Predictors of unfavorable perinatal outcomes in premature rupture of membranes

Introduction / Введение

Preterm labor (PL) worldwide is one of the major risk factors for fetal and neonatal adverse outcomes resulting in neonatal death in 70 % of cases, infant mortality – in 36 % of cases, and late pediatric neurological sequelae in 25–50 % of cases [1][2]. In 2020, PL rate in the Russian Federation was 59.9 thousand (total number of births was 1,220.8 thousand) [3]. Currently, a close relationship between PL and lower genital tract infection has been unambiguously established [4]. Upon this, one in 10 PL patients shows signs of intra-amniotic inflammation, subclinical in most cases that leads to a high risk of developing premature rupture of membranes (PROM) [5–7].

The leading role in PROM pathogenesis is played by extracellular matrix degradation along with amniotic membrane remodeling due to the infectious and inflammatory process [8]. Long-term persistent infection leads to emerging overt chorioamnionitis and fetal systemic inflammatory response syndrome [9]. Therefore, antibacterial therapy is indicated for PROM, because systemic antibiotics use is associated with markedly lowered incidence of chorioamnionitis, neonatal infection and perinatal mortality [10].

In 2017, W.H. Sim et al. emphasized dominance of neonatal sepsis as well as intraventricular hemorrhage (IVH), in PROM-related neonatal morbidity and mortality pattern at gestational age of < 24 weeks. Particularly, W.H. Sim et al. found that mean time interval between PROM onset and delivery ranged from 20 to 43 days. In case PROM occurred before 24 weeks of gestation, live birth rate comprised 63.6 %, whereas discharge survival rate was 44.9 %. Upon PROM, neonatal survival was better at later stages of pregnancy in case of sufficient amniotic fluid volume, and with maternal C-reactive protein (CRP) levels < 10.0 mg/L during the first 24 hours after hospitalization and PROM [9].

Prognostic role for residual amniotic fluid volume in PROM – amniotic fluid index (AFI) and/or maximum vertical pocket (MVP) magnitude have been currently of special interest [11]. Ultrasound (U/S) is widely used to assess amount of amniotic fluid and verify PROM (U/S) [9][11–17]. Globally, two approaches have been applied to U/S assess volume of amniotic fluid: MVP measurement [18][19] and AFI [20][21], both providing advantages over subjectively assessed parameter including those regarding predicting unfavorable perinatal outcomes [11].

А. Weissmann-Brenner et al. (2009) examined 102 singleton pregnancies with PROM and mean gestational age at PROM comprising 29.0 ± 5.3 weeks (range: 14.0–36.6 weeks), who concluded that at AFI < 10 cm, PROM was verified with sensitivity, specificity, positive and negative predictive values reaching 89.2, 88.5, 72.2 and 96.0 %, respectively. Mean AFI magnitude in PROM group and control group was 5.8 ± 3.6 cm (0.0–18.5 cm) and 13.7 ± 3.2 cm (7.3–24.4 cm) (p < 0.001), respectively [12].

Currently, a search for relationships between maternal and fetal outcomes in PROM-complicated pregnancies and AFI and/or MVP magnitude has been continued [11]. For instance, in 2019 E. Weiner et al. suggested that at admission to a third-level hospital PROM patients had amount of amniotic fluid independently associated with development of severe neonatal respiratory failure [14], which agrees with the data reported by W.H. Sim et al. (2017), where most common diseases in neonates born to mothers with PROM before 24 weeks of gestation were respiratory distress syndrome and bronchopulmonary dysplasia (BPD), whereas pulmonary hypoplasia was one of the most common causes of neonatal death [9].

