Introduction

Preterm birth (delivery occurring before 37 weeks of gestation) is the principal cause of neonatal mortality, morbidity, and major socioeconomic implications associated with hospital stays in surviving infants [1]. Tocolytic drugs with various mechanisms including β-mimetic agents are being used to manage the threat of preterm labor and prolong gestation.

Ritodrine is a tocolytic agent that binds to the β2-adrenergic receptor (ADRB2) on the outer membrane of myometrial cells. Binding increases the level of cyclic AMP (cAMP), which decreases intracellular calcium levels and leads to uterine contraction reduction [2]. Despite its wide usage in several European and Asian countries, the efficacy of ritodrine is not clinically consistent.

In our previous study, it was found that ADRB2 polymorphisms affected time to delivery in preterm labor women [3]. It was assumed that gene polymorphisms regulating the level of cAMP, the key factor in uterine smooth muscle relaxation mechanism of ritodrine, can also affect the tocolytic efficacy of ritodrine therapy on pregnant women.

Phosphodiesterase (PDE) terminates the signaling pathway of myometrial relaxation by degrading cAMP to the inactive 5′-AMP (Supplementary Fig. 1) [4]. The PDE4 family is one of the most predominant PDE families that display high affinity to cAMP, of which the PDE4A, 4B, 4C, and 4D mRNA subtypes are present in human myometrial smooth muscle cells [5, 6]. Among them, PDE4D and PDE4B2 have been shown to decrease β2-agonist-induced cAMP levels [7,8,9]. While PDE4D was found to be prominently expressed in myometria of women regardless of pregnancy, PDE4B2 was known to be more abundant in the myometrium of pregnant women than that of non-pregnant woman, suggesting an increase of PDE4B2 expression during pregnancy [6].

PDE4 selective inhibitors are clinically used for treatment of chronic obstructive pulmonary disease and are currently being evaluated in clinical trials for the treatment of inflammatory disorders. The potential of PDE4 selective inhibitors as a novel therapeutic strategy for preterm labor has been explored in previous literature; inhibition of spontaneous contractions in near-term myometria [10] and prevention of inflammation-driven preterm delivery have been documented [11].

Although the PDE4 family could affect preterm labor, no study has assessed the effects of PDE4 gene polymorphisms on ritodrine therapy for preterm labor women. Therefore, we aimed to investigate the possible effects of PDE4D and PDE4B2 gene polymorphisms on tocolytic effects and adverse drug events (ADEs) of ritodrine therapy in patients with preterm labor. The current study was based on the hypothesis that genetic variation in the predominant type of PDE4, which enhances the degradation of cAMP [12], alters ritodrine efficacy due to its association with physiological processes of the β2 agonist-induced cAMP pathway and its role as a signaling component governing the desensitization of β2-adrenergic receptor mediated signals [13].

Methods

Study patients

This prospective study was carried out from January 2010 to December 2014 at Ewha Womans University Mokdong Hospital. This study was approved by Ethics Committee of the Ewha Womans University Mokdong Hospital Institutional Review Board (IRB No.: 217-1-26). Eligible patients were enrolled if they met the following criteria: gestational age of 20 to 36 weeks, preterm labor with intact membrane, ≥ 18 years of age, uterine contractions with a frequency of 3 per 10 min with cervical change, and provision of written informed consent. Severe placenta abruptio, pre-eclampsia, fetal distress, severe oligohydroamnios, placenta previa, severe spontaneous premature rupture of membrane, and women whose continuation of pregnancy would be dangerous were excluded. Patients treated with ritodrine to prevent uterine contraction during the McDonald operation or selective termination were excluded at the end of study. This study was conducted according to International Ethical Guidelines for Biomedical Research Involving Human Subjects by the Council for International Organizations of Medical Sciences and approved by the Ewha Womans University Mokdong Hospital Ethical Review Committee (IRB number: 217-1-26). Informed consent was obtained from all patients before participating in the study.

