Study design, setting, and ethics statement

This was a multicenter open-label randomized controlled trial (ACTRN12609000699268) at 5 tertiary care centers in New Zealand, Australia, and The Netherlands. Appropriate ethics and governance approvals were obtained at each center. All women participating in the trial provided written informed consent.

Participants

Through the duration of the trial women referred for antenatal care were screened for eligibility. Women were eligible for inclusion if they were >6 ⁺⁰ and <16 ⁺⁰ weeks; gestation with a viable singleton pregnancy confirmed by ultrasound scan and at risk of preeclampsia and/or IUGR based on their obstetric history with: (1) previous preeclampsia delivered <36 ⁺⁰ weeks in their last ongoing pregnancy reaching >12 weeks; or (2) previous SGA infant <10th customized birthweight percentile delivered <36 ⁺⁰ weeks in their last ongoing pregnancy reaching >12 weeks with no major fetal anomaly; or (3) previous SGA infant ≤3rd customized birthweight percentile delivered at any gestation in their last ongoing pregnancy reaching >12 weeks with no major fetal anomaly. Eligibility criteria were checked against medical records. Women were excluded from the trial if they met ≥1 of the exclusion criteria: any contraindication to LMWH use; need for anticoagulant use in pregnancy such as previous thrombosis or antiphospholipid syndrome; previous successful pregnancy with LMWH treatment; multiple pregnancy; known preexisting type 1 or 2 diabetes; renal disease (with serum creatinine >150 umol/L); thrombocytopenia (platelet count <80 × 10 ⁹ /L); or known major fetal anomaly/chromosomal abnormality.

We included women with previous preeclampsia and/or SGA as these diseases are both placental in origin with considerable overlap in pathological mechanisms. LMWH is likely to exert any therapeutic benefit via effect(s) on placentation and therefore has potential to impact both diseases, which often coexist, particularly when disease is preterm.

Randomization and interventions

After confirming eligibility and obtaining consent participants were randomly assigned in a 1-to-1 ratio to standard high-risk care or standard high-risk care plus enoxaparin 40 mg (4000 IU) by subcutaneous injection (Clexane, Sanofi-Aventis, Auckland, New Zealand) daily from recruitment until 36 ⁺⁰ weeks or delivery, whichever occurred sooner. A 40-mg dose of enoxaparin was selected as the standard dose used for venous thromboembolism prophylaxis and, as per manufacturer’s direction, no adjustment was made for body mass index. Standard high-risk care was defined as care coordinated by a high-risk antenatal clinic service, aspirin 100 mg daily until 36 ⁺⁰ weeks, and–for women with prior preeclampsia–calcium 1000-1500 mg daily until 36 ⁺⁰ weeks. This was an open-label trial with all participants, clinicians, and investigators aware of trial group assignment.

A computer-generated randomization program balanced in blocks of 5 was used with stratification for recruiting site and thrombophilia status. Participants who were tested prior to trial involvement were assigned to: (1) positive thrombophilia or (2) negative thrombophilia. Women not tested or partially tested (with negative result) were assigned: (3) unknown thrombophilia status. Samples were then taken prior to any use of LMWH and tested for lupus anticoagulant, anticardiolipin antibodies, antithrombin III deficiency, protein C deficiency, protein S deficiency, factor V Leiden, and the prothrombin gene mutation. These results were used in the final analysis but not revealed to participant, clinicians, or investigators during the trial period.

All participants were assigned a sequential trial identifying number according to thrombophilia status. At the lead recruiting site (National Women’s Health, Auckland City Hospital) randomization occurred by telephone to the hospital pharmacy clinical trials service and trial group allocation was made by study identifying number. In all other sites sequential sealed opaque envelopes for each study identifying number were opened to reveal treatment allocation.

Procedures

For women receiving LMWH, an educational session regarding injection technique was provided by a research midwife prior to treatment commencement. LMWH was resupplied on a monthly basis from hospital pharmacies. Participants were asked to return sharps disposal boxes with used syringes and any unused treatment. Indications to stop treatment <36 ⁺⁰ weeks included: need for delivery <36 ⁺⁰ weeks (stopped once decision made for delivery or 12 hours prior to induction of labor or elective cesarean delivery, whichever was later), episode of threatened preterm labor (treatment recommenced if symptoms settled), and clinical evidence of placental abruption or thrombocytopenia with platelet count <80 × 10 ⁹ /L.

Serum samples were taken at recruitment (used as baseline) and at 20 and 30 weeks’ gestation in a subgroup of participants (n = 127). Samples were centrifuged and stored at –80°C for later assessment of sFlt-1, sEng, PlGF, sVCAM-1, and ET-1 by enzyme-linked immunosorbent assay according to manufacturer’s instructions (R&D Systems, Minneapolis, MN). Optical density was determined using X-Mark microplate spectrophotometer (BioRad, Gladesville, Australia) and analyte levels were determined using Microplate manager 6 software (BioRad). Analytes were measured in 1 batch in a blinded fashion with the cases and controls mixed.