S. Storness-Bliss et al. (2012) sought out for a relationship between maternal and fetal outcomes in PROM-complicated pregnancies at gestational age of < 24 weeks by dividing patients into 2 groups: group A – AFI ≥ 1 cm and group B – AFI < 1 cm (severe oligohydramnios). It was concluded that a larger volume of amniotic fluid after PROM was related to a better prognosis for the fetus, whereas time interval from PROM to delivery extended without increasing maternal complications rate [15]. S.T. Vermillion et al. (2000) suggest that increased risk of developing perinatal infection and shortening of time interval from PROM to delivery at gestational age of 24–32 weeks may occur more frequently if the AFI level after PROM was < 5 cm [16]. M. Bhagat and I. Chawla (2014) conducted the study aimed at assessing a relation between perinatal outcomes and AFI by focusing primarily at its level of < 5 cm and > 5 cm [17].

Taking into account the lack of consensus regarding the most prognostically relevant AFI and/or MVP level in PL with PROM, we attempted to conduct current study to identify objective predictors of adverse fetal and neonatal outcome.

Aim: to identify predictors of unfavorable perinatal outcomes associated with PROM.

Materials and Methods / Материалы и методы

Study design / Дизайн исследования

A single-center retrospective cohort study enrolling women with PROM at 28 0/7 to 36 6/7 weeks of gestation was conducted at the Perinatal Center (PC), Vorokhobov City Clinical Hospital № 67, Moscow Healthcare Department. For this, pregnant women were hospitalized in the PC from 01.01.2023 to 01.11.2023 covering entire anhydrous interval from PROM onset. Paired children were treated in the PC from birth until discharge.

Inclusion and exclusion criteria / Критерии включения и исключения

Inclusion criteria: gestational age > 28 0/7 to <3 7 0/7 weeks; required neonatal intensive care unit (NICU) transfer immediately after birth; history of long-term premature rupture of membranes (PROM) > 18 hours.

Exclusion criteria: congenital malformations that affect the outcome; chromosomal and/or genetic syndromes; transfer of a neonate to another clinic.

Study methods / Методы исследования

Obstetric history / Акушерский анамнез

Patient obstetric history was assessed for the following clinical and laboratory parameters: maternal age; number of pregnancies and births; gestational age at delivery (in days); anhydrous period duration (in hours); hyperthermia (> 37.5 °C) and tachycardia (more than 100 beats/min) in pregnant women before delivery; anhydramnios, chorioamnionitis (assessed by clinical criteria); cesarean section (CS); AFI at PC admission and before delivery; MVP before delivery (by U/S, mm; blood CRP level before and after delivery; cervical bacterial growth before delivery.

Amniotic fluid index measurement / Определение индекса амниотической жидкости

AFI measurement was conducted by conditionally dividing pregnant woman's abdomen area into four quadrants: pigmentation along the abdomen white line or linea nigra guide for conditionally dividing the abdomen into right and left halves; the navel and a horizontal line conditionally drawn through it serve as a guide for conditionally dividing the abdomen into lower and upper halves. Ultrasound device sensor was positioned strictly perpendicular to the surface of bed on which patient was located during examination. In each quadrant, MVP was measured. Depth of the maximum free pocket depth of amniotic fluid in each quadrant was assessed. Amniotic fluid index was calculated as the sum of the vertical lengths of MVP. Oligohydramnios was defined as AFI magnitude ≤ 5th percentile calculated for gestational age, polyhydramnios – as an AFI magnitude ≥ 95th percentile for gestational age [22].

Maximum vertical length of free pocket of amniotic fluid / Величина максимального вертикального размера свободного кармана околоплодных вод

MVP per se was measured in the right corner of the uterine contour, to be further divided into 3 categories: (a) < 2 cm – oligohydramnios; (b) a pocket depth of 2 to 8 cm – normal amniotic fluid amount; (c) a pocket depth of > 8 cm – polyhydramnios [18][19].

