- Email: [email protected]
Tumor necrosis factor-α is elevated in plasma and amniotic fluid of patients with severe preeclampsia
Tumor necrosis factor-a is elevated in plasma and amniotic fluid of patients with severe preeclampsia Michael J. Kupferminc, MD, Alan M. Peaceman, MD, Thomas R. Wigton, MD, Karen A. Rehnberg, BA, and Michael L. Socol, MD
Chicago, Illinois OBJECTIVE: Our purpose was to investigate whether markers for activation of the immune system are present in patients with preeclampsia by assessing maternal plasma and amniotic fluid for tumor necrosis factor-a and interleukin-1 [3. STUDY DESIGN: Twenty-one patients with severe preeclampsia composed the study group (group A). An antepartum comparison group was composed of healthy nulliparous patients not in labor and matched for gestational age (group B). Another control group consisted of term nulliparous patients in labor with uneventful pregnancies (group C). Maternal plasma samples were collected from all patients at recruitment and from patients in groups A and C immediately after delivery and again 20 to 24 hours post partum. Amniotic fluid was also collected from patients in groups A and C during labor. All samples were collectively assayed for tumor necrosis factor-a and interleukin-1 [3 by specific enzyme-linked immunoassays. RESULTS: Before labor tumor necrosis factor-a was detected more frequently in the plasma of preeclamptic patients than in the plasma of patients in group B (12/16 vs 5/16, P < 0.05) and in higher concentrations (median 35 pg/ml vs median 0 pg/ml, p < 0.05). Although tumor necrosis factor-a was frequently detected in the plasma of patients in group C in early labor (16/20), concentrations were higher in the four preeclamptic patients first sampled in early labor (210 pg/ml vs 65 pg/ml, p < 0.05). Similarly, amniotic fluid levels of tumor necrosis factor-a were increased in preeclamptic patients compared with control patients. At delivery tumor necrosis factor-a was more likely to be identified in the plasma of preeclamptic patients and was found in higher concentrations, but by 20 to 24 hours post partum measurements in the preeclamptic and control patients were similar. There were no differences in the frequency with which interleukin-1 [3 was detected or the concentration of interleukin-1 [3 in any of the samples. CONCLUSIONS: Tumor necrosis factor-a is increased in the plasma and amniotic fluid of patients with severe preeclampsia. These data are suggestive of a role for abnormal immune activation in the pathophysiologic mechanisms of preeclampsia. (AM J OBSTET GVNECOL 1994;170:1752-9.)
Key words: Preeclampsia, cytokine, tumor necrosis factor-a, interleukin-IJ3
In spite of intensive efforts to delineate the pathophysiologic mechanisms of preeclampsia, neither a specific cause nor a pathogenesis has been identified. Recent data suggest that endothelial damage, vasoconstriction, placental ischemia, and enhanced coagulation are caused by an as yet unknown serum factor produced by the trophoblast, 1. 2 and a product of the immune system could be this factor. Abnormal immune system activation by the fetal allograft has been suggested as contributing to the development of preeclampsia." but From the Section of Maternal-Fetal Medicine, Department of Obstetrics and Gynecology, Northwestern University Medical School and Northwestern Memorial Hospital. Presented at the Sixty-first Annual Meeting of The Central Association of Obstetricians and Gynecologists, White Sulphur Springs, West Virginia, October 28-30, 1993. Reprint requests: Michael}. Kupferminc, MD, 333 E. Superior St., Suite 410, Chicago, 1L 60611 Copyright © 1994 by Mosby-Year Book, Inc. 0002-9378/94 $3.00 + 0 6/6/55190
1752
no specific markers of immune activation have been identified in patients with this syndrome. Cytokines are peptide mediators of the immune system. 4 Tumor necrosis factor-a (TNF-a) is a cytokine derived from macrophages and lymphocytes and is involved with such essential biologic functions as immunoregulation, modulation of cell growth and differentiation, coagulation, and endothelial cell function. 4.7 Endothelial cell injury after TNF -a-mediated activation of the immune system may result in the secretion of vasoactive substances, an increase in vascular permeability, and intravascular coagulation,4.7 all of which are pathophysiologic changes observed in preeclampsia. Interleukin-IJ3 (IL-IJ3), another cytokine produced by macrophages, is believed to be one of the most potent regulators of prostaglandin biosynthesis by human decidua." It is secreted with TNF-a in response to inflammation and may serve as a marker for infection. 4 • 8. 9 Given the importance of prostaglandin production in
Kupferminc et al.
