Maternal risk factors associated with preterm birth after IVF/ICSI

In vitro fertilization/intracytoplasmic sperm injection (IVF/ICSI).is associated with an increased risk of preterm (33rd - 37th gestational week), and early preterm birth (20th - 32nd gestational week). The underlying general and procedure related risk factors are not well understood so far. 4,328 infertile women undergoing IVF/ICSI were entered into this study. The study population was divided into three groups: a) early preterm birth group (n=66), b) preterm birth group (n=675 ) and c) full-term birth group (n=3653). Odds for preterm birth were calculated by stepwise multivariate logistic regression analysis. We identied seven independent risk factors for preterm birth and four independent risk factors for early preterm birth. Older (>39) or younger (<25) maternal age (OR:1.504, 95%CI: 1.108-2.042,P=0.009; OR: 2.125, 95%CI: 1.049-4.304,P=0.036, respectively), multiple pregnancy (OR: 9.780, 95%CI: 8.014-11.935,P<0.001; OR: 8.588, 95%CI: 4.866-15.157,P<0.001, respectively), placenta previa (OR: 14.954, 95%CI: 8.053-27.767,P<0.001; OR: 16.479, 95%CI: 4.381-61.976,P<0.001, respectively), and embryo reduction (OR: 3.547, 95%CI: 1.736-7.249,P=0.001; OR: 7.145, 95%CI: 1.990-25.663,P=0.003, respectively) were associated with preterm birth and early preterm birth, whereas gestational hypertension (OR: 2.494, 95%CI: 1.770-3.514,P<0.001), elevated triglycerides (OR: CIs and P


Introduction
Preterm delivery accounts for more than 75% of perinatal morbidity and mortality worldwide 1 . Furthermore, those infants who do survive have higher rates of long-term morbidities, including cardiovascular diseases 2 as well as neurologic and developmental disabilities, compared to infants born full term. Known maternal risk factors for preterm birth in the general population include having a previous premature birth, twin pregnancy, an interval of less than six months between pregnancies, history of multiple miscarriages or abortions, smoking cigarettes or using illicit drugs, cardio-metabolic diseases such as hypertension or diabetes, and infections, particularly of the amniotic uid and lower genital tract 3-7. Conceiving through in vitro fertilization represents another risk factor in a subgroup of women undergoing assisted reproduction technologies (ART). The risk factors increasing the likelihood for preterm birth in this particular population are, however, as of today, not well understood. Numerous studies 8 analyzed maternal and offspring outcomes after ART, whereas larger sized studies focusing speci cally on the risk factors for preterm birthand especially early preterm birth especially associated to poor offspring outcome are lacking. However, this is a clinically important topic, since the understanding of ART related risk factors for preterm birth might identify changeable factors offering potential treatment options to improve offspring short term and long term outcome in women undergoing ART. The aim of the current study was to search for risk factors for preterm and early preterm birth in a large cohort of women who underwent ART.

Description of the cohort
From the primary dataset of 4349 treated women, we excluded 13 cases in which the gestational age was unclear and 8 cases that were post-term pregnancies. Thus nally 4,328 cases were included into the study, 3653 of them were full-term deliveries. The prevalence of preterm birth and early preterm birth was 15.5% and 1.8% respectively (Table 1 and Supplementary Figure 1    Moreover, univariate analysis comparing early preterm birth and full-term birth showed that 10 factors were signi cantly different between the full-term birth group and the early preterm birth group (P<0.05), including 3 maternal parameters (age, apolipoprotein A1, and thrombin time), 5 pregnancy related factors (multiple pregnancy, embryo reduction, and placenta previa), 3factors related to the IVF/ICSI procedure (blastocyst transfer, treatment cycles and number of embryos transferred), 1 offspring related factor (infant sex) ( Table 1-3).

Multivariate analysis
The above 14 detected factors in the univariate analysis were entered into the Logistic multivariate analysis. No other factors were considered as confounders. After stepwise regression analysis, multivariate analysis showed that 7 factors (older or younger maternal age, multiple pregnancy, embryo reduction,placenta previa, gestational hypertension, higher triglycerides and shorter activated partial thromboplastin time) were left in the multivariate analysis model for preterm birth. With regard to early preterm birth, the above described 10 factors, see above, were entered into the multivariate analysis. After stepwise regression analysis, 4 factors (older or younger maternal age, multiple pregnancy, embryo reduction and placenta previa) remained signi cant in the multivariate analysis model for early preterm birth (Table 4, supplementary table 1).