Severity of respiratory failure and central hemodynamic disorders / Определение тяжести дыхательной недостаточности и нарушений центральной гемодинамики

Respiratory failure (RF) severity in preterm neonates was assessed to calculate surfactant (poractant alfa) administration in delivery room at a standard dose of 200 mg/kg, repeated at a dose of 100 mg/kg; a need for invasive mechanical ventilation (IMV) or transfer to high frequency oscillatory ventilation (HFOV) in delivery room was taken into account; neonates were also assessed within the first 3 days of birth for maximum mean air way pressure (МАР) and fraction of inspired oxygen (FiO₂). Neonatal hemodynamic impairment was assessed by using modified inotropic index (MII) [23]. Magnitude of neonatal multiple organ dysfunction syndrome (MODS) within the first 72 hours after birth was analyzed by using modified Neonatal Multiple Organ Dysfunction score (NEOMOD) [23][24].

Analysis of children morbidity / Анализ заболеваемости детей

Neonatal morbidity was assessed by the following nosology entities: early-onset neonatal sepsis [25], congenital pneumonia, pulmonary hemorrhage, pneumothorax, multiple skin ecchymoses and hemorrhages, surgical-stage retinopathy of prematurity, IVH grade 3, periventricular leukomalacia (PVL), severe BPD, surgical-stage necrotizing enterocolitis (NEC). The following neonatal laboratory parameters were selected for analysis [26][27]: blood leukocyte and normoblast counts, hemoglobin level and neutrophil index (NI) within the first 24 hours of life, procalcitonin (PCT) and CRP level, bacterial growth from sterile (blood) and non-sterile loci (nasopharynx and anus) within the first 3 days of life by using polymerase chain reaction (PCR).

Study groups / Группы обследованных

A total of 176 birth histories and 176 medical records of preterm neonates were examined by dividing a cohort of examined children into 2 groups based on neonatal outcomes: group with favorable (Group 1) outcome and adverse (Group 2) outcome. The following study indicators were classified as adverse outcomes: antenatal fetal death, neonatal death, IVH grade 3, PVL, severe BPD, surgical-stage NEC (Fig. 1).

Statistical analysis / Статистический анализ

The perinatal center doctors collected the data on study indicators and filled in the depersonalized database, which in turn included 180 parameters visualized using Microsoft Office Excel 2016 (Microsoft, USA). Data statistical processing was carried out using Python v 3.11 (USA). Normal distribution for study indicators was assessed using Shapiro–Wilk criterion. In case the data showed no normal distribution, nonparametric methods were used to obtain more accurate data. The following statistical indicators were determined:

• medians (Me) – to determine the central tendency for all observations accompanied with most common values;
• quartile range [ Q₁; Q₃] to determine magnitude of data spread from the median;
• Mann–Whitney U-test for analyzing a group with independent samples;
• Pearson χ² criterion was used for assessing categorical variables, where most significant inter-group differences were of interest. In case number of table values was less than 10, Fisher exact criterion was used. Spearman rank correlation coefficient (Rs) was used to test a relationship between quantitative variables, indicating a correlation and strength between them.

Relative indicators were as follows:

• odds ratio (OR) with 95 % confidence interval (CI) was used to estimate probability of any event in a certain group compared to another;
• relative risk (RR) with 95 % CI demonstrated higher or lower risk of an event in a certain group compared to the other. To determine how accurate (quantitatively) the data were for prediction, ROC analysis was used to assign a patient to a risk of an adverse outcome.

To assess diagnostic value of quantitative indicators in predicting a certain outcome, including the probability of outcome calculated by using a regression model, the ROC curve analysis was used to determine optimal division of quantitative indicator magnitude allowing for classifying patients by degree of outcome risk, combining best sensitivity and specificity. Quality of the predictive model obtained by this method was assessed based on the area under the ROC curve with standard error and 95 % CI as well as level of statistical significance. Differences were considered statistically significant at p ≤ 0.05, using a two-sided p-level of significance.

Results / Результаты

Major quantitative obstetrical clinical and laboratory parameters assessed in pre- and postnatal period are shown in Table 1.

