Missed abortion is a common obstetrical condition with a 20% incidence of all clinically diagnosed pregnancies. It has received considerable attention in the past two decades as there is an increased focus on medical rather than surgical options. This is due to the introduction of new drugs, such as misoprostol and mifepristone, which have emerged as effective options for medical termination. Medical termination carries the advantage of outpatient treatment, lifting much of the burden on maternity hospitals [1, 2]. Misoprostol, a synthetic prostaglandin agonist originally used to treat peptic ulcer, was accidentally discovered to be an abortifacient agent. Compared with a surgical method, it has been the focus of many studies, regarding its dosing and route of administration, either alone or in combination with other drugs, with varying success rates and safety profiles [3, 4]. Treatment failure is a well-acknowledged subject by most obstetric practitioners and has been investigated by many researchers; some relate it to maternal parameters and others to the drug chosen. Ultimately, those with failed treatment will need dilatation and curettage [5]. Predicting the causes of treatment failure has gained much interest. The first report (in the form of a case report in non-conclusive concepts) on the successful use of Doppler to predict the medical option was produced in the past few decades. In 2009, an evaluation of the effectiveness of uterine arteries in predicting pregnancy outcome was reported in the Journal of Reproduction of Domestic Animals, after which many studies on human reproduction emerged, relating pregnancy outcome and the fate of on-going pregnancy [6].

Fortunately, uterine artery Doppler is easily evaluated in obstetrics and gynecology (OBG) clinics; the changes in the normal pattern of uterine artery indices, PI and RI, during the first trimester were recorded. Doppler study stands out as a non-invasive investigation with high acceptance rates and minimal side effects [7]. Little is known about the role of Doppler in predicting the success of medically terminated first-trimester abortion, predicting the causes remains a challenge. We hypothesized that a doppler study, measuring both resistance index (RI) and pulsatility index (PI), of uterine arteries, performed before undergoing treatment might predict the success of medically terminated first-trimester missed abortions and the time required to complete it. The importance and originality of this study are that it explores the relationship between Doppler indices in patients with a first trimester missed abortion and the success of medical termination, along with carefully selected maternal parameters that are valuable to clinicians.

Statistical analysis of data was performed using SAS (Statistical Analysis System-version 9.1). The test of normality, using the Shapiro-Wilk test, showed that the data were not normally distributed; therefore, the means were compared, using the Mann-Whitney test. Chi-square test or Fisher test (as applicable) was used for the descriptive measures. P 0.05 was considered statistically significant. The criterion between RI and PI was assessed by using the partial least square curve (PLS) and the coefficient marker of matrix mathematics. Thus, the eigenvalue was used for multiple independent parameters, such as PI and RI and the dependent parameter, the gestational age of the dead fetus. Positive eigenvalues above 1 indicate a correlation between the two studied parameters, especially the higher positive values. On the other hand, eigenvalues below 1 or in the negative range indicate that there is no correlation between the two measured parameters.

A receiver operating characteristic curve was used to identify the validity of the marker (RI) as an indicator of pregnancy loss. The marker cut-off value was generated according to the area under the curve. The analysis was performed using MedCalc Software. The sensitivity (a probability that the test will be positive when the infection is present).

The specificity (a probability that the test will be negative when the infection is absent).

MedCalc Statistical Software version 16.4.3 (MedCalc Software bvba, Ostend, Belgium; https://www.medcalc.org; 2016) SAS.2010.SAS/STAT Users Guide for Personal Computer. Release 9.13.SAS Institute, Inc., Cary, N.C., USA.

Finally, Poisson regression was used to measure the risk ratio of surgical intervention.

A total of 78 patients with a confirmed first-trimester abortion participated in this study; some of their maternal parameters are presented in Table 1. The age, BMI, and gestational age of dead fetus were not statistically significant with P = 0.13, P = 0.13, and P = 0.35, respectively; however, parity and delivery mode were significant with P 0.0001 and P 0001, respectively, which were also significant for all participants and the individual groups. The differences in the means of PI and RI are highlighted in Table 2; it can be seen that PI for Group I plus II and Group III are 1.53 (0.75-2.70) and 1.58 (1.10-2.10), respectively. On the contrary, RI for Group I plus II and Group III are 0.71 (0.50-1.00) and 0.80 (0.69-0.92), respectively. Closer inspection of the table shows that RI has a better predictive value than PI with P = 0.02 and P = 0.62, respectively.