Drug administration

Ritodrine (Lavopa®; JW Pharmaceutical, Seoul, Korea) was given by intravenous infusion at an initial rate of 0.05 mg/min with an increase of 0.05 mg/min every 10 min, until the desirable uterine response was obtained. Intravenous treatment was discontinued during uterine quiescence. Patients who achieved uterine quiescence received maintenance therapy via an infusion of 0.05 mg/min for 12 to 48 h.

Outcomes and data collection

The primary endpoint was time to delivery. The secondary endpoint was the proportion of patients who suffered from adverse events by ritodrine. Adverse events were defined as cases of drug cessation or dose reduction due to heart rate ≥ 100 bpm and symptoms of palpitation, dyspnea, and shortness of breath.

Selection of single nucleotide polymorphism (SNP)

SNPs of PDE4D and PDE4B2 were selected based on other studies [14,15,16] and genetic information from the UCSC Genome Browser. Linkage disequilibrium (LD) data and minor allele frequency (MAF) data in Japanese and Han Chinese populations from Haploreg ver. 2 [17] and tagger function within the Haploview v4.2 [18] program were incorporated to assort PDE4D and PDE4B2 gene SNPs. Additionally, rs1042719 of the ADRB2 gene was included for multivariate analysis based on our previous study [3].

Genotyping methods

Genomic DNAs were extracted from ethylenediaminetetraacetic acid–blood samples using QIAamp DNA Blood Mini Kits (QIAGEN GmbH, Hilden, Germany). The following PDE4D and PDE4B2 SNPs were analyzed by a SNaPshot assay using single base primer extension assay with ABI PRISM SNaPShot Multiplex kits (Applied Biosystems): rs1544791 (c.42+42302 A>G), rs983280 (c.42+36247 G>A), rs1504982 (c.272+218525 A>G), rs10940648 (c.272+218885 A>G), rs829259 (c.*2515 A>T), rs598961 (c.232-1587 G>A), rs2180335 (c.590-12261 A>G), and rs17128809 (c.231+2702 A>T).

Statistical analyses

The interval from start of treatment to delivery was analyzed with Kaplan–Meier survival data analysis method (log-rank test). Cox’s proportional-hazards model was used for exploratory multivariate analysis for the interval from start of treatment to delivery. Categorical variables were analyzed using chi-squared test or Fisher’s exact test. The tests for polymorphisms correspond to the best-fitting models of inheritance, according to descriptive statistics. A multivariable logistic regression model was used to identify independent predictors after adjusting other variables.

The model fit of the prediction model was assessed by computing the Hosmer-Lemeshow goodness-of-fit test. Discrimination of the model was further assessed by an analysis of the area under receiver operating curve (AUROC), which assesses the ability of the risk factor to predict ritodrine-induced ADEs. The data were analyzed using Statistical Package for Social Sciences Version 20.0 for Windows (SPSS, Chicago, IL, USA). A P value < 0.05 was considered statistically significant. Post hoc power analysis was performed using PROC POWER of the SAS version 9.4 (SAS Institute, Cary, NC, USA).

Results

Two hundred and sixteen patients were enrolled. Ten patients who used ritodrine as a premedication for McDonald operation, 14 patients who already had severe symptoms that led to preterm labor before hospital admission, and 22 patients without blood samples were excluded. Forty-six patients were ultimately excluded and 170 patients were included. Seven patients who had underlying cardiovascular diseases were additionally excluded for analysis of ritodrine-induced ADEs.

A post hoc power analysis was performed on the basis of Rozenberg et al.’s study [19], provided that overall observed MAFs of given SNPs were 30%. A sample size of 170 patients achieved 79% power to detect a clinically meaningful decrease of 35% in time to delivery at the two-sided α = 0.05 level in a dominant model.

In baseline characteristics, the mean maternal and gestational age at admission was 31.0 ± 3.7 years and 29.5 ± 3.9 weeks, respectively. The mean weight and BMI was 62.4 ± 8.2 kg and 24.1 ± 3.0 kg/m2, respectively. Gestational age at start of drug therapy (P < 0.001) and modified Bishop score (P = 0.002) were significant factors for time to delivery (Table 1).

