was NSAIDs: 49 mothers (28%) received indomethacin, 33 (19%) received sulindac, and 32 (18%) received both indomethacin and sulindac for antenatal tocolysis (Table 2). The most commonly used indomethacin dosing regimen was an initial 100 mg loading dose followed by 50 mg every six hours for 48 to 72 hours. The most commonly used sulindac‐dosing regimen was 200 mg twice daily.

Table 2.

Comparison of tocolytic agent use between cases of neonatal hypertension and controls

Characteristic	Infants with HTN (cases), N = 40	Infants without HTN (controls), N = 134	P value
Exposure to indomethacin	16 (40.0%)	33 (24.5%)	0.058
Gestational age started, weeks	23.87 ± 1.81	24.24 ± 2.31	0.583
Gestational age stopped, weeks	25.10 ± 1.88	25.81 ± 2.07	0.251
Duration, days	7.06 ± 10.69	3.75 ± 2.96	0.241
Total dose, mg	637.50 ± 620.35	695.31 ± 662.97	0.773
Latency to delivery, days	7.63 ± 13.35	10.69 ± 16.58	0.525
Exposure to sulindac	10 (25.0%)	23 (17.2%)	0.267
Gestational age started, weeks	24.30 ± 1.51	24.30 ± 2.15	0.998
Gestational age stopped, weeks	25.70 ± 1.69	26.80 ± 2.25	0.177
Duration, days	9.20 ± 12.46	15.87 ± 13.56	0.194
Total dose, mg	3370.00 ± 4905.34	5658.18 ± 5635.45	0.278
Latency to delivery, days	6.00 ± 11.05	6.61 ± 10.35	0.880
Exposure to both indomethacin and sulindac	10 (25.0%)	22 (16.4%)	0.219
Duration, days	19.40 ± 21.95	21.09 ± 13.85	0.792
Total dose, mg	4230.00 ± 5571.56	6805.00 ± 5694.86	0.246
Exposure to calcium channel blockers (Nifedipine)	7 (17.5%)	32 (23.9%)	0.396
Duration, days	1.14 ± 1.21	7.78 ± 10.97	0.002
Number of tocolytic agents
0	9 (22.5%)	33 (24.6%)	0.346
1	9 (22.5%)	38 (28.4%)
2	9 (22.5%)	31 (23.1%)
3	4 (10.0%)	17 (12.7%)
4	6 (15.0%)	6 (4.5%)
5	3 (7.5%)	9 (6.7%)

Thirty‐nine patients (22%) received CCBs as tocolytic agents during this study period. Among those, 16 patients (9%) received both NSAIDs and CCBs (three in cases vs 13 in controls, P = 0.672), and 23 patients (13%) received only CCBs (four in cases vs 19 in controls, P = 0.493).

As compared to controls, the total duration of antenatal CCB exposure was significantly shorter in HTN cases (1.4 ± 1.2 vs 7.8 ± 11.0 days, P < 0.001; Table 2). There was no statistically significant difference between cases and controls in regards to indomethacin and sulindac exposure, GA at the start or stop of drugs, total doses, duration of therapy, or latency period (Table 2).

HTN cases had earlier GA at delivery, lower birth weight, and higher CGA at discharge (Table 3). HTN cases also had an increased incidence of bronchopulmonary dysplasia, surfactant administration, and respiratory distress syndrome. Furthermore, HTN cases required more hours of respiratory support (1737 vs 1144 hours) and had increased incidence of umbilical artery catheterization (42.5% vs 25%; Table 3). There was no statistical difference between groups regarding sex, neonatal death (less than 28 days of life), need for inotropes or indomethacin in the neonatal period, the presence of patent ductus arteriosus, intraventricular hemorrhage, or pulmonary hypertension (Table 3). Additionally, of the 40 neonatal HTN cases identified, 32 (80%) required treatment.

Table 3.

