Clinical and anamnestic risk factors and prediction models for the development of fetal growth restriction

Fetal growth restriction (FGR) is a pressing issue in obstetrics and perinatology. FGR is one of the so-called “great obstetrical syndromes,” with consequences associated with a high incidence of perinatal morbidity and mortality [1]. The incidence of FGR varies from 5% to 23% [2, 3]. The causes of this complication include a variety of factors, including placental, fetal, maternal, and genetic factors as well as their combinations [4]. The pathophysiological basis of FGR related to the placenta is placental insufficiency, which manifests as decreased placental perfusion, reduced metabolic processes, and consequently impaired delivery of oxygen as well as various micro- and macronutrients to the fetus [5, 6]. Additionally, fetal malformations, chromosomal abnormalities, and congenital metabolic disorders can also contribute to FGR [7]. Thanks to advancements in molecular biology and genetics, variants in the genes of both the fetus and mother that may influence the development of FGR have been identified [4, 8].

The so-called "maternal factors" play a significant role in the development of FGR. These factors include advanced reproductive age, excess body weight, low socioeconomic status, teratogenic factors, and specific characteristics of obstetric and gynecological history [9–11]. Identifying predictors is crucial for developing methods to prevent pregnancy complications and improve and personalize the management of pregnant women affected by this pathology.

This study aimed to analyze the clinical and anamnestic risk factors and develop predictive models for early- and late-onset fetal growth restriction.

Materials and methods

This cross-sectional study was conducted at V.I. Kulakov NMRC for OG&P of Minzdrav of Russia. This study included 382 pregnant women who were divided into two groups. The first group (study group) consisted of 110 pregnant women with FGR, of which 34 had early-onset (subgroup 1A) and 76 had late-onset (subgroup 1 B) FGR. The second group (control group), consisted of 272 women with physiological pregnancies without FGR.

The study population consisted of pregnant women aged 18–45 years with a singleton pregnancy at 22–40 weeks, who provided informed consent. The exclusion criteria were the presence of somatic diseases in the decompensated stage, fetal malformations, and maternal acute infectious diseases.

The baseline evaluation included the patient's history of somatic and obstetric-gynecological diseases, as well as a detailed examination of the current pregnancy, encompassing all aspects of both obstetric and extragenital complications.

Statistical analysis

Statistical analysis was performed using the SPSS 25.0, StatTech v. 4.2.7 (Stattech LLC, Russia). Data were analyzed using nonparametric statistical methods. Qualitative features were presented as both absolute and relative values. The binary variables were compared using contingency tables. During the development of the mathematical predictive model, variance and discriminant analyses, the binary logistic regression method and ROC analysis were used. To assess the presence of a relationship between a certain outcome (FGR) and risk factors, crude and adjusted odds ratios with 95% confidence intervals (95% CI) were calculated. During the ROC analysis, ROC curves were constructed, the area under them (AUROC) was determined, and the sensitivity and specificity of the model were also calculated. The value of a quantitative variable at the cut-off point was determined using the highest Youden index value. Statistical significance was set at p<0.05.

Results

The age of the study participants ranged from 18 to 42 years. No statistically significant differences in age or anthropometric parameters were found between the groups or within the subgroups of the study group. The median age in the study subgroup 1A was 31.0 (29.0; 35.0) years, in subgroup 1B – 33.0 (29.0; 35.0) years. In the control group, the median age was 31.0 (28.0; 34.0) years. The median body mass index (BMI) in subgroup 1A was 24.0 (22.0; 26.0) kg/m², in subgroup 1B – 25.0 (23.0; 29.0) kg/m², and 25.0 (23.0; 26.0) kg/m² in the control group. A comparison of educational level, familial and social status, and presence or absence of unhealthy habits revealed no statistically significant differences between the study groups (p>0.05).

The two groups of patients were comparable in terms of age at menarche and nature of the menstrual cycle (p>0.05).

A comparative analysis of the structure of extragenital (somatic) pathology revealed that in group 1, the following were documented in the medical history significantly more often than in group 2: chronic arterial hypertension (10/110, 9.0%), chronic pyelonephritis (8/110, 7.3%), and chronic renal failure (5/110, 4.5%). The prevalence of chronic arterial hypertension was 1% in group 1 and 6/272 (2.2%) in group 2 (p=0.045, OR=0.226, 95% CI 0.052–0.979). Similarly, chronic pyelonephritis was observed in 8/110 (7.3%) patients in group 1 and 2/272 (0.7%) patients in group 2 (p=0.025, OR=0.094, 95% CI 0.010–0.865).