Table 1 Effects of demographic characteristics on time to delivery
Full size table

All PDE4D and PDE4B2 SNPs were in Hardy–Weinberg equilibrium (Table 2). As shown in Supplement Fig. 2, rs1544791 and rs983280 of PDE4D gene showed linkage disequilibrium with D’ value of 0.963 and r2 value of 0.699. Among the PDE4 gene SNPs, only rs598961 showed significant association with time to delivery. Patients with variant homozygotes had significantly shorter median time to delivery than those with other genotypes (P = 0.031, Fig. 1).

Table 2 Effects of grouped genotypes on phosphodiesterase 4D and 4B on time to delivery
Full size table
Fig. 1
figure 1

Survival curve of PDE4B2 rs598961 by Kaplan–Meier survival analysis. a Survival curve comparing AG,GG group with AA group (P = 0.031). b Survival curve comparing each genotype group (P = 0.059)

Full size image

After adjusting rs1042719 of the ADRB2 gene and demographic variables with P < 0.05 in addition to age, the PDE4B2 rs598961 polymorphism showed that mutant homozygote carriers had shorter median time to delivery compared to patients with other genotypes (adjusted hazard ratio [AHR] 1.57, 95% confidence interval 1.02 to 2.40, P = 0.035). Patients with gestational age ≥ 32 weeks and modified Bishop score ≥ 2 had shorter time to delivery (AHR 4.40 and 1.97, respectively), after adjusting other factors. Since drug-related factors such as infusion rate, maximum infusion rate, total dose, and duration of infusion also could be strong confounders, Model II was constructed with the drug-related factors in addition to variable used in Model I. PDE4B2 rs598961 polymorphism remained a significant factor for time to delivery even after adjusting variables including drug-related factors. Among drug-related factors, average infusion rate was a significant factor; the hazard of delivery increased 1.04 times by every 1 cc/h increase (Table 3).

Table 3 Multivariate analysis (Cox’s proportional-hazards model) of median time to delivery
Full size table

In univariate analysis for ritodrine-induced ADEs, patients with height less than 160 cm (P < 0.001), weight less than 60 kg (P < 0.05), and BSA < 1.65 m2 (P < 0.01) showed significantly higher ADEs. Among SNPs studied, rs17128809 showed significant association with ADEs (Table 4).

Table 4 Effects of baseline characteristics on adverse drug reaction of ritodrine
Full size table

As presented in Table 4, height, weight, BSA, and rs17128809 were significant factors for ritodrine-induced ADEs in the univariate analysis. Since there was multicollinearity between height or weight and BSA, two models were constructed for the multivariate analysis Table 5. Model I included height and weight in addition to rs17128809, and Model II included BSA instead of height and weight of Model I. Height (AOR 3.43, 95% CI 1.60–7.35), weight (AOR 2.17, 95% CI 1.12–4.62), and BSA (AOR 2.64, 95% CI 1.25–5.76) were significant factors for ritodrine-induced ADEs after adjusting for confounders. Among SNPs studied, PDE4B2 rs17128809 was a significant factor; patients with AA genotype had 2.86 (Model I) and 2.59 (Model II) times higher ADEs compared to those with other genotypes.

Table 5 Multivariate analysis of factors on adverse drug reaction of ritodrine
Full size table

The Hosmer–Lemeshow test for Model I, which included height, weight and rs17128809, revealed a good fit (χ2 = 1.488, P = 0.960). The AUROC was 0.726 (Supplementary Fig. 3 (A)). In Model II, which included BSA and rs17128809, the Hosmer–Lemeshow test also revealed a good fit (χ2 = 1.037, P = 0.597), and the AUROC was 0.668 (Supplementary Fig. 3 (B)).

