Patients with amniotic fluid embolism (AFE) (major cardiac and pulmonary symptoms plus consumptive coagulopathy) have high circulating tissue factor concentrations. Recombinant factor VIIa (rVIIa) has been used to treat hemorrhage in AFE patients even though rVIIa can combine with circulating tissue factor and form intravascular clots. A systematic review was done of case reports from 2003 to 2009 of AFE patients with massive hemorrhage who were and were not treated with rVIIa to assess the thrombotic complication risk.
MEDLINE was searched for case reports of AFE patients receiving rVIIa (rVIIa cases) and of AFE patients who received surgery to control bleeding but no rVIIa (cohorts who did not receive rVIIa). Additional AFE case reports were obtained from the Food and Drug Administration, the Australian and New Zealand Haemostasis Registry, and scientific meeting abstracts. The risk of a negative outcome (permanent disability or death) in rVIIa cases versus cohorts who did not receive rVIIa was calculated using risk ratio and 95% confidence interval.
Sixteen rVIIa cases and 28 cohorts were identified who did not receive rVIIa. All patients had surgery to control bleeding. Death, permanent disability, and full recovery occurred in 8, 6, and 2 rVIIa cases and 7, 4, and 17 cohorts who did not receive rVIIa (risk ratio 2.2, 95% CI 1.4-3.7 for death or permanent disability vs. full recovery).
Recombinant factor VIIa cases had significantly worse outcomes than cohorts who did not receive rVIIa. It is recommended that rVIIa be used in AFE patients only when the hemorrhage cannot be stopped by massive blood component replacement.
What We Already Know about This Topic
The role of recombinant factor VIIa (rVIIa) in the treatment of amniotic fluid embolism, which can be complicated by hemorrhage, is unclear
What This Article Tells Us That Is New
In a review of case reports of amniotic fluid embolism from 2003 to 2009, the 16 patients who received rVIIa had significantly worse outcomes (major organ thrombosis and death) compared with 28 cohorts who did not receive this therapy
CONSUMPTIVE coagulopathy and major cardiac and pulmonary symptoms are the hallmarks of the syndrome of amniotic fluid embolism (AFE).1Because AFE is a rare condition that does not naturally occur in nonhuman animals, case reports are a major source of information about this syndrome. When caring for patients with massive postpartum hemorrhage, some anesthesiologists have administered procoagulant drugs, including recombinant factor VIIa (rVIIa) (NovoSeven®; Novo Nordisk A/S, Bagsvaerd, Denmark), aprotinin, tranexamic acid, and fibrinogen concentrate, in addition to blood components. This therapy generally works well because most postpartum hemorrhage is caused by uterine atony or birth trauma, conditions not associated with increased circulating tissue factor concentrations. However, a recent review also recommends the use of rVIIa to treat hemorrhage from intractable coagulopathy in patients with AFE, a disease that is associated with increased circulating tissue factor concentrations.2
Factor VIIa initiates the extrinsic coagulation pathway by combining with tissue factor.3Tissue factor is normally present in very low concentrations in the bloodstream. A cut in a blood vessel exposes tissue factor to circulating factor VIIa, which starts a cascade resulting in fibrin deposition and the reestablishment of vascular integrity.3In theory, the administration of rVIIa to patients with high circulating concentrations of tissue factor could lead to widespread inappropriate fibrin deposition in vital organs, including the brain, heart, kidneys, and liver. Circulating tissue factor concentrations are high in patients with disseminated intravascular coagulopathy (DIC), including patients with DIC caused by AFE.3The package insert for NovoSeven® warns that “patients with DIC … have an increased risk of developing thrombotic events ….”4
To assess the risk of thrombotic complications, we systematically reviewed case reports from 2003 to 2009 of patients with massive hemorrhage attributed to AFE who were and were not treated with rVIIa.
