Amniotic fluid embolism is the second leading cause of maternal death in the United States, at 100–150 deaths per year.1 The medical literature gives a broad range of the case fatality rate with some studies suggesting that over half of mothers died, whereas recent data from the United States showed a fatality rate as low as 26%.2,3 There is similar variation in assessment fetal prognoses. The disease is devastating, underscored by the fact that it was first recognized by fetal material found in maternal pulmonary vasculature at autopsy.4 The diagnosis of amniotic fluid embolism is made clinically in pregnant women who suddenly become extremely ill without obvious explanations. To date, animal models have not elucidated the basic mechanism of this disease.5,6
With an interest in its pathophysiology, we wished to test two hypotheses that involved maternal immune response to fetal antigen. The first theory was that clinical symptoms resulted from mast cell degranulation and release of histamine (the mechanism that underlies anaphylaxis).5,7,8 The alternative was that illness resulted from massive activation of the complement pathway. Although not a main objective of our study, we also sought to measure a specific fetal antigen, sialyl Tn, which has been proposed as a diagnostic test for amniotic fluid embolism.9,10 That provided a semiquantitative method to measure fetal antigen leakage into the maternal circulation and some limited, additional data on the utility of sialyl Tn in diagnosing amniotic fluid embolism.
Materials and Methods
We used the following three criteria for diagnosis and inclusion: currently pregnant or within 48 hours of delivery; one or more of hypotension (or cardiac arrest), respiratory distress, or disseminated intravascular coagulation severe enough to require medical treatment; and absence of other medical explanations for clinical course. Some women with amniotic fluid embolism might have secondary conditions, but the emphasis is on whether other medical problems seem most likely to cause the patients' sudden collapse.
The seven Japanese women were recruited from institutions all over Japan and came to the authors' attention chiefly through their 10-year involvement in a national maternal mortality surveillance project. Autopsies found fetal material in maternal pulmonary vasculature. The two American women were recruited over 2 years at a single Illinois hospital (Evanston Northwestern Healthcare) as part of a prospective effort to identify individuals with amniotic fluid embolism who might be suitable for laboratory testing.
The control group comprised 13 Japanese and nine American women who had uneventful vaginal deliveries. Control volunteers were recruited during a 6-month period from the authors' practices because no banked samples were available during the 10 years during which we collected data from patients who had amniotic fluid embolism. Given the anticipated range of complement levels in healthy adults, we estimated that 20 subjects would be needed to define normal complement levels in women who had uncomplicated vaginal deliveries. Informed consents were obtained in keeping with human research protocols approved by our respective institutional review boards. With patients too ill to consent, informed consent was obtained from a family member.
Each control subject had complement levels (C3 and C4) measured at routine phlebotomy upon admission during labor and the day after delivery. Blood samples were processed by the hospitals' clinical laboratories within 24 hours. For all nine women with amniotic fluid embolism, serum was collected within 14 hours of onset of symptoms and promptly frozen for later study. Urine samples were collected 12–24 hours after symptom onset and frozen for subsequent analysis.
Sialyl Tn was measured at Hamamatsu University using reagents supplied by Otsuka Pharmaceutical Company (Tokushima, Japan) as described.11 Serum tryptase, a marker for mast cell degranulation, was measured by Specialty Laboratories (Santa Monica, CA) and IBT Reference Laboratory (Lenexa, KS). Urinary histamine was assessed by Specialty Laboratories for the American women and by SRL Corporation (Tokyo, Japan) for Japanese women. The decision to use urinary histamine as an adjunct for examining mast cell degranulation was a late innovation of our study, permitting determinations in only the most recent study participants. Because of logistic considerations of processing control complement samples from several hospitals on two different continents, samples were evaluated locally, using laboratories with comparable normative complement ranges (Table 1). Each laboratory used the immunologic rate nephelometric method with an inter-assay variability less than 10%. Before the control complement levels from different hospitals were combined, their mean values at each institution were compared and were statistically similar.
Statistical comparisons were made using Wilcoxon signed-rank test for comparison of related measurements in non-normal distributions, and Mann–Whitney tests for unrelated, non-normal distributions, with P < .05 statistically significant. Statistical analyses were performed using the Statistical Package for the Social Sciences (SPSS, Inc., Chicago, IL).
Among nine women with amniotic fluid embolism, six died. All had disseminated intravascular coagulation severe enough to require blood transfusions. Seven women also had significant respiratory distress. Two of nine delivered by cesarean, and their first clinical manifestations were after delivery. Five mothers had their first pregnancies when stricken. Although many presenting symptoms and signs started within minutes of each other, four presented with cardiovascular collapse (severe hypotension or arrest), two had respiratory distress first, and two had disseminated intravascular coagulation as their first sign of illness. Seven of nine had their first indications of disease shortly after birth. Only two women had comorbidities. One woman became ill during a planned cesarean for placenta previa at 35 weeks. The other also had diagnosis of sepsis and was included in the series only after autopsy documented significant fetal material in her pulmonary vasculature.
