- Top of page
- Disclosure of Conflict of Interests
Summary. Objective: To conduct a pilot randomized controlled trial of unfractionated heparin (UFH) in women considered at high risk of placental insufficiency in the second trimester. Methods: Women with either false-positive first trimester (pregnancy-associated placental protein-A [PAPP-A] < 0.35 MoM) or second trimester (alpha-fetoprotein [AFP] > 2.0 MoM, inhibin > 3.0 MoM, human chorionic gonadotropin > 4.0 MoM) serum screening tests or medical/obstetric risk factors were screened for placental insufficiency by sonographic evaluation of the placenta and uterine artery Doppler between 18 and 22 weeks. Thrombophilia screen-negative women with two or three abnormal test categories were randomized by 23+6 weeks to self-administration of subcutaneous unfractionated heparin (UFH) 7500 IU twice daily until birth or 34 weeks, or to standard care. Maternal anxiety and other maternal-infant outcomes were determined. Results: Thirty-two out of 41 eligible women consented, with 16 women randomized to UFH and 16 to standard care. There was no statistically significant difference identified between the two treatment groups (standard care vs. UFH) for the following: maternal anxiety score (mean [standard deviation]), 14.2 [± 1.6] vs. 14.0 [± 1.8]; birth weight (median [range]), 1795 [470–3295]g vs. 1860 [730–3050]g; perinatal death, 3 vs. 0; severe preeclampsia, 2 vs. 6; placental weight < 10th percentile, 7 vs. 4; or placental infarction, 4 vs. 3. Conclusion: Our study design identified women at high risk of adverse maternal-infant outcomes attributable to placental insufficiency. Women with evidence of placental insufficiency were willing to undergo randomization and self-administration of UFH without increased maternal anxiety.
- Top of page
- Disclosure of Conflict of Interests
Prophylactic anticoagulation during pregnancy, with the aim of preventing adverse maternal-infant outcomes associated with placental insufficiency, is an attractive concept proposed over 35 years ago . It is based on the observation that the extent of placental infarction correlates with the severity of preeclampsia and intrauterine growth restriction (IUGR) , and the assumption that heparin effectively prevents placental infarction . We recently performed a systematic review of the literature  that identified four randomized controlled trials, involving 324 women, where heparin (alone or in combination with dipyridamole) was compared with no treatment [4–7]. Collectively these studies found no significant differences in perinatal mortality or preterm birth < 34 weeks’ gestation, though heparin significantly reduced the risk of preeclampsia and birth weight < 10th percentile .
The original hypothesis that heparin promotes improved pregnancy outcomes via its anticoagulant properties within the spiral arteries and the intervillous space has never been adequately tested because, to date, no trial has included placental pathology. Among women with confirmed placental infarction, the rates of severe preeclampsia and IUGR are high, yet only a subset have a detectable thrombophilia disorder . Despite common minor side effects such as painful bruising and the rare but serious complication of vertebral collapse from osteopenia, thrombophilic women with previous pregnancy complications are increasingly prescribed prophylactic subcutaneous doses of either unfractionated heparin (UFH) or low-molecular-weight heparin (LMWH) in subsequent pregnancies . This practise lacks evidence and is the subject of an ongoing clinical trial (http://clinicaltrials.gov/ct2/show/NCT00967382). Preliminary evidence that heparin may exert beneficial effects in women with a negative thrombophilia screen  or promote improved implantation with in-vitro fertilization  underscores the importance of defining the potential benefits and risks of heparin in women considered at risk of placental insufficiency from adequately-powered randomized controlled trials.