Regarding age, parity, ordinal number of deliveries and duration of the anhydrous interval pregnant women did not significantly differ between study groups. However, gestational age at the time of birth (days) in group with unfavorable outcome (Group 2) was significantly lower (p < 0.001) comprising 193.0 [ 180.0; 198.0] vs. 238.0 [ 223.5; 247.0] (Group 1) days, respectively. Blood CRP level and leukocyte count before delivery in Group 2 was significantly higher than in Group 1. AFI magnitude at admission to PC and immediately before delivery as well as MVP level before delivery were found to be significantly lower in Group 2 compared with Group 1.

Nonparametric variables characterizing the course of pregnancy and childbirth in the examined women are presented in Table 2.

It turned out that surgical delivery via CS was required significantly more often in group with unfavorable neonatal outcome (Group 2). Antenatal fetal death was recorded in 7 of 176 (4.0 %) cases (Group 2). Fever (> 37.5 °C), tachycardia (> 100 beats/min), as well as bacterial growth from the cervical canal in pregnant women before delivery were recorded in both groups with near similar rate. However, it was noteworthy that pregnant women in Group 2 were significantly more often diagnosed with chorioamnionitis (p < 0.001) and anhydramnios (p = 0.003).

Quantitative clinical characteristics of premature neonates are presented in Table 3.

In Group 2 (unfavorable outcome), gestational age, birth weight, and Apgar score were significantly lower than in Group 1. No significant difference in sex bias between the children in both groups was found: 78 out of 135 (57.8 %) and 27 out of 41 (66.0 %) in Groups 1 and 2, respectively (OR = 1.41 [ 0.68; 2.93]; OR = 1.08 [ 0.92; 1.27]; p = 0.356). In total, 105 out of 176 (59.7 %) boys and 71 out of 176 (40.3 %) girls were enrolled in the study.

The major clinical characteristics describing respiratory failure severity in the examined children from both groups are presented in Table 4.

Children in Group 2 vs. Group 1 were characterized by significantly more severe RF course while assessing all clinical indicators related to RF severity: children more often required tracheal intubation (p < 0.001) and surfactant administration (p < 0.001) in delivery room; demonstrated a more pronounced need for routine mechanical lung ventilation through endotracheal tube (p = 0.029) and for transfer to HFOV (p < 0.001) within the first 72 hours of life. In delivery room, surfactant was administered to children usually at the 10th minute after birth (Table 5).

RF severity in premature neonates during early neonatal period is characterized by the need for applying additional oxygen during respiratory therapy, MAP level as well as requiring repeated surfactant administration within the first 72 hours of life. The latter was significantly more often (p = 0.031) required in Group 2 (Table 4). Moreover, children in Group 2 required significantly higher FiO₂ (p < 0.001). A need for cardiotonic support assessed by comparing peak MII values within the first 72 hours of life, did not differ significantly in both groups; related NEOMOD scale assessment revealed a more prominent MODS severity in Group 2 (p < 0.001), which undoubtedly affected neonatal morbidity (Table 6).

Children in Group 2 were more likely to develop multiple dermal ecchymoses and hemorrhages from birth (p < 0.001), pulmonary hemorrhage (p = 0.008), pneumothorax (p = 0.040), and surgical-stage retinopathy of prematurity (p = 0.001). Blood bacterial growth was reported only in 1 of 166 (0.6 %) children, from non-sterile foci (nasopharynx and anus) – in 61 of 168 (36.3 %) neonates. No significant differences between Groups 1 and 2 in bacterial growth from non-sterile loci (p = 0.228) were found. It is worth noting that in Group 2 vs. Group 1, Ureaplasma рarvum was significantly more often detected in the nasopharyngeal samples by PCR (p = 0.015).