The PLS curve in Fig. 2 shows the correlation between Doppler indices, PI and RI, as independent variables and the gestational age of the dead fetus as a dependent variable in the successful termination of abortion. Surprisingly, RI was the only parameter to reach a statistically significant correlation with the gestational age of the dead fetus with an eigenvalue of 2.77, contrary to that of PI (-0.8), ascribe as a successful predictor.

Fig. 2.

Partial PLS curve for generating a coefficient of correlation, Eigenvalues; both doppler indices were challenged to detect the most meaningful correlation, it was illustrated as RI to be a significant predictor for successful termination.

Receiver operating curve (ROC) analysis was used to calculate the cut-off value for RI; the result was 0.74, any reading below this number is more likely to have a successful medical induction of abortion, as shown in Fig. 3.

Fig. 3.

Receiver operation curve ROC for RI, it was generated to assess the validity of RI as a marker for termination success, a cut off value of RI of 0.74 was concluded with its associated sensitivity and specificity of 68.7, 56.7 respectively.

As shown earlier, not all the patients aborted within the first week; some failed and continued to the second and third weeks; hence those who aborted in the first week were considered the reference group. A risk ratio was calculated with the aid of Poisson regression to accurately predict the risk of surgical intervention specific to each group; as shown in Table 3, the risk ratios were estimated to be 1.3 and 4.7 for the second and third weeks, respectively.

Unique adaptations in placental morphology and function have increasingly been acknowledged and well described with Doppler-derived hemodynamic parameters. PI and RI are commonly used to monitor fetal well-being in clinical practice, but whether they significantly influence the type of termination of missed abortion have not been adequately investigated.

Increased parity is among the maternal parameters that best predict good responders. Prior studies by Kim et al. portrayed such association, which is contrary to that of Reeves et al. and Ehrnstén et al., who stated that the possibility of a successful outcome decreases as gestational age and parity advance since the stretching of uterine muscles and loss of muscle integrity lead to poor response to medical termination. This relationship was most striking in $\ge$ para 5 patients; this may explain the apparent contradiction with the current study as the mean parity was 4 in our patients [8, 9, 10]. This study shows there is a strong relationship between predicting good responder and vaginal delivery. This agrees with the results of Kim et al., who declared that vaginal delivery was associated with higher and more rapid success rates. Caesarian delivery tends to alter the success of a good responder because the scarred site will be considered a weak point in disturbing uniform myometrium and its contractility, not to overlook the increased penetration by trophoblasts to the myometrium, making it difficult to expel [8, 11]. Maternal age and gestational age of the dead fetus are not statistically significant; the same finding was reported in a study by Kim et al. In contrast, earlier findings by Ehrnstén et al. suggested that most distinct factor for surgical intervention risk was a 9-week gestational age or more at the time of treatment; the study results do not support this. This inconsistency may be due to the larger sample size. Moreover, the mean age of the participants was (8.4) weeks less than the cut-off value proposed by Ehrnstén L et al. [8, 10].

From the PLS curve and its generated eigenvalue, the RI but not the PI stands out as significantly correlated to the success of medical termination with a cut-off value of 0.74; below which medical treatment will be successful, and above which medical treatment will most likely fail, with increased hazards of surgical intervention. It is promising when this figure is compared with the one expressed by Sonek et al. where both RI and PI correlate with the success rate; this may be explained by the larger sample size used in their study [12]. There are similarities between the results of this study and the results of Guedes-Martins et al. that described the use of PI and RI as a predictor of medical management of incomplete abortion. Their results suggested that retained products of conception will affect PI and RI, causing both to increase, necessitating further surgical intervention. The researchers used both mifepristone and misoprostol for treatment; different settings, drugs, and another type of abortion were used. In addition, in line with this conclusion, Hameed et al. used color Doppler imaging instead of PI and RI in managing cases of incomplete abortion [13, 14].

The use of PI and RI were the focus of many studies that investigated their effect on predicting pregnancy and delivery outcomes rather than predicting abortion outcomes as in this study. RI was studied; its reference value was established before and in early pregnancy. Women who were deemed to have missed abortion were classified; for every 0.1 increase in RI, there is an associated 18.7 increase in the risk of abortion [1, 15].