Placental infarction is rare in otherwise normal pregnancies. In a recent single-centre retrospective cohort of 180 singleton pregnancies with placental infarction, the majority of placentas were small and exhibited histologic abnormalities in both the utero-placental vascular bed and in the development of the placental villi . These observations suggest that normal placental development confers intrinsic anticoagulant properties, via both hemostasis regulation at the surface of the villi and the attainment of a high utero-placental blood flow velocity within the placenta due to erosion of the distal segments of the spiral arterioles by the invasive extravillous trophoblast cells . The ability to non-invasively demonstrate several elements of placental function in the first half of pregnancy, in order to define the subsequent risk of adverse perinatal outcomes due to thrombotic vascular disease of the placenta, forms the rationale for restricting the evaluation of the role of heparin to a subset of so-called high-risk pregnant women with multiply-abnormal test results . These tests are presently available during standard antenatal care and comprise three categories: first, re-interpretation of the maternal serum screening tests for spina bifida and Down’s syndrome ; second, uterine artery Doppler at 19–22 weeks ; and third, placental morphology (size, shape and internal texture) . Used alone, these tests do not achieve sufficient test characteristics to merit introduction as a screening program for adverse perinatal outcomes , yet when combined, can identify a subset of women with over 40% positive predictive values for serious adverse maternal and/or perinatal outcomes [12,15]. Therefore in our experience, over 60% of so-called high-risk women considered at risk of placental insufficiency based solely on clinical risk factors have normal tests of placental function that over-ride their background risks, to result in near-term favourable outcomes . We therefore adopted the rationale of placental function testing in order to restrict the definition of ‘high-risk women’ to a subset of women with multi-parameter evidence of placental dysfunction in the second trimester. The aims of our pilot randomized trial were to assess the feasibility of this recruitment strategy and to monitor its impact upon psychological health.
- Top of page
- Disclosure of Conflict of Interests
During the study period, 41 eligible women were approached, of whom 32 were randomized (16 women to the standard care group and 16 women to the UFH group). Reasons for not participating in the trial included women of non-English speaking background who were unable to give informed consent, an unwillingness to agree to a 50% risk of self-administration of heparin, or partner objection to the potential risk of heparin. No eligible patient declined randomization to seek heparin elsewhere and all eligible women continued care in our clinic. Women allocated to self-inject UFH did not complain of significant discomfort and none discontinued UFH prematurely. Self-injection was ascertained at each visit and could not be ascertained visually in only 1/16 women due to dark skin color. No woman allocated to standard care was found to have any bruises suggestive of heparin administration. Complete pregnancy and birth data were available for all 32/41 eligible women randomized as shown in Fig. 1. No newborn had aneuploidy or any major congenital abnormality.
Baseline characteristics for the two treatment groups at the time of trial entry were comparable. The distribution of categories of abnormal placental function tests at trial entry was similar between the two treatment groups. However, within the category of abnormal maternal biochemistry, more women with elevated hCG or Inhibin were allocated to the UFH arm (Table 1). Seven women in the standard care group (43.8%) and six women in the UFH group (37.5%) had abnormalities in all three categories (maternal biochemistry, placental morphology and uterine artery Doppler), as shown in Table 2.
Table 1. Maternal demographics at trial entry
| ||Standard care (n = 16)||UFH (n = 16)|
|Gestational age at randomization (weeks)||22.1 (18.6–23.4)||22.6 (18.1–23.9)|
|Maternal age (years)||35 (25–42)||33.5 (25–42)|