Standard fetal growth parameters and uterine and umbilical arterial Doppler waveforms were recorded at 20 and 24 weeks’ gestation. Data from any additional antenatal ultrasound scans were collected. All participants received care through a high-risk antenatal clinic service including daily aspirin and, for women with prior preeclampsia, daily calcium. Care included regular blood pressure (BP) assessment, urine dipstick analysis for proteinuria, fetal well-being assessment, and review of any adverse events. Intrapartum and postnatal care was provided by local clinicians. Pregnancy, labor, postnatal, and neonatal data were collected by research staff from maternal and infant clinical records up to the time of hospital discharge.

Outcomes

The primary outcome was a composite of preeclampsia and/or SGA <5th percentile. All cases with primary outcome events were reviewed by trial investigators. Secondary outcomes included preeclampsia, severe preeclampsia, HELLP syndrome (hemolysis, elevated liver enzyme, and low platelet count), eclampsia, gestational hypertension, SGA <10th percentile, SGA <5th percentile, SGA <3rd percentile, placental abruption and antepartum hemorrhage, thrombocytopenia (platelet count <100 × 10 ⁹ /L) while on LMWH, stillbirth, induction of labor, cesarean delivery, postpartum hemorrhage, gestational age at birth, preterm birth, mean birthweight and mean birthweight percentile, neonatal death, neonatal intensive care admission and duration of admission, a composite outcome of severe neonatal morbidity (evidence of ≥1: intraventricular hemorrhage [grade 3 and 4], cystic periventricular leukomalacia, chronic lung disease, retinopathy of prematurity requiring treatment, or necrotizing enterocolitis requiring surgery), and maternal serum levels of sFlt-1, sEng, PlGF, sVCAM-1, and ET-1.

Preeclampsia was defined as new-onset hypertension (systolic BP ≥140 mm Hg and/or diastolic BP ≥90 mm Hg) >20 weeks’ gestation with evidence of significant proteinuria or any multisystem complication including hematological, liver, renal, and neurological involvement. Severe preeclampsia included preeclampsia associated with maternal death, persistent severe hypertension (systolic BP ≥170 mm Hg and/or diastolic BP ≥110 mm Hg), or a multisystem complication. Customized birthweight percentiles adjusting for infant sex, gestation at delivery, and maternal variables (parity, ethnicity, height, and early pregnancy weight) were used to identify SGA infants. Nonpregnant- and pregnancy-specific (as appropriate to timing of testing) reference ranges were used to define protein C deficiency, protein S deficiency, and antithrombin III deficiency.

Statistical analysis

A priori we estimated a risk of recurrence of preeclampsia and/or SGA <5th percentile for the included population of 25%. Based on published data we estimated the risk of recurrence would be reduced to 7% with the addition of LMWH therapy. We planned for a sample size of 160 participants (80 participants in each group, including a 5% dropout/early miscarriage rate) to achieve 80% power at a 2-sided significance level of .05 to detect a difference between 25-7%.

An independent data monitoring committee undertook a planned interim analysis once 98 participants were recruited and completed follow-up. Safety and primary outcome data were assessed and reported to the trial investigator group. Early discontinuation of the trial was to be considered if there was an excess of serious adverse events in the LMWH group or if a significant benefit had already been demonstrated (49 participants in each group to achieve 80% power at a 2-sided significance level of .05 to detect a difference between 25-4%).

All data were analyzed on an intention-to-treat basis. Continuous data are reported as median with minimum–maximum values (range). Categorical data are reported as number and proportion. Unless otherwise stated, statistical models were adjusted for thrombophilia status, recruitment center, inclusion criteria, and number of previous preeclampsia/SGA events (1 vs ≥2) as these variables were deemed the most likely to potentially influence the risk of recurrence and/or the care provided. Multiple logistic regression was used to analyze the binary primary and secondary outcomes. Multiple linear regression was used to analyze the continuous secondary outcomes. No missing data imputation was conducted. No multiple comparison adjustment was made. Two-sided P values <.05 were used to determine statistical significance and all confidence intervals (CI) are given at a 2-sided 95% level. Statistical analysis was conducted using software (SAS for Windows 9.4; SAS Institute Inc, Cary, NC). Levels of analytes measured in the serum samples were tested for normal distribution and statistically tested using nonparametric Mann-Whitney test. Data were expressed as mean ± SEM. Biomarker statistical analysis was performed using software (GraphPad Prism 6; GraphPad Software, La Jolla, CA).