As for those deemed to be experiencing pregnancy-induced hypertension (PIH), intrauterine growth restriction (IUGR) were reported to have higher Doppler indices during the pre-pregnancy and early pregnancy states. This was the basis for the therapeutic trial of added aspirin and progesterone drugs that decrease their levels close to the level of the general population [16, 17, 18]. Trayanov et al. compared RI levels in patients with threatened abortion and normal population reporting no statistical differences. Surprisingly, this was not observed in the study by Khong et al. that reported a high and significant RI upon a comparison between normal population and patient deemed to have a first trimester missed abortion and later on PETand IUGR [16, 19]. It should be highlighted that there are discrepancies in RI values that have been reported in different literature. This can be attributed to the differences in study design, sample size, and, most importantly, the criteria for selecting enrolled subjects. A deeper insight reveals that the outcome will be more meaningful if enrolled subjects were selected from the high-risk group rather than the general population. Furthermore, if these subjects were combined with maternal characteristics, the validity will be substantially higher [20]. To explain the study results, a search was made for factors that influence the RI, thereby causing the RI to rise and producing a poor respondent who is resistant to medical termination.

The RI is calculated as the ratio between the mean pressure and the mean volume of blood flow. Consequently, an increase in blood viscosity or an increase in the resistance or a reduction in the radius of blood vessels will be a probable cause [7]. A wide range of causative factors is implicated, keeping in mind the possibility of interference from multiple factors. RI values represent changes in the entire system and are unlikely to be altered by a single factor. Some of these causative factors include oxidative stress, hypoxia, alteration in placental gene expression and endothelial injury, failure of the second wave devascularization of spiral arteries in placental bed, and local release of vasoactive substance [19, 21].

The ROC analysis produced a cut-off point of 0.74 for RI, with associated sensitivity and specificity of 68.7 and 56.7, respectively. The relatively low sensitivity and specificity reached can be explained by the sample distribution. Groups I and II had relatively comparable numbers while group III had a markedly low number. This discrepancy in the distribution led to sampling bias. Moreover, sampling was made from one center; more diverse patients from multiple centers could have added the variance needed to achieve better sensitivity. Another probable factor is the heterogeneity of the causes of a first trimester abortion, and heterogeneity in the modified pattern of vascularity of the myometrium during early pregnancy, which affects artery indices. A bigger sample size and a more strict gestational age would have added more credibility to the results and not compromised them. It is advisable to overcome these challenges as they limit the validity of the conclusions reached.

We established the risk of surgical intervention on a poor responder. Those who aborted in the first week (good responders) were considered as the reference group. From that group, based on the weeks, a Risk Ratio was generated to predict the need for surgical intervention. It was clear that the more time elapses, the higher the risk was since it increased from 1.4 to 4.6 in the second and third weeks, respectively. The novelty of this study is that most of the earlier studies addressed the estimated RI in relation to the prosperity and health of on-going pregnancy. So far, very little attention has been paid to the role of Doppler in predicting response for a medical termination among patients with missed abortions. To the best of our knowledge, this is the first study, at least in Iraq, to correlate maternal and Doppler attributes in designing predictive module for patients with missed abortion. This study may serve as a base for further research that will address the implication of the RI in the termination of pregnancies at a higher gestational age.

Abortion is commonly practiced. There is a challenge to its management, whether to operate or to wait for a medical termination. Doppler indices were proposed as a means to predict a patient’s fitness for the best medical option. The main finding of this study is that the RI measurement may serve as an effective predictor of successful termination, along with increased parity and vaginal delivery. These findings may help patients to make the best and the most appropriate choice, guided by their clinician, based on their demographic characteristics and Doppler indices.

Wassan Nori: wrote the paper, designed the concept, collected the data, scientific editing and revision. Muna Abdulghani: collected the data, performed the experiments and dopplers study, scientific editing and revision. Ali B. Roomi: designed the concept, scientific editing and revision. Wisam Akram: conducted the statistical analyses, scientific editing and revision.

The study was supported by the staff of AL-Yarmouk hospital in the collection of samples that are gratefully acknowledged.

The authors declare no competing interests.