Comparison of postnatal factors between cases of neonatal hypertension and controls

Characteristic	Infants with HTN (cases), N = 40	Infants without HTN (controls), N = 134	P value
Gestational age at delivery, weeks	27.70 ± 2.73	28.91 ± 2.99	0.023
Birth weight, g	1065.0 ± 467.0	1222.1 ± 453.3	0.058
Birth weight percentile, %	42.28 ± 24.36	39.00 ± 25.51	0.473
Sex (male)	23 (57.5%)	72 (53.7%)	0.674
Surfactant	30 (75.0%)	59 (44.0%)	0.001
Intraventricular hemorrhage	7 (17.5%)	25 (18.8%)	0.853
Severe IVH	2 (5.0%)	12 (9.0%)	0.526
Need for inotropes	2 (5.1%)	14 (10.5%)	0.530
Bronchopulmonary dysplasia	18 (45.0%)	36 (27.1%)	0.032
Neonatal indicated indomethacin	3 (7.5%)	8 (6.0%)	0.717
Respiratory distress syndrome	38 (95.0%)	105 (79.5%)	0.022
Hours respiratory support	1737.13 ± 1094.90	1144.07 ± 1313.62	0.005
Patent ductus arteriosus	23 (57.5%)	57 (42.9%)	0.103
Umbilical artery catheterization	17 (42.5%)	33 (25.0%)	0.033
Pulmonary hypertension	4 (10.0%)	17 (12.8%)	0.637
CGA at discharge	41.86 ± 5.54	39.30 ± 5.44	0.010
Neonatal death	0	3 (2.2%)	0.340

Multivariable logistic regression revealed that indomethacin exposure was independently associated with a trend toward increased odds of neonatal HTN (OR 1.17 [1.00 to 1.38], P = 0.055; Table 4).

Table 4.

Factors potentially associated with neonatal hypertension

Covariate	Unadjusted OR	Adjusted OR	P valuea
Sulindac duration, days	0.99 (0.95‐1.04)	0.94 (0.86‐1.03)	0.164
Indomethacin duration, days	1.11 (1.00‐1.22)	1.17 (1.00‐1.38)	0.055
Doses of antenatal steroids	1.35 (0.94‐1.94)	1.34 (0.90‐1.98)	0.147
Calcium channel blockers	0.68 (0.27‐1.68)	0.50 (0.18‐1.41)	0.191
Maternal race (Caucasian)	0.91 (0.43‐1.91)	0.87 (0.39‐1.94)	0.739
Infant sex (male)	1.17 (0.57‐2.38)	1.15 (0.54‐2.44)	0.725
GA at birth	0.86 (0.75‐0.98)	0.89 (0.77‐1.02)	0.092

Adjusted for all covariates listed in table.

^a P value for adjusted OR.

4. DISCUSSION

With 15 million preterm births occurring per year in the US, choice of tocolytic agents can have a widespread impact on short‐ and long‐term neonatal outcomes.9 In this case‐control study of infants born prematurely at a single Level III NICU from January 1, 2007, to February 20, 2012, we found that antenatal NSAID tocolytic use associated with increased odds of HTN in the neonate. We also found that longer duration of CCB exposure was protective against the development of neonatal HTN. These findings have significant therapeutic implications, as there are currently no established consensus or guidelines regarding the preferred tocolytic therapy in the setting of preterm labor.

Tocolytic therapies are commonly used to provide an opportunity for antenatal corticosteroid administration, with the goal of decreasing neonatal morbidity and mortality. Of the tocolytic agents available, NSAIDs are one of the more commonly used agents. While NSAID medications are prescribed worldwide for other indications and are considered relatively safe in non‐pregnant patients, elevated blood pressure is a known side effect.2 While only occurring in up to 1% of adults, hypertension secondary to NSAID use occurs secondary to direct vasoconstriction from diminished prostaglandin release and indirectly due to increased intravascular volume from diminished natriuresis.10 The effect of NSAIDs on blood pressure in adults has a dose‐dependent effect, in that it is more likely to occur with longer duration of use and with higher dosages administered.2 NSAID use is also associated with resistant hypertension, and hypertensive patients with chronic conditions requiring anti‐inflammatory use are counseled to either avoid NSAIDs due to their known effect on blood pressure or hypertension management is changed to include calcium channel blockers or alpha‐agonists, as NSAIDs do not impact the efficacy of these agents' use.11, 12