Lethal outcome in total patient cohort was recorded in 7 of 169 (4.1 %) cases observed only in Group 2, with fatality rate for 7 of 34 children born alive comprising 21.0 % (OR = 73.91 [ 4.1; 1332.5]; OR = 11.57 [ 0.8; 167.4]). All children, IVH grade 3 among was recorded in 3.6 % (6 of 169 children) of cases, with that of found in Group 2 in 18.0 % (6 of 34 children) (OR = 61.81 [ 3.38; 1128.65]; OR = 11.57 [ 0.8; 167.4]). Surgical-stage NEC in total patient cohort during neonatal period manifested in 1 of 169 (0.6 %) children and, accordingly, in 1 of 34 (3.0 %) children in Group 2 (OR = 12.13 [ 0.48; 304.48]; OR = 3.21 [ 0.29; 35.44]). BPD incidence (oxygen requirement at 36 0/7 weeks of postconceptional age) was 7.1 % in total group (12 of 169 children) and 35.0 % – in Group 2 (12 of 34 children) (OR = 150.56 [ 8.61; 2633.61]; OR = 22.3 [ 1.47; 338.11]).

The indicators related to neonatal morbidity and outcomes in both groups were combined with temporal inflammation marker changes (Table 7).

Blood leukocyte count and NI during the first 24 hours of life in both groups were comparable whereas hemoglobin level in Group 2 was significantly lower (p < 0.001). Blood PCT and CRP level during the first 72 hours of life showed different patterns, however, both inflammation markers in children of Group 2 were significantly elevated. For instance, PCT magnitude during the first day of life in Group 2 was significantly higher (p = 0.017), that later (at the age of 24 to 48 hours after birth) became comparable with that of in Group 1, sustaining significantly higher magnitude in Group 2 at the age of 72 hours (p = 0.007). Blood CRP levels mirroring higher disease incidence in Group 2 were also significantly elevated in neonates with unfavorable outcomes between 24 and 48 hours after birth (p = 0.015). However, it was comparable in both groups during the first 24 hours and 48 to 72 hours of life.

To assess the impact of anhydrous interval duration on neonatal outcomes, a Spearman correlation analysis was performed (Table 8).

It was found out that anhydrous interval length impacts on neonatal outcomes more prominently in total patient cohort rather than specifically in each group. The longer anhydrous interval, the more significantly more often children subsequently required application of additional oxygen supply, had a higher NEOMOD score during the first three days of life, significantly lower blood hemoglobin levels and higher PCT magnitude during the first day of life.

Prenatal amniotic fluid index magnitude was found to be most powerful predictor of unfavorable outcome. This parameter determines unfavorable outcome with ROC-AUC accuracy = 0.791 (Table 9, Fig. 2). The study demonstrated that a cut-off level for unfavorable outcome was equal to 32.0 mm, i.e., if prenatal AFI ≤ 32.0 mm, then the probability for unfavorable outcome was higher; if prenatal AFI > 32.0 mm, then adverse outcome was observed at lower rate. The groups with unfavorable and favorable outcomes differed significantly in AFI levels before delivery > 32.0 mm (p = 0.001): group with unfavorable outcome: AFI before delivery ≤ 32.0 mm – 13 (76.47 %), AFI before delivery > 32.0 mm – 4 (23.53 %); group with favorable outcome: AFI before delivery ≤ 32.0 mm – 20 (32.26 %), AFI before delivery > 32.0 mm – 42 (67.74 %); OR = 6.82 [ 1.97; 23.58], p = 0.001.

Discussion / Обсуждение

Here, we set out to seek for predictors of adverse outcome in case of PROM development in PL. It allowed to reveal several clinical and laboratory variables that may be used in clinical practice as markers for fetal and neonatal negative prognosis. Unfavorable outcome in the fetus and premature neonate was significantly more often combined with obstetric indicators such as low gestational age and birth body weight, manifestations of lacked amniotic fluid (anhydramnios) and inflammatory process (chorioamnionitis), the need for delivery by CS, high CRP magnitude, as well as leukocytosis in pregnant women with PROM at the time of delivery, AFI before delivery ≤ 32.0 mm.