Discussion
Our study showed that older (>39) or younger (<25) maternal age, multiple pregnancy, placenta previa, and embryo reduction surgery were associated with an increased risk for both preterm birth and early preterm birth after IVF/ICSI. Gestational hypertension, higher triglycerides and a shorter activated partial thromboplastin time were only associated with an increased occurrence of preterm birth but not early preterm birth. However, the lack of some associations in the early preterm group was most likely due to the lower sample size.
In several studies it was already shown that ART is associated with adverse pregnancy outcomes, such as preterm birth 9 . The underlying risk factors for preterm birth in this population, however, are not fully established so far. The current study, investigating a cohort of women who underwent ART, identi ed several factors which were associated with preterm birth. The majority of the identi ed factors, such as maternal age 10 , multiple pregnancies, placenta previa, gestational hypertension, high triglycerides and hypercoagulability 11,12 , have previously been shown to be associated with an increased risk for preterm birth in the general population as well.
Our study showed an increased risk for preterm delivery in association with maternal age. Bothyounger and older women after ART treatment had an increase risk for preterm birth in our study as it was likewise seen in studies addressing this topic in the general population. Somewhat smaller study done in an ART populations mainly reported similar associations 13 . However, there is also a study with a similar study design as our study showing that women aged 25-29 were at an increased risk for preterm birth in comparison to women aged 30-34. Women aged ≥ 35 years did not display an increased risk of any type of preterm birth 14 . These previous ndings may suggest that while there is a positive association between maternal age and the risk for preterm birth, younger women who conceived via ART may display a higher risk for preterm birth compared to older women who conceived undergoing ART. As the proportion of women who conceive with ART also shows an age related increase, these results may also re ect the increased clinical risk of adverse birth outcomes among young women who needed ART to conceive 13 .
The current study also identi ed gestational hypertension as risk factor for preterm birth in women who needed ART to conceive. This nding is in line with the current literature, gestational hypertension was shown to be associated with an increased risk for preterm birth in both the general population and women who underwent ART. Moreover, it was shown that ART is associated with a higher frequency of gestational hypertension and preeclampsia as compared to natural pregnancy 15,16 .The underlying reasons for a higher frequency of gestational hypertension in ART pregnancies remain incompletely understood. Wang et al. showed in a large cohort comparing ART pregnancies to natural conception that ART is associated with a higher prevalence for gestational hypertension, yet this association disappeared once data was strati ed by multiple birth cases 17 . The authors concluded that multiple pregnancy which is associated with ART is the single most likely explanation for the increased rate of gestational hypertension among ART mothers.
Another risk factor for preterm birth found by the current study is placenta previa. Placenta previa is associated with preterm birth in the general population as well 18 . A study that investigated mothers who had conceived both naturally and via ART, showed that the risk of placenta previa was three-fold higher in the ART pregnancy 19 . The mechanisms underlying this phenomenon still have to be elucidated. It is hypothesized that ART related procedures, such as an induction of uterine contractions due to transcervical catheter insertion or the unique endocrinological environment with high estradiol concentrations following ART cycles might be responsible 20 .
Regarding maternal laboratory parameters, the current study demonstrated a positive association between maternal triglycerides, shorter activated partial thromboplastin time and preterm birth. Both high triglycerides and hypercoagulability were already shown to be associated with an increased risk for preterm birth in the general population 21,22 . Hypercoagulabilty and preterm birth has not yet been investigated extensively. Results from one small prospective study analyzing the relationship between maternal hypercoagulability and preterm labor in 76 women demonstrated a statistically signi cant procoagulant activity, expressed by a shorter prothrombin time and activated partial thromboplastin time, in pregnant women with premature uterine contractions who gave birth prematurely 22 . The presence of hypercoagulation before starting an IVF treatment was shown to be associated with negative IVF outcomes such as pregnancy loss. Our study clearly established hypercoagulability as an risk factor for preterm birth in ART and hence may help to clarify ongoing debates on this subject 23 .
With regard to triglycerides it was demonstrated that high maternal triglycerides levels during pregnancy are related to an increased risk for pretermbirth in both obese women and in women with normal BMI in the general population conceiving naturally 21 . There are studies that investigated lipid levels and ART outcomes showing that maternal triglyceride are inversely associated with live birth rate, however data regarding the relationship between maternal triglycerides and premature birth in the setting of ART are scarce 24 .
Our study is in good agreement with previous studies indicating that multiple pregnancy is a strong risk factor for preterm birth, both in the general population as well as in women who conceived trough ART 25 .
Multiple pregnancy is considered one of the largest hazard of ART. Until now ART is associated with a high number of multiple pregnancies due to the current policy transferring multiple embryossimultaneously to achieve a high pregnancy rate 26 . To reduce the risks associated with multiple pregnancy, embryo reduction is performed frequently. Previous meta-analyses have shown that embryo reduction improves outcomes in triplet pregnancies, but never to that degree of singleton pregnancies.
Aneffective method to reduce the risk of multiple births in ART is an elective single embryo transfer, a policy that is adopted by an increasing number of guidelines. However, studies also demonstrated that elective single embryo transfer is not associated with a reduction in the risk for preterm delivery 27 .
In conclusion, our studydemonstratedthat maternal age, multiple pregnancy, embryo reduction and placenta previa could increase the risk of preterm birth in womenundergoing IVF/ICSI. During ART treatment, the numbers of embryo transfers per cycle should be reduced to two or even to one thus reducing the need for embryo reduction procedures or multiple pregnancy. Strengthening antenatal care is necessary during pregnancy, especially for the patients with placenta previa. The nding that coagulation abnormalities are linked to preterm birth needs independent con rmation and if con rmed may stimulate clinical trials testing drugs interfering with the coagulation system as it is for example done in pregnant women suffering from activated protein C resistance 28 .