Despite these known adverse effects in adults, NSAIDs continue to serve as common tocolytic agents for preterm labor. Of concern, our data suggest that antenatal indomethacin exposure is associated with elevated blood pressure in the preterm neonate. Systemic hypertension in infants is uncommon, and limited literature suggests a prevalence of 0.7% to 3% in the neonatal intensive care setting.13, 14 Its onset can be insidious, as often infants are asymptomatic, and hypertension is only picked up on routine vital sign monitoring.15 Exposure to potential agents that increase risk the development of hypertension is a concern, as premature infants are already at higher risk for hypertension as adults due to their decreased nephron endowment.16, 17 Nephrogenesis is complete at 34 to 36 weeks gestation; when born before this time, nephrogenesis can continue; however, it is often suboptimal and is thought not to progress beyond 40 days of life. Further, premature infants are often subjected to multiple renal insults in early life, related to hypoperfusion/sepsis, need for umbilical artery catheterization, and required use of nephrotoxic agents, each of which puts them at higher risk for cardiovascular disease later in life. Several studies have eloquently demonstrated that decreased birth weight, a proxy for gestational age, is significantly associated with decreased number of nephrons, increased single nephron size (ie, compensatory hypertrophy, which is known to be associated with progressive kidney disease), and hypertension in adulthood.18, 19

It is becoming increasingly established that the antecedents to adult disease begin in childhood (likely prenatally), bringing greater urgency to efforts that aim to determine risk factors for the development of adult disease such as hypertension. In fact, one of the priorities of the American Heart Association is to increase efforts toward primordial prevention of cardiovascular disease (CVD).20, 21 Primordial prevention aims to prevent CVD risk factors from ever developing, which is different from primary prevention that aims to prevent diseases from occurring via risk‐factor reduction. Animal and human studies have consistently described the importance of the intrauterine environment, and the potential impact of premature exposure to the extrauterine environment on later outcomes and risk factors.22 Also, seminal longitudinal cohort studies have shown that blood pressure in childhood tracks into adulthood, and this tracking is more likely to occur with increasing numbers of comorbid CVD risk factors such as an elevated body mass index, a family history of hypertension and the presence of left ventricular hypertrophy.23, 24, 25 Determining the impact of prenatal medications on the development of CVD risk factors is therefore essential to decreasing the burden of CVD and the related morbidity and mortality.

Several systematic reviews and meta‐analyses have shown that preterm and very low birth weight infants (even after weight adjustment) have higher resting blood pressure than infants born full term.26, 27, 28 Furthermore, Pejovic et al29 showed that blood pressures rise after the first week of life in both preterm and term infants, although the rate‐of‐rising is faster in preterm infants. It is not clear if intrauterine exposures such as tocolytic agents contribute to either association. Our data also suggest that longer antenatal CCB exposure may be protective against neonatal HTN. This observation is interesting, as CCB are often among the first line treatments for neonatal hypertension.8, 15 Calcium channel blockers exert their antihypertensive effects by causing relaxation of the cardiac and vascular smooth muscle. The resulting vasodilatation and decrease in total peripheral resistance occur via blockage of extracellular calcium influx into smooth muscle cells. It is possible, that when used for tocolysis, the prenatal kidney benefits from increased renal blood flow that could manifest as reduced hypertension risk in infancy. This is in contrast to the other tocolytic agents commonly utilized, such as NSAIDs that inhibit prostaglandins. Prostaglandins are important for renal development, with evidence suggesting adverse renal function with both maternal and neonatal use of NSAIDs. These findings need to be investigated further, as they suggest a differential effect of tocolytic agents on the risk of hypertension in neonates.