In terms of the frequency of history of gynecological pathology, the groups were comparable for most nosological forms, except for chronic endometritis [8/110 (7.3%) vs. 58/272 (21.3%), p=0.02, OR=0.82 (95% CI 0.015–10.353]) (by groups, respectively).

A comparative analysis of parity revealed that in group 2, there were significantly more women with a history of childbirth than in group 1 (p<0.05). There were no significant differences in the number of births (p=0.124) or pregnancy losses (p=0.308) between the study groups.

The diagnosis of FGR in previous pregnancies was significantly more often noted in group 1 than in group 2 [8/110 (7.3%) vs. 2/272 (0.7%), p=0.025, OR=0.094 (95% CI 0.010–0.865]). No statistically significant differences were found in the other complications of previous pregnancies.

The results of the analysis of the current pregnancy data in two subgroups (1A and 1B) of group 1 compared to group 2 are presented in Table 1. Statistically significant differences were found in the incidence of placenta previa for group 2 and subgroup 1A (p=0.003), as well as in the incidence of chronic arterial hypertension and preeclampsia for subgroup B and group 2 (p=0.023 and p=0.007, respectively).

The results of this analysis were used to develop mathematical models to predict the development of early- and late-onset FGRs in the pregnant mothers enrolled in the study.

The predictive model for the likelihood of developing early-onset FGR included the following indicators: pregnancy over 35 years of age, parity, smoking before pregnancy, FGR in previous pregnancies, obesity, threatened miscarriage, gestational diabetes mellitus, preeclampsia, placenta previa, and smoking during current pregnancy.

The analysis of the statistically significant dependence of the model predictors on the probability of early-onset FGR is presented in Table 2.

The obtained dependence can be described by the following equation:

P=1/(1+e-z) ×100%,

z=-2.565+0.455X_{pregnancy over 35 years of age}+0.299X_Obesity-1.416X_{Number of births} +2.821X_{Fetal growth restriction in previous pregnancies}+0.106X_{Threatened miscarriage}+0.969X_{Gestational diabetes mellitus}+0.235X_Preeclampsia+2.900X_{Placenta previa}+22.934X_{Smoking during pregnancy}-20.590X_{Smoking before pregnancy},

where P is the probability of developing early-onset FGR.

The resulting regression model was statistically significant (p=0.015).

Assessment of the dependence of the probability of developing early-onset FGR on the value of the logistic function P by ROC analysis yielded the following curve (Fig. 1).

The area under the ROC curve was 0.823±0.067, 95% CI 0.692–0.955. The model was found to be statistically significant (p<0.001).

The threshold value of the logistic function P at the cutoff point, which corresponded to the highest value of the Youden index, was 0.074. The probability of developing the early-onset FGR was predicted at a function value of P higher than or equal to this value. The sensitivity and specificity of this model were 93.3 and 62.5%, respectively (Fig. 2). A predictive mathematical model was also developed to determine the probability of developing late-onset FGR depending on parity, age at pregnancy over 35 years, smoking before pregnancy, obesity, threatened termination of pregnancy, gestational diabetes mellitus, preeclampsia, and smoking during the current pregnancy.

The observed dependence was described by the equation:

P=1/(1+e-z) ×100%,

z=-2.494+0.136X_{Number of births}+0.209X_{pregnancy over 35 years of age} +0.977X_Obesity+0.769X_{Threatened miscarriage}+0.029X_{Gestational diabetes}+1.283X_Preeclampsia+0.845X_{Smoking during pregnancy}+0.533X_{Smoking before pregnancy},

where P is the probability of developing the late-onset FGR.

The resulting regression model was statistically significant (p=0.004).

The ROC curve of the dependence of the probability of developing the late-onset FGR on the value of the logistic function P is shown in Figure 3; Table 3 presents the ROC analysis data.

The area under the ROC curve was 0.754±0.048, 95% CI 0.659–0.849. The resulting model was statistically significant (p<0.001).

The threshold of the logistic function P value at the cut-off point, which corresponded to the highest value of the Youden index, was 0.155. The development of late-onset FGR was predicted by a P value greater than or equal to this value. The sensitivity and specificity of the model were 81.6 and 62.7% (Fig. 4).