Materials and Methods
The Washington University in Saint Louis Human Research Protection Office approved this study. We searched MEDLINE for case reports describing the use of rVIIa in patients with AFE (rVIIa cases). We retrieved rVIIa cases using the following medical subject heading terms: “amniotic fluid embolism” or “anaphylactoid syndrome of pregnancy” combined with “recombinant factor VIIa,”“VIIa,” or “NovoSeven.” After determining the time period during which these cases were reported, we researched MEDLINE using the terms “amniotic fluid embolism” or “anaphylactoid syndrome of pregnancy” for AFE cases during that time period in which surgery was used to control bleeding and in which there was no use of rVIIa (cohorts who did not receive rVIIa). Surgery is used to control postpartum hemorrhage only when the bleeding cannot be stopped by standard medical means; we added this requirement to restrict the cohorts who did not receive rVIIa to patients with massive hemorrhage. We hand-searched the abstracts of the Society for Obstetric Anesthesia and Perinatology and the American Society of Anesthesiologists from 1999 to 2009 for additional rVIIa cases and for cohorts who did not receive rVIIa. We filed a Freedom of Information Act request with the Food and Drug Administration to obtain information on adverse drug effects with rVIIa. We requested information from the Australian and New Zealand Haemostasis Registry, the Northern Europe Factor 7a in Obstetric Hemorrhage Registry, and the American Society of Anesthesiologists Closed Claims Project on rVIIa use in AFE patients.5,–,7We contacted the authors of articles and abstracts for more information as needed. Non-English case reports were translated by anesthesiologists who are native speakers of the non-English languages. Demographic and clinical course details were scrutinized to avoid duplicate reporting of patients.
We considered the diagnosis of AFE confirmed if patients had at least one cardiac (hypotension or cardiac arrest) and one pulmonary (dyspnea, cyanosis, hypoxemia, or respiratory arrest) symptom in addition to DIC. This is the same definition of AFE used by the AFE registries in the United States and the United Kingdom.1,8Patients were not included if the diagnosis of AFE could not be confirmed. We defined massive transfusion as the transfusion of ≥10 units of packed red blood cells (PRBC). We obtained the fresh frozen plasma (FFP):PRBC ratio by dividing the number of FFP units received by the number of PRBC units received. We classified patients as having had minor surgery if the maximum intervention for bleeding (dilation and curettage, B-Lynch suture placement, intrauterine balloon catheter placement) did not include exploratory laparotomy, hysterectomy, or uterine artery embolization or ligation.
The risk of a negative outcome (permanent disability or death) in rVIIa cases versus cohorts who did not receive rVIIa was calculated using risk ratio and 95% CI. To determine whether rVIIa cases and cohorts who did not receive rVIIa differed, we compared age (normally distributed) using independent samples Student t test, the number of PRBC units and the FFP:PRBC ratio (not normally distributed) using the Mann–Whitney U test, and surgery type (minor vs. not) and transfusion requirement (massive transfusion vs. not) using Fisher exact test. P < 0.05 was considered significant. SAS version 9.2 (SAS Institute, Cary, NC) was used for analyses.
Our first MEDLINE search yielded five rVIIa cases.9,–,13One rVIIa case received both rVIIa and aprotinin.13Two patients mentioned in several review articles as having had successful outcomes after receiving rVIIa for AFE did not have AFE according to the strict diagnostic criteria we used and are not included in this review.14,15Three additional rVIIa cases were described in Society for Obstetric Anesthesia and Perinatology abstracts.16,–,18Four rVIIa cases were obtained from the Food and Drug Administration (unpublished data, Food and Drug Administration file #2009–4281, October 13, 2009); however, one of the Food and Drug Administration rVIIa cases also was described in a published case report.10Five rVIIa cases were obtained from the Australian and New Zealand Haemostasis Registry.19No additional rVIIa cases were obtained from the Northern Europe Factor 7a in Obstetric Hemorrhage Registry or the American Society of Anesthesiologists Closed Claims Project. Surgery to control bleeding was performed in all 16 rVIIa cases: hysterectomy (n = 7), hysterectomy plus a subsequent exploratory laparotomy (n = 2), hysterectomy plus right salpingo-oophrectomy (n = 1), exploratory laparotomy (n = 3), exploratory laparotomy plus additional surgery (n = 2), and intrauterine balloon catheter insertion (n = 1) (table 1). All case reports for retrieved rVIIa cases were published, presented, or registered in English or German between May 1, 2003 and May 3, 2009.
Our second MEDLINE search identified 181 articles discussing AFE or anaphylactoid syndrome of pregnancy between 2003 and 2009. These articles contained details of 78 patients identified by the authors as having AFE. Five patients, described previously, met our definition of AFE and received rVIIa. Forty-seven patients either did not meet our definition of AFE (n = 35) or did not have surgery to control bleeding (n = 12). Twenty-six patients (cohorts who did not receive rVIIa) met our definition of AFE, underwent a surgical procedure to control bleeding, and did not receive rVIIa.20,–,43Five of these patients received fibrinogen concentrate,20,27,30,31,41two patients received tranexamic acid,26,41two patients received aprotinin,27,29and one patient received prothrombin complex concentrate.41Information on two additional cohorts who did not receive rVIIa was available in Society for Obstetric Anesthesia and Perinatology abstracts.44,45The surgical procedures in these 28 control subjects were hysterectomy (n = 12), hysterectomy plus a subsequent exploratory laparotomy (n = 2), uterine artery embolization or ligation (n = 5), exploratory laparotomy (n = 5), B-Lynch suture placement during cesarean delivery (n = 1), and dilation and curettage (n = 3) (table 2). Case reports on all retrieved cohorts who did not receive rVIIa were published or presented in English or French between August 11, 2003 and September 7, 2009.