Volume 170, Number 6 Am J Obstet Gynecol
the pathophysiologic mechanisms of preeclampsia, it is possible to postulate a role for both IL-l J3 and TNF-o: in the development of this disease, We investigated whether activation of the immune system could be detected in patients with preeclampsia by assessing maternal plasma and amniotic fluid for the presence ofTNF-o: and IL-IJ3, Material and methods
Three groups of patients were recruited from those patients delivered at Northwestern Memorial Hospital between July 1992 and April 1993. The study cohort (group A) was composed of 21 patients with severe preeclampsia without underlying chronic hypertension or renal disease. All preeclamptic patients had persistent elevation of blood pressure ~ 140/90 mm Hg in conjunction with proteinuria. Preeclampsia was characterized as severe because of blood pressure ~ 160/110 mm Hg in 12 patients; proteinuria> 5 gm per 24 hours in three patients; platelet count < 100,000/mm2 in two patients; hemolysis, elevated liver enzymes, low platelets (HELLP syndrome) in three patients; and eclampsia in one patient. Nineteen of the preeclamptic patients were nulliparous. Seventeen patients in this group were not in labor but were delivered because of the severity of the disease; the other four patients were seen in early labor (regular uterine contractions with a cervix dilated to <4 cm). None of these pregnancies were complicated by preterm labor, premature rupture of membranes, or clinical chorioamnionitis. An antepartum comparison group (group B) consisted of healthy nulliparous patients not in labor and matched for gestational age with the nonlaboring patients in group A for which a prelabor sample was available (n = 16). None of these patients had preeclampsia or other pregnancy complications. Another control group (group C) consisted of nulliparous patients' in early labor with uneventful pregnancies at 38 to 40 weeks' gestation. The study was approved by the Northwestern University Institutional Review Board, and informed consent was obtained from each patient at the time of recruitment. Maternal venous blood samples were collected on hospital admission from patients in group A and C and during an outpatient prenatal visit for patients in group B. Additional maternal blood samples were obtained from patients in group A and C immediately after delivery and again at 20 to 24 hours post partum. Amniotic fluid specimens were collected during labor at a cervical dilatation of 2 to 4 cm from 17 preeclamptic patients and from 19 patients in group C. For patients in group A a specimen of amniotic fluid was collected from free-flowing fluid at the time of artificial rupture of membranes (n = 9), after placement of an intrauterine pressure catheter (n = 5) or by needle aspiration at
1753
cesarean section (n = 3). For patients in group C the number of samples collected by each method was 13, three, and three, respectively. In four nonlaboring preeclamptic patients amniotic fluid was collected by amniocentesis before induction of labor (n = 3) or during cesarean section for suspected fetal compromise (n = 1).
Sample processing. Blood and amniotic fluid specimens were collected in endotoxin-free tubes. The blood and amniotic fluid samples were centrifuged, respectively, at 1000g and 200g for 10 minutes. The plasma and amniotic fluid supernatants were then stored in polypropylene tubes at -70 0 C and collectively assayed for TNF -0: and IL-1 J3 by a technician blinded to patient group assignment. TNF-o: was measured by an enzyme-linked immunoassay (Human TNF-o:, Endogen, Boston, Mass.), which incorporates a double-antibody sandwich with a mouse anti-TNF-o: antibody as the capture reagent and M-301A mouse immunoglobulin conjugated to horseradish peroxidase as the detecting antibody. The assay is specific for TNF-o: and does not bind human lymphotoxin (TNF-J3), human IL-10:, human IL-2, human IL-3, human IL-6, human IL-7, human IL-8, or interferon gamma. The assay is designed to measure levels of human TNF -0: between 5 and 1715 pg/ml with interassay and intraassay coefficients of variation of 9.3% and 8.3%, respectively. We determined the minimal detectable concentration of TNF -0: in our plasma specimens to be 10 pg/ml. This amount was determined as the amount of TNF -0: that provides absorbance at two SDs above the absorbance of the diluent alone. In addition, we determined the minimal detectable concentration of TNF -0: in amniotic fluid to be 60 pg/ml by means of pooled amniotic fluid as the diluent, which was shown to be negative for TNF -0: with a standard diluent. The matrix effect of the amniotic fluid was determined to be 2 SDs above the difference between the absorbance of the diluent and the absorbance of the amniotic fluid. IL-1J3 was also measured by an enzyme-linked immunoassay (Human IL-1J3, Endogen), which incorporates a double-antibody sandwich with a monoclonal mouse anti-IL-IJ3 antibody as the capture reagent and M-301A mouse immunoglobulin conjugated to acetylcholinesterase as the detecting antibody. The assay is specific for IL-IJ3 and has a cross reactivity of 1.5% with human IL-lo:. It does not cross-react with human IL-2, human IL-3, human IL-6, human IL-7, human IL-8, human TNF -0:, or human interferon gamma. The assay is designed to measure levels of human IL-l J3 between 4 and 540 pg/ml, with interassay and intraassay coefficients of variation of 7.8% and 6.3%, respectively. The minimal detectable concentration of IL-1 J3 in maternal plasma was determined to be 10 pg/ml, and the mini-
1754
June 1994 Am J Obstet Gynecol
Kupferminc et al.