|Nulliparity||4 (25)||6 (37.5)|
|BMI (kg m−2)||27.0 (20.4–41.7)||28.3 (20.0–45.5)|
|BMI ≥ 30||4 (25)||8 (50)|
| Caucasian||8 (50)||6 (38)|
| Asian||7 (44)||6 (38)|
| African||1 (6)||4 (25)|
|Previous complex obstetric history*||8 (50)||6 (37.5)|
| Stillbirth (> 20 weeks)||4 (25)||4 (25)|
| Preterm delivery (< 32 weeks)||2 (12.5)||3 (18.8)|
| ≥ 2 pregnancy losses < 20 weeks||1 (6.3)||1 (6.3)|
| Previous placental abruption||0||1 (3.1)|
| Previous preeclampsia||2 (12.5)||1 (6.3)|
| Previous SGA infant||6 (37.5)||1 (6.3)|
|Complex medical history||2 (12.5)||0|
| Chronic renal disease||1 (6.3)||0|
| Chronic hypertension||2 (12.5)||0|
Table 2. Distribution of abnormal placental function tests at randomization
| ||Standard care (n = 16)||UFH (n = 16)|
|Abnormal maternal biochemistry||13 (81.3)||15 (93.8)|
| PAPP-A < 0.35 MoM*||8 (57.1)||8 (53.3)|
| AFP > 2.0 MoM†||9 (60)||10 (71.4)|
| Inhibin > 3.0 MoM‡||1 (14.3)||4 (80)|
| Total hCG > 4.0 MoM§||1 (7.7)||8 (61.5)|
|Abnormal placental morphology||14 (87.5)||11 (68.8)|
| Thickness > 4 cm or >50% maximum length||6 (37.5)||4 (25)|
| Maximum length < 10 cm||9 (56.3)||6 (37.5)|
| Heterogenous texture/ECLs||7 (43.8)||4 (25)|
|Abnormal uterine artery Doppler||12 (75)||12 (75)|
|All three categories abnormal||7 (43.8)||6 (37.5)|
Maternal health outcomes
There were no statistically significant differences identified between the two treatment groups in the mean anxiety score, or in the number of women with an EPDS score ≥ 12 (Table 3). There were no statistically significant differences identified between the two treatment groups in the occurrence of all preeclampsia, or severe preeclampsia ± HELLP syndrome. Only four women (12.5%) had healthy term pregnancies (delivery > 37 weeks without evidence of preeclampsia or IUGR). In five women (15.6%), the lowest platelet count before delivery was < 120 × 10−9 L−1. This was asymptomatic in four women, while one woman developed an abruption. She was admitted with mild vaginal bleeding and some uterine activity. She received Celestone for fetal lung maturation. Two days later, an abruption was clinically evident and she required Caeserean delivery under general anesthesia. A significant abruption was confirmed.
Table 3. Maternal outcomes
|Variable||Standard care (n = 16)||UFH (n = 16)||RR (95% CI)||P value|
|Mean anxiety score||14.2 ± 1.6||14.0 ± 1.8|| ||0.574|
|EPDS ≥ 12||1 (6.3)||2 (12.6)||1.66 (0.16–17.10||0.641|
|Caesarean delivery||8 (16)||9 (56.3)|| ||1.000|
|Severe preeclampsia||2 (12.5)||6 (37.5)||0.33 (0.08–1.41)||0.220|
|HELLP||0||1 (6.3)||0.33 (0.01–7.62||1.000|
|Placental abruption||0||1 (6.3)||0.33 (0.01–7.62)||1.000|
Delivery and infant health outcomes
No statistically significant differences were identified between the two treatment groups in gestational age at birth, Apgar score of < 7 at 5 min of age, infant birth weight < 3rd centile, perinatal mortality, neonatal intensive care unit admission, or composite neonatal morbidity (Table 4).
Table 4. Neonatal outcomes
|Variable||Standard care (n = 16)||UFH (n = 16)||RR (95% CI)||P value|
|Mean gestational age at birth (weeks)||35.6 (28.4–38.3)||34.3 (27.9–38.4)|| ||0.481|
|Preterm birth at < 32 weeks gestation||5 (31.3)||5 (31.3)||1.00 (0.36–2.79)||1.000|
|Male||9 (56)||9 (56)|| ||1.000|
|Median birth weight (g)||1795 (470–3295)||1860 (730–3050)|| ||0.759|
|Birth weight < 10th centile (SGA)||12 (75)||6 (37.5)||2.00 (1.00–4.00)||0.073|
|Birth weight < 3rd centile||5 (31.3)||4 (25)||1.25 (0.41–3.82)||1.000|
|IUGR||4 (25)||1 (6.3)||4.00 (0.50–31.98)||0.333|
| 1||8 (0–9)||8 (1–9)|| ||0.928|
| 5||9 (0–9)||9 (6–9)|| ||0.585|
|NICU admission*||7 (46.7)||10 (62.5)||0.70 (0.36–1.37)||0.479|
|Perinatal mortality||3 (18.8)||0||7.00 (0.39–125.44)||0.226|
|Composite neonatal morbidity*||3 (20)||3 (18.8)||1.00 (0.24–4.23)||1.000|
A total of 31/32 placentas were submitted for pathological examination by a dedicated perinatal pathologist (SK) who was blinded to treatment allocation. Only five placentas were normal at delivery (weight > 10th percentile for gestation with no gross lesions; UFH group n = 4; standard care group n = 1). No statistically significant differences were found between the two groups for gross pathology or vascular lesions (infarction, intervillous thrombosis and hemorrhage) (Table 5).