All of the patients in both groups had an intense DIC and bled profusely from the uterus, the bladder, the mouth, the incision, and venipuncture sites. All patients received pressors to treat hypotension or cardiac arrest, although it is not possible from the case reports to determine a complete list of the pressors used or the doses administered. Heroic efforts were used to obtain sufficient blood products to treat one patient; the US Navy diverted two ships for blood donation during unscheduled port visits for a cohort who did not receive rVIIa: the 20-yr-old wife of a Navy sailor.38
Approximately 88% (14 of 16) of rVIIa cases had a negative outcome (permanent disability or death) compared with 39% (11 of 28) of cohorts who did not receive rVIIa (risk ratio 2.2, 95% CI 1.4–3.7) (table 3). Among those who lived, 75% (6 of 8) of rVIIa cases had permanent disability (coma, stroke × 2, memory loss, pulmonary hypertension, new systemic hypertension) compared with 19% (4 of 21) of cohorts who did not receive rVIIa (left-sided motor weakness, short term memory loss × 2, panhypopituitarism) (risk ratio 4.0, 95% CI 1.5–10.4).
Recombinant factor VIIa cases and cohorts who did not receive rVIIa did not differ significantly by age (rVIIa cases: mean 35.0 yr, 95% CI 31.4–38.5; cohorts who did not receive rVIIa: mean 32.9 yr, 95% CI 30.7–35.0; P = 0.27), whether or not the surgery was minor (rVIIa cases 1 of 16 (6%), cohorts who did not receive rVIIa 4 of 28 (14%), P = 0.64), whether or not the patient received a massive transfusion (rVIIa cases 13 of 14 (93%), cohorts who did not receive rVIIa 15 of 22 (68%), P = 0.12), the number of PRBC units transfused (rVIIa cases: median 16, minimum 3, maximum 66; cohorts who did not receive rVIIa: median 11.5, minimum 2, maximum 80; P = 0.22), or the FFP:PRBC ratio (rVIIa cases: median 0.67, minimum 0.13, maximum 4.33; cohorts who did not receive rVIIa: median 0.91,minimum 0.30, maximum 3.00; P = 0.14).
Patients with a negative outcome (permanent disability or death) versus full recovery did not differ significantly by age (negative outcome mean 34.3 yr, 95% CI 31.6–37.0; full recovery mean 32.7 yr, 95% CI 30.3–35.2; P = 0.40), whether the surgery was minor (negative outcome 1/25 (4%), full recovery 4 of 19 (21%), P = 0.15), whether they received a massive transfusion (negative outcome 16 of 19 (84%), full recovery 12 of 17 (71%), P = 0.43), the number of PRBC units transfused (negative outcome median 16, minimum 3, maximum 80; full recovery median 13, minimum 2, maximum 54; P = 0.23) or the FFP:PRBC ratio (negative outcome median 0.74, minimum 0.13, maximum 4.33; full recovery median 0.92, minimum 0.30, maximum 3.00; P = 0.25).
Thrombosis in major organs is the most serious complication of rVIIa administration. Our review suggests that clinically significant thrombosis is associated with rVIIa treatment in AFE patients with DIC. Patients who received rVIIa had significantly worse outcomes than cohorts who did not receive rVIIa. (risk ratio 2.2, 95% CI 1.4–3.7).