Study design, setting, and ethics statement

This was a multicenter open-label randomized controlled trial (ACTRN12609000699268) at 5 tertiary care centers in New Zealand, Australia, and The Netherlands. Appropriate ethics and governance approvals were obtained at each center. All women participating in the trial provided written informed consent.

Participants

Through the duration of the trial women referred for antenatal care were screened for eligibility. Women were eligible for inclusion if they were >6 ⁺⁰ and <16 ⁺⁰ weeks; gestation with a viable singleton pregnancy confirmed by ultrasound scan and at risk of preeclampsia and/or IUGR based on their obstetric history with: (1) previous preeclampsia delivered <36 ⁺⁰ weeks in their last ongoing pregnancy reaching >12 weeks; or (2) previous SGA infant <10th customized birthweight percentile delivered <36 ⁺⁰ weeks in their last ongoing pregnancy reaching >12 weeks with no major fetal anomaly; or (3) previous SGA infant ≤3rd customized birthweight percentile delivered at any gestation in their last ongoing pregnancy reaching >12 weeks with no major fetal anomaly. Eligibility criteria were checked against medical records. Women were excluded from the trial if they met ≥1 of the exclusion criteria: any contraindication to LMWH use; need for anticoagulant use in pregnancy such as previous thrombosis or antiphospholipid syndrome; previous successful pregnancy with LMWH treatment; multiple pregnancy; known preexisting type 1 or 2 diabetes; renal disease (with serum creatinine >150 umol/L); thrombocytopenia (platelet count <80 × 10 ⁹ /L); or known major fetal anomaly/chromosomal abnormality.

We included women with previous preeclampsia and/or SGA as these diseases are both placental in origin with considerable overlap in pathological mechanisms. LMWH is likely to exert any therapeutic benefit via effect(s) on placentation and therefore has potential to impact both diseases, which often coexist, particularly when disease is preterm.

Randomization and interventions

After confirming eligibility and obtaining consent participants were randomly assigned in a 1-to-1 ratio to standard high-risk care or standard high-risk care plus enoxaparin 40 mg (4000 IU) by subcutaneous injection (Clexane, Sanofi-Aventis, Auckland, New Zealand) daily from recruitment until 36 ⁺⁰ weeks or delivery, whichever occurred sooner. A 40-mg dose of enoxaparin was selected as the standard dose used for venous thromboembolism prophylaxis and, as per manufacturer’s direction, no adjustment was made for body mass index. Standard high-risk care was defined as care coordinated by a high-risk antenatal clinic service, aspirin 100 mg daily until 36 ⁺⁰ weeks, and–for women with prior preeclampsia–calcium 1000-1500 mg daily until 36 ⁺⁰ weeks. This was an open-label trial with all participants, clinicians, and investigators aware of trial group assignment.

A computer-generated randomization program balanced in blocks of 5 was used with stratification for recruiting site and thrombophilia status. Participants who were tested prior to trial involvement were assigned to: (1) positive thrombophilia or (2) negative thrombophilia. Women not tested or partially tested (with negative result) were assigned: (3) unknown thrombophilia status. Samples were then taken prior to any use of LMWH and tested for lupus anticoagulant, anticardiolipin antibodies, antithrombin III deficiency, protein C deficiency, protein S deficiency, factor V Leiden, and the prothrombin gene mutation. These results were used in the final analysis but not revealed to participant, clinicians, or investigators during the trial period.

All participants were assigned a sequential trial identifying number according to thrombophilia status. At the lead recruiting site (National Women’s Health, Auckland City Hospital) randomization occurred by telephone to the hospital pharmacy clinical trials service and trial group allocation was made by study identifying number. In all other sites sequential sealed opaque envelopes for each study identifying number were opened to reveal treatment allocation.

Procedures

For women receiving LMWH, an educational session regarding injection technique was provided by a research midwife prior to treatment commencement. LMWH was resupplied on a monthly basis from hospital pharmacies. Participants were asked to return sharps disposal boxes with used syringes and any unused treatment. Indications to stop treatment <36 ⁺⁰ weeks included: need for delivery <36 ⁺⁰ weeks (stopped once decision made for delivery or 12 hours prior to induction of labor or elective cesarean delivery, whichever was later), episode of threatened preterm labor (treatment recommenced if symptoms settled), and clinical evidence of placental abruption or thrombocytopenia with platelet count <80 × 10 ⁹ /L.

Serum samples were taken at recruitment (used as baseline) and at 20 and 30 weeks’ gestation in a subgroup of participants (n = 127). Samples were centrifuged and stored at –80°C for later assessment of sFlt-1, sEng, PlGF, sVCAM-1, and ET-1 by enzyme-linked immunosorbent assay according to manufacturer’s instructions (R&D Systems, Minneapolis, MN). Optical density was determined using X-Mark microplate spectrophotometer (BioRad, Gladesville, Australia) and analyte levels were determined using Microplate manager 6 software (BioRad). Analytes were measured in 1 batch in a blinded fashion with the cases and controls mixed.