Discussion

cAMP hydrolysis by PDE4 is a key signaling component in the G protein–coupled receptors signaling pathway that consequently results in myometrial relaxation [12, 13, 20]. In fact, the influence of PDE enzyme on β agonist response was indicated by several studies [8, 21]. PDE selective inhibitors showed their influence upon β agonists by potentiating the utero-relaxing effects of terbutaline when two agents were used in human myometrium [21]. In a study comparing the human airway smooth muscles of normal and asthmatic subjects, the contribution of PDE activity in modulating β2 agonist-induced cAMP production was assessed. This study revealed an overexpression of PDE4D in asthmatic airway smooth muscle cells compared to control cells [8]. Another study examining PDE expression in human myometrium showed that PDE4B2 expression increased in near-term myometrium compared to myometrium in the non-pregnant state [22].

PDE4D gene polymorphisms influence various diseases such as asthma and chronic obstructive pulmonary disease (COPD) [8, 23]. A pharmacogenetic study elucidated the correlation between PDE4D genetic variations and β agonists among asthma patients by analyzing four SNPs that were also included in our study [14]. Two SNPs, rs1544791 and rs1504982, showed significant associations with responsiveness of β2 agonist administration. However, both SNPs failed to achieve statistical significance in our study.

In PDE4B2 variants, PDE4B2 rs598961 showed a significant influence on altering the ritodrine response in both univariate and multivariate analyses. PDE4B2 rs598961 is located in the intron region of the PDE4B2 gene, and this genetic information is known to be uninvolved in protein production. However, intron regions have the potential to affect mRNA splicing and alter protein expression or activity [24, 25].

Among the SNPs studied, PDE4B2 rs17128809 was a significant factor for ritodrine-induced ADEs from multivariate analysis. The results suggested that patients with the rs17128809 AA genotype have a 2.6~2.9 times higher risk for ritodrine-induced ADEs. Therefore, it would be advisable to detect pregnant patients with the AA genotype of rs17128809 on an early stage, which would then allow methods to prevent ritodrine-induced ADEs, including more cautious monitoring during ritodrine administration.

Expectedly, older gestational age at start of drug therapy and higher modified Bishop score were associated with higher hazard ratio of delivery. On the contrary, lower height and weight or lower BSA was significantly related to ritodrine-induced ADEs. This pattern might be explained by the conjuncture that shorter or lighter women must exhibit higher drug concentrations, leading to more ADEs. However, a more scientific explanation in this matter would require further research.

This study is limited by its small sample size and the fact that it is a single-center study. With respect to rs1544791, the median time to delivery for wild-homozygotes carriers, heterozygote carriers, and mutant homozygote carriers were 1073, 980, and 1944 h, respectively, showing that the difference in time to delivery between wild-type allele carriers and mutant homozygote carriers was almost two-fold (1040 vs 1944); however, statistical significance was not found. This was possibly due to the small sample size; the post hoc power analysis showed that power to detect significance of rs1544791 was less than 50% (47.9%). Another limitation is a lack of a detailed mechanism for each SNP associated with ritodrine response. Therefore, our hypothesis requires further independent validation using a multi-centered research with a large sample size, along with verification to clarify how these genetic polymorphisms affect the efficacy and safety of ritodrine.

To our knowledge, this is the first report evaluating potential effects of PDE4 genes on efficacy and adverse drug events of ritodrine in preterm labor patients. Multiple test correction was not performed in this study to avoid the possible loss of the true positives. Especially, Bonferroni correction was not applied as it is considered excessively conservative for a hypothesis-generating study, while this study is intended to find candidate genes affecting efficacy and adverse drug events of ritodrine. We have found possible genetic effects of PDE4; however, it should be implemented with caution considering the risk of false-positive results and must be verified by further replication studies.

Conclusions

This pharmacogenomic study suggested that PDE4 genetic polymorphisms impact individual susceptibility to β2-adrenergic receptor targeted therapy in patients with preterm labor. Since the study population was not ethnically diverse, further studies are required on other populations.