Table 5. Placental pathology
|Variable||Standard care* (n = 15)||UFH (n = 16)||P value|
| Placental weight < 10th centile||7 (46.7)||4 (25)||1.000|
| Accessory lobe||1 (6.7)||1 (6.3)||1.000|
| Two vessel cord||3 (20)||0||0.101|
| Velamentous umbilical cord insertion†||3 (21.4)||1 (6.7)||0.330|
| Infarction||4 (26.7)||3 (18.8)||0.685|
| Intervillous thrombosis||7 (46.7)||2 (13.3)||0.109|
| Large retroplacental or subchorionic hemorrhage||5 (33.3)||1 (6.3)||0.083|
- Top of page
- Disclosure of Conflict of Interests
The increasing use of prophylactic doses of heparin in the belief that it will improve a range of pregnancy outcomes attributable to placental dysfunction is concerning for several reasons, including cost, side effects, and provision of safe regional anesthesia at delivery. The range of inadequately powered randomized controlled trials suggesting clinical benefit encompasses implantation at in vitro fertilization , and perinatal outcomes in both thrombophilia screen positive  and negative [5,31,32] women with prior adverse outcomes. Conversely, two recent trials found no benefit of heparin in preventing recurrent miscarriages in either the presence or absence of thrombophilia [31,33]. The rationale of such studies is based on the assumption that heparin prevents placental damage via an anticoagulant effect, though none of these studies assessed the target organ by performing placental pathology at delivery. Furthermore, the low rate of serious adverse perinatal outcomes in these trials suggests that a majority of participants probably had adequate placental function to sustain their near term pregnancies.
Recognizing the need to develop a cost-effective strategy to address the role of heparin in this context, we performed a pilot randomized controlled trial in women with a prenatal diagnosis of placental insufficiency in the second trimester. The inclusion criteria were based on our previously validated model for identifying women with positive predictive values for adverse perinatal outcomes attributable to placental dysfunction up to 50%  and a high likelihood of ischemic-thrombotic placental pathology . We excluded a minority of otherwise eligible women due to participation in an ongoing multicenter trial in thrombophilia screen-positive women (http://clinicaltrials.gov/ct2/show/NCT00967382) and chose a higher (7500 IU/BD) dose of the low-cost unfractionated type of heparin to explore efficacy. We achieved a high rate of enrollment with no postrandomization cross-over in either direction. We were able to analyse 31/32 placentas after delivery. As expected in this pilot trial sample size, we found no statistically significant differences in the prespecified maternal and infant health outcomes. Importantly, there were no differences identified in maternal anxiety or risk of depression during pregnancy.
The presence of maternal anxiety has been well recognized among healthy pregnant women and demonstrated to be associated with negative expectations about motherhood , poorer adjustment to motherhood  and postpartum mood disorders, particularly postpartum depression [36,37]. Furthermore, there is a well-described literature identifying that women experience increased anxiety if they have pregnancy complications, including miscarriage , fetal aneuploidy  and need for invasive prenatal testing [40,41] or investigation , in addition to medical complications, including gestational diabetes . The use of the six-item short form Spielberger State-Trait Inventory Self Evaluation Questionnaire  has been well reported and validated as a reliable assessment of anxiety during pregnancy. Importantly, our pilot randomized trial did not identify any increase in maternal anxiety score or risk of depression in this high-risk antenatal population, attributable to regular self-administration of subcutaneous injections.