Table I. Maternal clinical characteristics
Maternal age (yr) Gestational age (wk) Race White Black Hispanic Parity 0 ~1
Vaginal delivery Cesarean delivery Length of labor (hr)t
Group A (n = 21)
Group B (n = 16)
Group C (n = 20)
26.6 ± 9.5 34.3 ± 4.5
28.6 ± 3.8 34.6 ± 4.4
27.9 ± 4.7 39.2 ± 1.1*
7 12 2
7 9 0
6 14 0
19 2 17 4
16 0 NA NA NA
20 0 17 3
11.4 ± 6.7
10.6 ± 3.3
Group A, Preeclamptic patients; group B, antepartum control group; group C, intrapartum control group. NA, Not available. *p < 0.0001 compared with group A. tVaginal delivery only.
mal detectable concentration in amniotic fluid was determined to be 40 pg/ml. These minimal concentrations were determined as described for the TNF-a assay. Statistical analysis. All TNF-a and IL-I ~ results are presented as medians with ranges. Categoric data were analyzed with 1(2 or Fisher's exact test. Continuous data were compared with the Mann-Whitney U test or Student t test. Statistical significance was assumed at a p of <0.05. Results
The maternal clinical characeristics of the three groups of patients are presented in Table I. No significant differences were noted between groups A and B in maternal age, gestational age, race, and parity. Groups A and C differed only in mean gestational age. The length of labor, defined by the interval from the onset of regular uterine contractions until delivery, was similar among those patients delivered vaginally. Maternal plasma concentrations ofTNF-a at recruitment in the preeclamptic patients and comparison groups are displayed in Fig. I. Sixteen plasma samples were obtaine'd from the 17 preeclamptic patients not in labor. TNF -a was detected more frequently in the preeclamptic patients than in the antepartum control patients in group B (12/16 vs 5/16, P < 0.05), and the concentrations of TNF-a were significantly higher (35 pg/ml [0 to 160 pg/ml] vs 0 pg/ml [0 to 60 pg/ml], p < 0.05). TNF-a levels within groups A and B were not correlated with gestational age. The frequency of detection and plasma concentrations of TNF-a were higher in the term control patients in labor (group C) compared with patients in group B (16/20 vs 5/16, P < 0.01; 65 pg/ml versus 0 pg/ml, p < 0.01). However, when the plasma levels of TNF -a from the four preeclamptic patients in labor were compared with those of patients in group C, TNF-a levels were significantly
increased (210 pg/ml [130 to 350 pg/ml] vs 65 pg/ml [0 to 280 pg/ml]; p < 0.05). Maternal plasma TNF-a concentrations immediately after delivery for the preeclamptic patients and patients in group C are depicted in Fig. 2. For those parturients delivered vaginally TNF-a was detected more frequently (15/17 vs 7/16, P < 0.01) and at higher concentrations (130 pg/ml [0 to ll20 pg/ml] vs 0 pg/ml [0 to 490 pg/ml], p < 0.01) in the preeclamptic patients in spite of a similar length of labor. If data from patients delivered by cesarean section (group A: failed attempt at induction of labor [3], suspected fetal compromise [1]; group C: breech presentation [3]) are included, the detection rates (17/20 vs 7/19, P < 0.01) and concentrations ofTNF-a (80 pg/ml vs 0 pg/ml, p < 0.01) were ! still significantly different between the two groups. By 20 to 24 hours post partum the frequency of TNF-a detection (16/21 vs 16/18) and the TNF-a concentrations (30 pglml [0 to 140 pglml] vs 50 pglml [0 to 140 pg/ml]) in the preeclamptic patients and patients in control group C were similar. Amniotic fluid. As shown in Fig. 3, TNF-a was detected more frequently in the amniotic fluid samples obtained from preeclamptic patients during labor than in laboring control patients (17/17 vs 12/19, P < 0.01), and TNF-a concentrations were significantly higher in the preeclamptic patients (310 pg/ml [60 to 490 pg/ml] vs 60 pg/ml [0 to 250 pg/ml], p < 0.0001). The amniotic fluid TNF-a concentrations collected from the four preeclamptic patients not in labor were also significantly elevated (270 pg/ml [180 to 350 pglml]). Within each group of patients the frequency of detection and TNF-a concentrations were similar in free-flowing amniotic fluid compared with amniotic fluid collected by other means. Maternal plasma and amniotic fluid IL-lJi.Plasma IL-l ~ was detected on recruitment in only one patient