Low gestational age and lower birth weight objectively lead to a worse prognosis in premature neonates, as evidenced by numerous studies published in Russia and globally [28][29]. For instance, N. Younge et al. (2017) showed that among neonates born at 220–236 weeks of gestation, mortality within a few weeks was 97–98 %, so that as few as 1.0 % survived without neurodevelopmental disorders; among those born at 240–246 weeks of gestation, 55 % of neonates survive, but only 32 % had no neurological deficits at the age of 18–22 months of life [30].

Chorioamnionitis [9] and oligo- and anhydramnios are highlighted as most typical PROM-related maternal medical issues, with the latter often being combined with the need for CS delivery [17] that agrees with our data. It is worth noting that chorioamnionitis was significantly more common in children with unfavorable outcome that was not observed for fever in pregnant women before delivery, i.e. chorioamnionitis in some cases was diagnosed in the absence of fever. This finding was reported earlier as a subclinical chorioamnionitis accompanied by the lack of typical clinical symptoms, including fever. In this case, a subclinical infection may manifest, i.e., as a premature prenatal rupture of the membranes [31][32].

Elevated maternal CRP level before delivery we found in group with unfavorable outcome that was also paralleled with higher chorioamnionitis incidence further corroborate the data demonstrating that CRP is the most accurate marker of infectious and inflammatory diseases (sensitivity is 68.7 %, specificity is 77.1 %) [33]. Of special importance is the data on prenatal amniotic fluid index magnitude that turned out to be the strongest predictor of unfavorable outcome, which probability is higher if the prenatal AFI is ≤ 32.0 mm (ROC-AUC = 0.791). The studies discuss various AFI levels ranging from 1 to 10 cm used as prognostic criteria for maternal and neonatal outcomes [9][12][15–17]; however, prenatal AFI ≤ 32.0 mm found by us was first determined as a strong predictor for fetal and neonatal negative prognosis. A retrospective study by E.J. Okonek et al., conducted in the USA and published in April 2025, also parallels with our data on respiratory outcomes, particularly highlighting a need for frequent HFOV use, mechanical lung ventilation, surfactant administration, and treatment of pulmonary hypertension [34].

Thus, women with PROM are recommended to be referred to hospitals able to provide high-technology respiratory support to preterm neonates.

Not surprisingly, Group 2 neonates in our study were significantly more likely to exhibit MODS (including more severe RF) from birth, as evidenced by lower Apgar scores at the end of the 1^st and 5^th minutes of life, a higher need for invasive respiratory support starting in delivery room, for repeated surfactant administration, supplemental oxygen, and higher MAP scores at first 72 hours of life. We found that a more frequent combination of widespread ecchymosis in premature neonates at birth with adverse outcome may be considered as a negative prognosis predictor. Similar pattern we also observed for higher PCT and CRP magnitude in children during the first 72 hours of life as well as for lower hemoglobin level within the first 24 hours after birth and negative outcomes. A certain discovery was also that Ureaplasma рarvum was significantly more often detected in the nasopharyngeal swabs in children with negative prognosis.

Study limitations / Ограничения исследования

The limitations of our work were presented by retrospective study design which infer errors in data collection as well as the lack of chorioamnionitis histological verification. In terms of further prospective studies to validate the identified predictors, especially in multicenter cohorts, it is essential to take into account verification of chorioamnionitis diagnosis not only clinically, but also to align with histological data, especially in case of examining fever-free mothers.

Conclusion / Заключение

It may be concluded that lower gestational age and birth weight, anhydramnios and chorioamnionitis, CS, higher preterm CRP level and leukocyte count as well as AFI before delivery ≤ 32.0 mm are the major predictors for unfavorable perinatal outcomes in premature rupture of membranes. Severe hypoxia at birth and severity of multiple organ dysfunction, severe RF and postnatal multiple ecchymoses, neonatal Ureaplasma parvum, lower hemoglobin levels, and higher PCT and CRP levels during the first 72 hours of life are associated with unfavorable outcomes.