Discussion

The results of this study align with data from both foreign and Russian studies. Specifically, studies by Lees et al. and others have reported a relationship between obesity and chronic maternal hypertension and the development of FGR [12, 13]. Arterial hypertension is a common complication during pregnancy. Zhang H.G. et al. examined 100 newborns in the intensive care unit of mothers whose pregnancies were complicated by preeclampsia (n=53) and gestational hypertension (n=47). A control group of 100 healthy infants from the physiology department, born to mothers with uncomplicated pregnancies, was also included. The findings indicated that chronic hypertension and preeclampsia were significantly associated with the development of FGR, along with complications, such as placental abruption and fetal distress (p<0.05) [14]. In the present study, a history of chronic endometritis was significantly more common in pregnant women with FGR. The existing literature on the impact of chronic endometritis before pregnancy on the subsequent development of FGR is relatively limited. A study by Pestrikova T.Yu. et al., placental bed biopsies were taken from pregnant women with FGR and those with normal pregnancies who underwent cesarean sections. The results revealed that one-third of the patients in the FGR group showed no inflammatory changes in the placenta, whereas over 70% of the patients in the normal pregnancy group did. Statistically significant differences were also found between the groups in terms of the number of inflammatory cells and the area of necrotic foci based on the morphological study of the placental bed (p<0.05). The pathological reactions identified in the placental bed of pregnant women with FGR were interpreted by the authors as signs of chronic endometritis, manifesting during pregnancy as basal deciduitis [15].

Analysis of the current pregnancy course in our study found that complications, such as preeclampsia, threatened miscarriage, and chronic arterial hypertension, were significantly more prevalent in the FGR group than in the control group. Numerous studies indicate that the pathophysiology of preeclampsia and FGR is linked to mechanisms involving impaired cytotrophoblast invasion and remodeling of spiral arteries, which subsequently leads to significant structural changes in the placenta [16–18]. Consequently, preeclampsia is often associated with FGR during pregnancy, particularly in cases of early-onset FGR. In the present study, significant differences in the incidence of preeclampsia were observed only in the late-onset FGR group compared with the control group, which may be attributed to the small sample size of subgroup 1A (early-onset FGR).

The relationship between threatened miscarriages and FGR has been explored in various studies. A systematic review published in 2010 indicated that the risk of having a low-birth-weight baby was higher in women experiencing threatened miscarriage in the first trimester (OR=1.83, 95% CI 1.48–2.28) compared than in women without this complication (p<0.0001). The overall risk ranged from 1.1 to 3.7 [19]. Additionally, the association between threatened miscarriage accompanied by vaginal bloody discharge and FGR has been investigated. It is presumed that bloody discharge during pregnancy may indicate placental dysfunction, which could manifest at a later stage [20–22].

Thus, both the results of the current study and the conclusions of foreign colleagues indicate that threatened miscarriage, especially in the presence of vaginal bloody discharge, should be considered as a risk factor for the development of FGR. Among the analyzed clinical and anamnestic factors, the mathematical model for determining the probability of developing early-onset FGR included pregnancy at age > 35 years, parity, smoking before pregnancy, FGR in previous pregnancies, obesity, threatened miscarriage, gestational diabetes, preeclampsia, placenta previa, and smoking during current pregnancy. Statistically significant predictors of early-onset FGR in our study were previous pregnancies and placenta previa during the current pregnancy. In the presence of placenta previa, the chance of developing early FGR increased by 3.9 times. With a history of FGR, the chance of developing early FGR increased by 16.8 times. The predictive model for calculating the probability of developing late-onset FGR included the following variables: parity, pregnancy over 35 years of age, obesity, threatened miscarriage, gestational diabetes, preeclampsia during the current pregnancy, and smoking during and before pregnancy. Statistically significant predictors of late-onset FGR in our study were maternal obesity and preeclampsia during current pregnancy. The presence of maternal obesity during pregnancy increased the chances of developing late FGR by 2.7 times, while the presence of preeclampsia increased the chances by 3.6 times. It should be noted that factors such as threatened miscarriage and smoking before and during pregnancy tended to be statistically significant in the development of late FGR, and with an increase in sample size, they may attain statistical significance for this pathology. These factors in the sample of pregnant women we studied should be considered significant predictors of FGR.

Conclusion

The results of our study showed that women with chronic arterial hypertension, pyelonephritis, and chronic endometritis could be classified as a risk group for the development of FGR. This syndrome in previous pregnancies and placenta previa during the current pregnancy are significant predictors of early-onset FGR, increasing the risk of developing this pathology by 16.8 and 3.9 times, respectively. Maternal obesity and preeclampsia are significant predictors of late-onset FGR, increasing the risk of developing this pathology by 2.7 and 3.6 times, respectively. Thus, predictive mathematical models for the development of early-onset (sensitivity, 93.3%; specificity, 62.5%) and late-onset (sensitivity, 81.6%; specificity, 62.7%) FGR, developed based on the identified and most significant clinical and anamnestic factors using the binary logistic regression method, make it possible to reduce the incidence of pregnancy complications and improve perinatal outcomes.