Methods
Ethics, inclusion and exclusion criteria, data collection The Note: Gestational age was determined as the gestational 17 th day when 6-8 cell embryo was transferred into the uterus, and gestational 19 th when blastocyst was transferred into the uterusafter treatment in the fresh embryo transfer recipients.
Full-term birth was de ned as a live birth with a gestational age between 37 but not over 42 weeks (37 weeks ≤gestational age<42 weeks). Preterm birth was de ned asalivebirthwith a gestational age of at least 20but not over 37 weeks (20 weeks ≤gestational age<37 weeks) 31 . Early-preterm birth was de ned as a live birth with a gestational age between 20 but not over 32 weeks (20 weeks ≤gestational age<32 weeks) 32 .
All patient's data (clinical data as well as laboratory data) used in our study were extracted from the routine electronic patient records used in our hospital.
Clinic data collection A structured medical history was taken. The following risk factors for preterm birth were examined in this study: Basic parameters about the mother, pregnancy history, gynecological complications, and relevant risk factors during IVF/ICSI procedure came from the case report in the hospital. Furthermore, blood test results from maternal blood taken before the beginning of superovulation was extracted from the case report. Pregnancy related factors and offspring data were followed up strictly by special nurse.

Patient and Public Involvement
This study is a retrospective study. Data were obtained through the electronical medical record system of the hospital. Patients were not directly involved in this study. The patients were unaware of the results of the study.

Statistical analysis
Continuous variables are represented as mean±standard deviation for normally distributed variables and student's unpaired t-test was used for comparison of variables between two groups. Continuous variables are represented as median and quartiles M(Q 1 -Q 3 ) for non-normally distributed variables and Mann-Whitney nonparametric test was used for comparison of variables between groups. Categorical variables are described as frequency and percentages. Pearson's chi-square test was used for testing qualitative data and Fisher's exact test was used when the expected frequencies were <5%.
Multivariate Logistic regression analyses with step forward selection using the likelihood method were applied to examine the association between the patient's characteristics and the risk of preterm birth.
Analyzed variables with P<0.05 in the univariate analysis were entered into the multivariate analysis. No other factors were considered as confounders. Results are represented as ORs with corresponding 95% CIs and P values.
Statistical package for social sciences (SPSS version 22.0, Chicago, IL, USA) was used to perform all data analyses and a two-sided P value <0.05 was considered to be statistically signi cant.
All methods were carried out in accordance with relevant guidelines and regulations of the People Republic of China.