The strengths of this pilot trial are that it utilized high quality randomized trial methodology, and provided valuable information regarding the feasibility of any further definitive studies. Our findings are consistent with those presented in the Cochrane Systematic Review evaluating the role of antithrombotic therapy for women at risk of placental dysfunction . When our data and that of Gris et al.  are incorporated into the meta-analysis for the outcome of perinatal mortality, there is no statistically significant difference identified between the use of heparin therapy and no treatment, although the combined available sample size is underpowered (five studies, 429 participants; 6/228 heparin therapy vs. 17/201 no treatment; risk ratio 0.42; 95% CI 0.18–1.01). Utilizing the information available from this updated meta-analysis, a definitive trial with sample size of approximately 1200 women would be required to detect a 50% reduction in the risk of perinatal death from 8.46% to 4.23% (alpha 0.05; power 80%). A similar sized trial would be able to detect a more modest reduction of 30% in a composite outcome of perinatal death or serious infant morbidity from 25% to 17.5% (alpha 0.05; power 90%), as suggested by this pilot study.
A novel aspect of our trial design is the adoption of a strategy to identify elements of placental dysfunction using a combination of reinterpretation of maternal serum screening test data, placental morphology and uterine artery Doppler . An extensive interest in the concept of multi-parameter screening for placental dysfunction to date has been limited by focusing on just one element of adverse outcome, for example preeclampsia , and the fact that the blood tests used in risk modelling are performed in a batched retrospective manner and are thus not available for real-time clinical decision-making . We operated a real-time strategy by the integration of maternal serum screening data  with ultrasound imaging [12,16]. While maternal serum screening test abnormalities do not achieve sufficient precision to report risks of adverse perinatal outcomes in the care of low-risk women , when used in combination with two additional ultrasound tests (placental morphology and uterine artery Doppler) a subset of women with up to 50% positive predictive value for perinatal death and delivery before 34 weeks' gestation can be identified . Our current study validates this logic in a prospective manner because six infants, including all three perinatal deaths, were born before 32 weeks while only four women had healthy term pregnancies (delivery > 37 weeks with no evidence of IUGR or preeclampsia). By comparison, one perinatal death at 24 weeks amongst 110 women randomized to low-molecular-weight heparin was found in a similar trial based on clinical risk factors alone .
Our study is the first randomized controlled trial to report the effects of heparin on its target organ. The assumption in previous trials, that infarction is the dominant pathology mediating perinatal death and adverse outcomes, may be flawed because the majority of infarcted placentas exhibit other significant pathologies that limit placental function. These elements are being small for gestational age, with abnormal placental villous structure and non-infectious inflammatory lesions . We found one-third of the placentas were small for gestational age, consistent with previous data showing a 7-fold increased risk of a small placenta in clinically high-risk women with abnormal uterine artery Doppler . Placental infarction was detected in only 7/31 (22.5%) placentas, while lesions attributable to accidental hemorrhage (intervillous thrombosis, large retroplacental or subchorionic hemorrhage) were collectively more common and found mostly in the control arm. This observation is both surprising and reassuring because investigators are frequently concerned that heparin anticoagulation may increase the risk of hemorrhagic complications in pregnancy, especially abruption. Prenatal screening of placental function may aid future trials of heparin by defining the type of underlying placental insufficiency so as to focus on those at most risk of infarction. As an example, the subset of women with low PAPP-A and small abnormal placentas  are destined to develop severe IUGR rather than preeclampsia, due to defective formation of a gas and nutrient-exchanging placenta; such pregnancies may not respond to the anticoagulant action of heparin [45,46].