Kupferminc et al. 1755
Volume 170, Number 6 Am J Obstet Gynecol
375
1200
•
350 325
1100 1000
300
•• • •
275 250
~
E
CD
B a LL
z
I-
• •
225 200 175
o o o
150 125 100 50 25 0
•
o o o
75
-+ • •
000
co
..i8000
·...0006·. ·...···. ···...·. ·············......~ ...···. ·..~·.. n
Group A
= 16
n
=4
Group B
n
= 16
Group C
n
= 20
Fig. 1. Plasma TNF-a detection rates and concentrations on recruitment were higher in nonlaboring preeclamptic patients (group A) than in antepartum control patients (group B). TNF -a levels were also increased in preeclamptic patients in early labor compared with control patients in early labor (group C). Open circles, Patients not in labor; closed circles, patients in early labor; horizontal bars, median concentrations in each group.
in each of the three groups. Plasma IL-lj3 was detected at delivery in only three of 20 preeclamptic patients and four of 19 patients in group C. IL-lj3 was not detected in any of the plasma samples collected 20 to 24 hours post partum. The frequency of IL-lj3 detection in amniotic fluid obtained in labor and the median concentrations of IL-lj3 were also comparable between patients in group A and group C (7/17 vs 9/19; 0 pg/ml [0 to 660 pg/ml] vs 0 pg/ml [0 to 660 pglml]). Comment
Our data indicate that TNF -a is detected more frequently and at higher concentrations in the plasma of patients with severe preeclampsia than in patients with uncomplicated pregnancies. The increase in plasma TNF -a noted before labor was also found to be independent of gestational age. Although labor appeared to increase the detection rate and concentration of TNF-Ci. for control patients, further elevations in TNF-a concentrations were seen in preeclamptic parturients. Additionally, amniotic fluid levels of TNF-Ci. in early labor were also significantly elevated in patients with preeclampsia. Plasma TNF-Ci. concentrations remained el-
.
900
800
-E ~
01
B
a
LL
z
I-
700 600 500 400
300 200
• •
.....
• •
• • o ......··_······. ······rA/ItI········. . · ·. ·••·••.a:·.989········· . ·. . . . .
100
Group A n = 20
Group C n = 19
Fig. 2. Plasma TNF-a detection rates and concentrations at delivery were higher in preeclamptic patients (group A) than in control patients (group C). Closed circles, Vaginal deliveries; open circles, cesarean births; horizontal bars, median concentrations in each group.
evated in preeclamptic patients at delivery, but by 20 to 24 hours post partum the plasma TNF -Ci. measurements were comparable to those of the control patients. The increase in TNF -Ci. concentrations in the postpartum control patients compared with delivery values may be a secondary effect of the reparative process at the placental site. In contrast to TNF-Ci., no increase in IL-lj3 was detected in plasma or amniotic fluid samples from preeclamptic patients. These data support a role for TNF -Ci., but not IL-lj3, in the pathogenesis of preeclampsia. TNF -Ci. was not consistently found to be either present or absent in all samples from a given patient. This observation suggests a complex or nonlinear mechanism of TNF -a secretion. Because the half-life of TNF-Ci. is only a few minutes in the circulation, 10 a single blood sample may fail to detect a periodic elevation or decline in TNF -Ci. levels. Further, an unknown portion ofTNF-Ci. may be bound to its soluble receptor and may not be detected by the bioreactive or immunoreactive assays. 10 TNF -Ci. and IL-lj3 are not considered normal constituents of amniotic fluid, but both have been detected in amniotic fluid during parturition. II Intraamniotic macrophages, decidual macrophages, and trophoblast
1756 Kupferminc et aI.
••
500
450
•••
400
8
350
~
a a
Ii.
zt-
June 1994
Am J Obstet Gynecol
300
--•
• • 200 • 8 •• 150 • • 100 -_. _-_. _. . . . ._. _. . . . . _. . . _. . ;_. . _. . _. _. . . _. _. . . . . -A-.._. . . .. 50 250
..