Standard fetal growth parameters and uterine and umbilical arterial Doppler waveforms were recorded at 20 and 24 weeks’ gestation. Data from any additional antenatal ultrasound scans were collected. All participants received care through a high-risk antenatal clinic service including daily aspirin and, for women with prior preeclampsia, daily calcium. Care included regular blood pressure (BP) assessment, urine dipstick analysis for proteinuria, fetal well-being assessment, and review of any adverse events. Intrapartum and postnatal care was provided by local clinicians. Pregnancy, labor, postnatal, and neonatal data were collected by research staff from maternal and infant clinical records up to the time of hospital discharge.

Outcomes

The primary outcome was a composite of preeclampsia and/or SGA <5th percentile. All cases with primary outcome events were reviewed by trial investigators. Secondary outcomes included preeclampsia, severe preeclampsia, HELLP syndrome (hemolysis, elevated liver enzyme, and low platelet count), eclampsia, gestational hypertension, SGA <10th percentile, SGA <5th percentile, SGA <3rd percentile, placental abruption and antepartum hemorrhage, thrombocytopenia (platelet count <100 × 10 ⁹ /L) while on LMWH, stillbirth, induction of labor, cesarean delivery, postpartum hemorrhage, gestational age at birth, preterm birth, mean birthweight and mean birthweight percentile, neonatal death, neonatal intensive care admission and duration of admission, a composite outcome of severe neonatal morbidity (evidence of ≥1: intraventricular hemorrhage [grade 3 and 4], cystic periventricular leukomalacia, chronic lung disease, retinopathy of prematurity requiring treatment, or necrotizing enterocolitis requiring surgery), and maternal serum levels of sFlt-1, sEng, PlGF, sVCAM-1, and ET-1.

Preeclampsia was defined as new-onset hypertension (systolic BP ≥140 mm Hg and/or diastolic BP ≥90 mm Hg) >20 weeks’ gestation with evidence of significant proteinuria or any multisystem complication including hematological, liver, renal, and neurological involvement. Severe preeclampsia included preeclampsia associated with maternal death, persistent severe hypertension (systolic BP ≥170 mm Hg and/or diastolic BP ≥110 mm Hg), or a multisystem complication. Customized birthweight percentiles adjusting for infant sex, gestation at delivery, and maternal variables (parity, ethnicity, height, and early pregnancy weight) were used to identify SGA infants. Nonpregnant- and pregnancy-specific (as appropriate to timing of testing) reference ranges were used to define protein C deficiency, protein S deficiency, and antithrombin III deficiency.

Statistical analysis

A priori we estimated a risk of recurrence of preeclampsia and/or SGA <5th percentile for the included population of 25%. Based on published data we estimated the risk of recurrence would be reduced to 7% with the addition of LMWH therapy. We planned for a sample size of 160 participants (80 participants in each group, including a 5% dropout/early miscarriage rate) to achieve 80% power at a 2-sided significance level of .05 to detect a difference between 25-7%.

An independent data monitoring committee undertook a planned interim analysis once 98 participants were recruited and completed follow-up. Safety and primary outcome data were assessed and reported to the trial investigator group. Early discontinuation of the trial was to be considered if there was an excess of serious adverse events in the LMWH group or if a significant benefit had already been demonstrated (49 participants in each group to achieve 80% power at a 2-sided significance level of .05 to detect a difference between 25-4%).

All data were analyzed on an intention-to-treat basis. Continuous data are reported as median with minimum–maximum values (range). Categorical data are reported as number and proportion. Unless otherwise stated, statistical models were adjusted for thrombophilia status, recruitment center, inclusion criteria, and number of previous preeclampsia/SGA events (1 vs ≥2) as these variables were deemed the most likely to potentially influence the risk of recurrence and/or the care provided. Multiple logistic regression was used to analyze the binary primary and secondary outcomes. Multiple linear regression was used to analyze the continuous secondary outcomes. No missing data imputation was conducted. No multiple comparison adjustment was made. Two-sided P values <.05 were used to determine statistical significance and all confidence intervals (CI) are given at a 2-sided 95% level. Statistical analysis was conducted using software (SAS for Windows 9.4; SAS Institute Inc, Cary, NC). Levels of analytes measured in the serum samples were tested for normal distribution and statistically tested using nonparametric Mann-Whitney test. Data were expressed as mean ± SEM. Biomarker statistical analysis was performed using software (GraphPad Prism 6; GraphPad Software, La Jolla, CA).