Seven of nine women had assays for evidence of anaphylaxis—five for serum tryptase alone, one for tryptase and urinary histamine, and one for urinary histamine (Table 2). Of six tryptase tests, four were done within 6 hours of symptom onset, whereas others were done at 12 and 14 hours, respectively. Eight assays for mast cell degranulation in seven women were negative. Individual complement levels for the amniotic fluid embolism subjects are given in Table 2. Mean postpartum C3 levels, 117.2, decreased by 19.3 ± 7.3 mg/dL standard error of the mean (SEM) (8%) from intrapartum value of 127.5 in normal laboring women (P = .003). The corresponding decrease for C4, 31.1 to 29.4, was 3.6 ± 1.6 mg/dL (SEM) (5%) (P = .021). Both means were within the range of normal. The complement levels for amniotic fluid embolism subjects were severely depressed compared with post-partum controls (C3 of 44.0 and C4 of 10.7; P = .018 for C3, P = .012 for C4). Compared with controls, those values were clearly in the clinically abnormal range. The mean complement levels of the groups are shown in Figure 1.
When complement levels were examined as categorical variables, abnormal compared with normal, 100% of controls (22) had normal C3 and C4 levels. In the amniotic fluid embolism group, seven of eight (87.5%) had abnormal C3 levels, and eight of eight (100%) had abnormal C4 levels. When compared using Fisher exact test, both analyses were highly statistically significant (P < .001).
Fetal levels of sialyl Tn are also presented in Table 2. Seven of nine women had levels that were two standard deviations above the mean. In particular, all six women who died had elevated values, whereas only one of the three survivors did.
The most difficult aspect of designing a study of women with amniotic fluid embolism is determining the inclusion criteria because there is no standard clinical definition or laboratory test for survivors. More confusing are the variety of clinical presentations, including seizures or even coma as presenting signs.12 A previous attempt at providing a specific set of diagnostic criteria was too restrictive, not accommodating women who die within a few minutes of clinical illness, as occasionally happens.13
The high mortality rate among Japanese women might be from ascertainment bias because the authors were participants in a national maternal mortality research program. Those who died had fetal material in their pulmonary vasculature, but the sensitivity of that finding is not known. Although fetal material suggests the diagnosis and was a key feature in early descriptions of the disease, its absence in women who otherwise had clinical manifestations of amniotic fluid embolism is problematic. We were also reluctant to apply a diagnostic requirement to those who died that could not be assessed in survivors. Therefore, we did not make that autopsy finding an inclusion requirement in our study, but we did use it to help justify inclusion of the dead patient with concomitant sepsis. All patients had severe disseminated intravascular coagulation, so we do not believe that these findings apply necessarily to women without clinically evident coagulopathy.
Since 1926, when amniotic fluid embolism was first described, investigators speculated about an allergy, hypersensitivity reaction, or anaphylaxis affecting etiology.14–17 One group suggested that complement activation might be important in this condition, although they did not measure serum complement in afflicted women.18,19 Others linked adult respiratory distress syndrome in patients with amniotic fluid embolism to complement activation by classical or alternate pathways.20 The current study sought specific evidence for mast cell degranulation or complement activation in clinically ill women.
We used mast cell degranulation and serum tryptase and urinary histamine as markers. Tryptase has been used as an indicator for anaphylaxis because it is released with histamine during mast cell degranulation. Commercially available as a clinical test, tryptase has also been used to document anaphylaxis on postmortem serum samples.21 Tryptase should be assayed 1–6 hours after onset of symptoms, so we also checked for urinary histamine when possible, because it can persist in urine longer and has been used to detect anaphylaxis in other clinical settings (Greenberger PA, Miller M. Urine histamine during episodes of anaphylaxis [Abstract]. J Allergy Clin Immunol 1994;93:302). Among seven women evaluated for anaphylaxis, five had blood or urine sampled in a timely fashion. The remaining two with 12–14 hour intervals in measuring tryptase levels could be expected to have reduced sensitivity. Data suggest that mast cell degranulation might not play a central role in amniotic fluid embolism.
Complement levels were decreased uniformly in amniotic fluid embolism cases, whereas postpartum control values remained consistently in normal range. The biologic explanation for the 8% decrease in complement levels in our control group after labor is not known, but it is unlikely this decrease is clinically meaningful. A previous series of ten obstetric patients showed that complement values remained within normal range during labor and were comparable to those of other healthy adults.22 Significantly decreased complement levels among women with amniotic fluid embolism suggest that complement activation has an important role in the pathophysiology of the disease.