There were fewer cases of neonatal HTN identified than anticipated, but this may be related to our conservative definition of hypertension. First, there is the consideration of how blood pressure was measured. Given that the infants diagnosed with hypertension were beyond 35 weeks corrected age, they did not have invasive blood pressure monitoring, which is considered the gold standard and would have given the most accurate blood pressure measurements. We also are unable to present specific data on the location of where the blood pressure was obtained (eg, arms vs legs), due to limitations inherent to chart review. However, all of the neonates were from the same NICU, where the neonatal nurses undergo standardized training regarding accurate blood pressure measurement. Second, the diagnosis of hypertension in infancy and childhood is percentile based. Hypertension is diagnosed when blood pressures are measured at or above the 95th percentile on at least three separate occasions. In the neonatal period, antihypertensive therapy is not recommended routinely until the SBP is sustained above the 99th percentile (77 to 100 mm Hg for infants born at 26 EGA and 40 weeks EGA, respectively.8 It is for this reason that we limited our definition to neonates with SBP above 100 mm Hg because it would be unlikely to misclassify normotensive infants with neonatal hypertension. Further, to be included as a case, the neonatal problem list was required to include hypertension. As a result, neonates with SBP above 100 mm Hg may have been missed if their problem list did not include hypertension, leading to a diagnosis bias. However, SBP greater than 100 mm Hg was above the 99th percentile for those in the study; the problem list likely identified most neonates with a BP in this range.8

Another limitation of the study relates to the increased prevalence of neonatal HTN risk factors among the cases versus controls. For example, the HTN case group had a higher incidence of surfactant use, earlier GA at birth, bronchopulmonary dysplasia, respiratory distress syndrome, hours of respiratory support, and umbilical artery catheterization (Table 3). This may suggest that these neonates were sick at the time of delivery, which may have an impact on their development of neonatal hypertension. The limited literature that has studied neonatal hypertension confirms these characteristics as risk factors for the condition.13, 30, 31 Adjustment for these risk factors in multivariable regression analysis attempted to account for this potential confounding.

As noted earlier, prematurity could be another contributing factor to the development of neonatal hypertension.13, 17 As noted in Table 3, gestational age at birth was statistically different between the case and control groups, with the HTN cases having a lower gestational age at birth (P = 0.023). However, when controlled through logistic regression, there was no statistical difference in gestational age at birth between the two groups (P = 0.092; Table 4). It is plausible that gestational age at delivery could independently increase the risk of neonatal hypertension, but the sample size of this study limits our ability to explore this further. Future studies with a larger sample size would need to be conducted to answer this important clinical question.

While tocolytics are used to allow time for administration of antenatal corticosteroids, a better understanding of possible risks and adverse outcomes is urgently needed. Additionally, without evidence‐based recommendations regarding the preferred first‐line tocolytic agent for preterm labor, further elucidating the individual risks of various tocolytic medications, prior to exposing the pregnant woman and the fetus, is essential. Improved understanding of the risk profiles of available tocolytic medications and their short‐ and long‐term effects on the fetus and neonate will help to identify a preferred tocolytic agent with the safest profile for use in preterm labor. To that end, the correlation we describe between antenatal exposure to NSAIDs and neonatal hypertension warrants further investigation in larger studies to assess short‐ and long‐term neonatal effects after antenatal exposure to NSAIDs. Such data will aid in achieving primordial prevention of cardiovascular disease risk factors by decreasing the prevalence of neonatal hypertension.

CONFLICT OF INTEREST

The authors have no conflicts of interest to report.

Habli M, Clifford CC, Brady TM, et al. Antenatal exposure to nonsteroidal anti‐inflammatory drugs and risk of neonatal hypertension. J Clin Hypertens. 2018;20:1334‐1341. 10.1111/jch.13354