Two potential limitations of our study design deserve discussion. First, we chose UFH rather than LMWH. This decision was made on cost within the grant budget, to avoid pharmaceutical support. Both related trials that were larger [2,5,32] used LMWH and were not conducted independently of the drug manufacturer. Heparins have complex cellular actions that include regulating angiogenesis . UFH and LMWH differ in their in vitro angiogenic responses and in their survival properties for metastatic cancer patients . We recently found that LMWH interacts with first trimester placental villi to exert a stronger proangiogenic response than UFH in vitro . As UFH may be less potent than LMWH, we therefore chose a higher dose (7500 IU BD) to ensure efficacy as a prophylactic drug. Our observation of a higher rate of severe preeclampsia in the UFH arm of this pilot trial is surprising because UFH is predicted, from our in vitro data, to exert a partial proangiogenic response when it interacts with placental villi. Recently, Sela et al.  demonstrated that heparin promotes release of bound sFlt1 from the human placenta to increase circulating sFlt1 levels two-fold. As circulating maternal sFlt1 levels rise several weeks prior to the development of severe preeclampsia , we speculate that the higher rate of severe preeclampsia observed in our pilot study could be due to the selection of women with a high rate of preclinical placental dysfunction, in whom UFH may further elevate sFlt1 to exacerbate disease risk. An alternative explanation may relate to sample size because an excess of women allocated to the UFH arm had an abnormal biochemical test result due to elevated hCG or inhibin, which we more recently have recognized to be associated with severe preeclampsia from placental pathology . Given the small sample size, we accept that no specific conclusion can be drawn on the maternal benefits or risks of heparin, other than to underscore that our trial design, by virtue of its high negative predictive value , substantially reduces the proportion of women exposed to heparin in a trial context. In future studies, we suggest that serial determination of maternal circulating pro- and anti-angiogenic growth factors followed by detailed placental analysis will effectively test the hypothesis that heparin exerts negative effects upon high-risk pregnancies with placental dysfunction via augmented release of sFlt1. Given that LMWH exerts stronger proangiogenic effects in vitro  we accept that it will be important to use LMWH in future adequately-powered trials of this design to exploit this proangiogenic action of heparin in pregnancy.
A second potential limitation of our study was the ongoing use of low-dose aspirin in the standard care arm. For safety reasons, our trial design precluded the ongoing use of low-dose aspirin in the UFH treatment arm while a decision to take low-dose aspirin was allowed in the standard care arm to facilitate recruitment. A recent systematic review of the role of low-dose aspirin in the prevention of preeclampsia and/or IUGR in women with abnormal uterine artery Doppler ultrasound demonstrated that a reduction in preeclampsia was confined to women who started the drug in early pregnancy . An equal subset of women randomized to UFH, who therefore discontinued aspirin, had at least been exposed to aspirin from the first trimester until randomization. Given the low efficacy of aspirin in preventing preeclampsia following second trimester randomization , together with the placental pathology findings, demonstrating that placental infarction was not the dominant pathology, we believe that this difference in aspirin utilization does not undermine our general conclusions. Finally, the low rate of placental infarction in both arms is a valid counter-argument against the criticism that this trial design commenced heparin too late, or that it differed in the proportion of women who continued to take aspirin.
Given the cost and potential side effects of heparin during pregnancy, we believe the deployment of a strategy of ‘placental function screening’ to reduce trial entry to women with a high positive predictive value of extreme preterm delivery in association with placental developmental and vascular pathology, is an appropriate clinical strategy. This is based on a previously-validated screening program in a clinically high-risk cohort, where the positive predictive value for significant adverse perinatal outcomes was up to 50% . The high rate of serious outcomes in this pilot trial was further validated by placental pathologic analysis. Our research ethics board approval did not include consent to capture data for either screen negative women or eligible women who did not consent to randomization. We therefore estimated the number of screened women based on our prior cohort data  and thus could not derive the false-negative rate of adverse perinatal outcomes, and the overall effectiveness of this study design. Nevertheless our study design and findings, including placental pathology data, challenge the widely-prevailing view that pregnant women should receive prophylactic heparin to improve perinatal outcomes based solely on clinical risk factors for placental insufficiency.