The medical management of coagulopathy-induced hemorrhage is an unresolved clinical problem. rVIIa, which activates the extrinsic coagulation pathway by combining with circulating tissue factor, is the drug most likely to trigger inappropriate fibrin deposition.3Thrombosis also can occur with tranexamic acid or aprotinin, which inhibit fibrinolysis.46Among the five patients in this review treated with aprotinin and/or tranexamic acid, multiple organ failure and death occurred in two patients: one who also received rVIIa and one who did not.13,26,27,29,41Anticoagulation might prevent the initiation of consumptive coagulopathy but is also likely to worsen an already massive hemorrhage. An agent that may prove useful in the future is fibrinogen concentrate, which recently became commercially available in the United States. Four of the five patients treated with fibrinogen concentrate, all cohorts who did not receive rVIIa from Europe, made full recoveries, whereas the fifth patient, who was also treated with tranexemic acid, died.20,27,30,31,41
Recombinant factor VIIa has been used successfully to treat postpartum hemorrhage in patients who do not have high circulating tissue factor concentrations, such as those with uterine atony, uterine rupture, and abnormal placentation.6,7,19,47,48Patients with preeclampsia or preterm prelabor rupture of membranes have moderately high circulating tissue factor concentrations; however, patients with the HELLP variant of preeclampsia (Hemolysis, Elevated Liver enzymes, and Low Platelet count) have been treated successfully with rVIIa.19,48,49,50
Coagulopathy can be both the cause of massive hemorrhage in AFE and the result of insufficient coagulation factor replacement during massive transfusion.51Trauma patients who have had massive tranfusion frequently develop an acute coagulopathy because of tissue trauma, shock, hemodilution, hypothermia, acidemia, and inflammation. Studies in trauma patients have shown that replacement of PRBC, FFP, and platelets in a ratio of 1:1:1 usually provides adequate concentrations of all clotting factors and fibrinogen. However, if fibrinogen is less than 1 g/l, cryoprecipitate or fibrinogen concentrate should be given.51,52Correcting the coagulopathy and stopping the hemorrhage as soon as possible are important to avoid the potentially lethal complications of acidosis and hypothermia.
Amniotic fluid, the amniochorion, and the placenta are rich sources of both procoagulant and anticoagulant factors. Surfactant, a lipoprotein produced by fetal lungs and present in increasing amounts in amniotic fluid with increasing gestational age, is structurally similar to tissue thromboplastin and possesses significant thromboplastic activity.53There is a close correlation between amniotic fluid surfactant concentration and the ability of amniotic fluid to shorten plasma recalcification time (r = 0.77).53,54Amniotic fluid and the amniochorion contain a cysteine protease that directly activates factor X.55Cells in amniotic fluid externalize phosphatidylserine, which promotes the activation of factor X and prothrombin.56Amniotic fluid cells also externalize tissue factor, the most potent naturally occurring procoagulant.56Amniotic fluid irreversibly aggregates platelets.54The term placenta contains high concentrations of anticoagulants, including tissue factor pathway inhibitor-2 (a strong inhibitor of activated factor X), thrombomodulin, and annexin V.57Many of these factors are essential for the maintenance of pregnancy. Transgenic mice with low tissue factor concentrations have a 42% incidence of fatal midgestational maternal hemorrhage.57Thrombomodulin knockout mice resorb all embryonic tissue in midgestation.57Maternal and fetal survival require sequestration and local application of powerful procoagulant and anticoagulant factors; loss of the coagulation balance or the sequestration can be fatal.
The mortality rates we observed in rVIIa cases (50%) and cohorts who did not receive rVIIa (25%) are consistent with other reports of AFE mortality. The United Kingdom Amniotic Fluid Embolism Register reported a mortality rate of 29.5% from 1997 to 2004.8A population-based study in the United States from 1998 to 2003 found a case fatality rate of 21.6%.58Previous reports found mortality rates of 40–80%.1Although AFE is a rare disease, it is currently the second leading cause of direct maternal death in the United States.59
Because this is a retrospective study, we cannot determine whether rVIIa caused or was merely associated with poor patient outcome. However, the low incidence of AFE (7.7 per 100,000 births in the United States) makes a prospective, randomized trial of AFE therapy impossible.58The patients in this review were described in case reports, presented at conferences, or entered into registries of complications. Although none of the demographic differences between rVIIa cases and cohorts who did not receive rVIIa were significant, rVIIa patients were 2.1 yr older, had a slightly lower FFP:PRBC ratio (0.67 vs. 0.91), and were slightly more likely to have major surgery (94%vs. 86%) and have massive transfusion (93%vs. 68%). Sicker patients, including patients in whom DIC could not be controlled with blood component therapy, may have been selected to receive rVIIa. It is possible that the rVIIa patients may have required more pressor support than the cohorts who did not receive rVIIa. There may have been underreporting of patients with successful results or of maternal deaths in either the rVIIa patients or the cohorts who did not receive rVIIa. We used rigid criteria for the diagnosis of AFE and systematic search criteria to minimize these potential sources of error.
In conclusion, the use of rVIIa to treat postpartum hemorrhage in patients with severe consumptive coagulopathy and AFE was associated with worse outcome compared with cohorts who did not receive rVIIa. Most of the survivors treated with rVIIa suffered permanent damage that could have been caused by organ thrombosis. We recommend that the initial therapy of AFE-associated consumptive coagulopathy should consist of blood component replacement, including PRBC, FFP, platelets, cryoprecipitate, and possibly fibrinogen concentrate. We recommend that rVIIa be used in AFE patients only when the hemorrhage cannot be stopped by massive blood component replacement.