One concern about the fetal antigen sialyl Tn as a diagnostic test for amniotic fluid embolism was sensitivity. Although elevated in all women who died, it was abnormal in only one of three survivors. An unexpected paradox suggested by our data was that complement activation might be a better diagnostic test because it was positive in all women in whom it could be measured. Other studies should resolve issues of sensitivity and specificity of both assays in other obstetric populations.
Complement activation occurs in seriously ill patients with adult respiratory distress syndrome.18 Whether that is the mechanism that incites amniotic fluid embolism or a secondary phenomenon is not known. In that context, the two women who did not have respiratory distress as part of their clinical presentations still had depressed levels of complement. That might suggest a primary effect of complement activation. Involvement of C4 points to the classical activation pathway and a specific response to antigen.23
1. Berg CJ, Atrash HK, Koonin LM, Tucker M. Pregnancy-related mortality in the United States, 1987–1990. Obstet Gynecol 1996;88:161–7.
2. Hogberg U, Joelsson I. Amniotic fluid embolism in Sweden, 1951–1980. Gynecol Obstet Invest 1985;20:130–7.
3. Gilbert WM, Danielson B. Amniotic fluid embolism: Decreased mortality in a population-based study. Obstet Gynecol 1999;93:973–7.
4. Steiner PE, Lushbaugh CC. Maternal pulmonary embolism by amniotic fluid as a cause of obstetric shock and unexpected death in obstetrics. JAMA 1941;117:1245–54, 1340–5.
5. Adamsons K, Mueller-Heubach E, Myers RE. The innocuousness of amniotic fluid infusion in the pregnant rhesus monkey. Am J Obstet Gynecol 1971;109:977–84.
6. Hankins GD, Snyder RR, Clark SL, Schwartz L, Patterson WR, Butzin CA et al. Acute hemodynamic and respiratory effects of amniotic fluid embolism in the pregnant goat model. Am J Obstet Gynecol 1993;168:1113–30.
7. Morgan M. Amniotic fluid embolism. Anaesthesia 1979;34:20–32.
8. Benson MD, Lindberg RE. Amniotic fluid embolism, anaphylaxis, and tryptase. Am J Obstet Gynecol 1996;175:737.
9. Kobayashi H, Ohi H, Terao T. A simple, noninvasive, sensitive method for diagnosis of amniotic fluid embolism by monoclonal antibody TKH-2 that recognizes NeuAcα2-6GalNac. Am J Obstet Gynecol 1993;168:848–53.
10. Ohi H, Kobayashi H, Hirashima Y, Yamazaki T, Kobayashi T, Terao T. Serological and immunohistochemical diagnosis of amniotic fluid embolism. Semin Thrombosis Hemostasis 1998;24:479–84.
11. Kjeldsen T, Clausen H, Hirohashi S, Ogawa T, Iijima H, Hakomori S, et al. Preparation and characterization of monoclonal antibodies directed to the tumor-associated O-linked sialosyl-2--6 alpha-N-acetylgalactosaminyl (sialosyl-Tn) epitope. Cancer Res 1988;48:2214–20.
12. Courtney LD. Review: Amniotic fluid embolism. Obstet Gynecol Surv 1974;29:169–77.
13. Benson MD. Nonfatal amniotic fluid embolism. Three possible cases and a new clinical definition. Arch Fam Med 1993;2:989–94.
14. Meyer JR. Embolia pulmonar amnio caseosa. Brasil-Medico 1926; 2:301–3.
15. Mallory GK, Blackburn N, Sparling HJ, Nickerson DA. Maternal pulmonary embolism by amniotic fluid: Report of three cases and discussion of the literature. N Engl J Med 1950;243:583–7.
16. Attwood HD. Fatal pulmonary embolism by amniotic fluid. J Clin Pathol 1956;9:38–46.
17. Shnider SM, Moya F. Amniotic fluid embolism. Anesthesiology 1961;22:108–19.
18. Jacob HS, Hammerschmidt DH. Tissue damage caused by activated complement and granulocytes in shock lung, post perfusion lung, and after amniotic fluid embolism: Ramifications for therapy. Ann Chir Gynaecol 1982;71(Suppl 196):3–9.
19. Hammerschmidt DE, Ogburn PL, Williams JE. Amniotic fluid activates complement. A role in amniotic fluid embolism syndrome? J Lab Clin Med 1984;104:901–7.
20. Hollingsworth HM, Irwin RS. Acute respiratory failure in pregnancy. Clin Chest Med 1992;13:723–40.
21. Schwartz LB. Tryptase from human mast cells: Biochemistry, biology and clinical utility. Monogr Allergy 1990;27:90–113.
22. Kovar IZ, Riches PG. C3 and C4 complement components and acute phase proteins in late pregnancy and parturition. J Clin Pathol 1988;41:650–2.
© 2001 The American College of Obstetricians and Gynecologists
23. Liszewski MK, Atkinson JP. The complement system. In: Paul WE, ed. Fundamental immunology. New York: Raven Press, 1993:917–39.