0 n
=
Group A 4 n 17
=
••••••• Group C n = 19
Fig. 3. Amniotic fluid TNF-a detection rates and concentrations were higher in preeclamptic patients (group A) than in control patients (group C). OPen circles, patients not in labor; closed circles, patients in early labor; horizontal bars, median concentrations in each group.
are all possible sources, because all are capable of cytokine synthesis. 9. 11·1' Among women with negative amniotic fluid cultures, TNF-a was detected in the amniotic fluid of 20% of patients in labor at term and in 18.7% of patients in pre term labor. II In comparable groups of patients IL-l~ was detected in the amniotic fluid of 61.5% and in 47.6% of patients, respectively.9 TNF-a and IL-l ~ are present in higher concentrations in patients with preterm labor and microbial invasion of the amniotic fluid cavity,9. II and it has been postulated that these cytokines play a role in the pathophysiologic mechanisms of preterm labor in this setting. 8. 9. II. 14 Although we did not perform cultures of the amniotic fluid, increased TNF-a levels in the amniotic fluid secondary to infection is unlikely, because detection rates and levels of IL-l ~ were similar to those reported for patients with preterm labor and lower than those for term parturients with negative cultures of the amniotic fluid. 9 The increased detection of TNF-a in our amniotic fluid samples may be related to the underlying disease process or to the stage of labor at collection. Immunologic stimuli occurring as a part of the adaptation of the immune system to pregnancy may account for the detection of TNF-a in plasma and amniotic fluid in some normal pregnant patients. On
the other hand, the increased detection rate and concentrations of TNF-a in preeclamptic patients may follow abnormal activation of the immune system. Indeed, one postulated mechanism for the development of preeclampsia involves abnormal activation of the immunologic system against the fetal allograft. 3 Alloantigenic fetal tissue may be exposed to the maternal immune system as a result of impaired immunologic suppression, causing a maternal immune response similar to graft rejection, manifested in part by activation of macrophages and lymphocytes with secretion ofTNF-a. Consistent with this premise is the observation that TNF-a has been shown to be an important mediator in transplantation rejection. 15 An altered immune response to the fetal allograft may also play a role in the impaired trophoblastic invasion into maternal spiral arterioles that has been reported in preeclampsia by 16 to 20 weeks' gestation. I Recent observations suggest that TNF-a is capable of limiting trophoblast invasion in the uterus, as well as inhibiting deoxyribonucleic acid synthesis and trophoblastic cellular proliferation in vitro!6. 17 Furthermore, TNF-a is also increased in midtrimester amniotic fluid of pregnancies resulting in fetal growth retardation,18 which has similar pathophysiologic characteristics to preeclampsia. Involvement of the cellular branch of the immune system and its products, the cytokines, in preeclampsia remains poorly understood, but the presence of and biologic capabilities of TNF-a strongly suggests that a role in the pathogenesis of preeclampsia is plausible. TNF-a is involved in many pathologic processes such as infection, tumor cytotoxicity, and endotoxic shock. 5-7 TNF-a may cause tissue damage either directly through activation of protease, collagenase, or phospholipase A2 enzymes or through oxygen radicals. 5 Direct damage to vascular endothelial cells, reduced regional blood flow, occlusion of vessels, and increased endothelial permeability are characteristic effects of pathologically secreted TNF_a. 5-7. 16 TNF-a also enhances procoagulant activity, by inducing endothelial cell tissue factor, and suppresses protein C activation, an antithrombotic mechanism that functions on the surface of quiescent endothelial cells. 5. 7 Finally, TNF-a increases the release of vasoactive substances such as endothelin,19 which is increased in preeclamptic patients,20 and platelet-derived growth factor, a potent constrictor of vascular smooth muscle 21 that is released from endothelial cells after exposure to serum from preeclamptic patients in vitro. 22 Our preliminary data support the involvement of abnormal immune activation in the pathogenesis of preeclampsia. TNF-a may play an active role in mediating the endothelial damage, ischemia, and enhanced coagulation seen in preeclampsia. The low detection rates and concentrations of IL-l~ imply different pat-
Kupferminc et al.
Volume 170, Number 6 Am J Obstet Gynecol
terns of secretion for the two cytokines in the setting of preeclampsia, as opposed to contemporary elevation in response to infection or inflammation. It remains to be determined whether increased TNF-a levels are present in patients with mild preeclampsia and are detectable before clinical manifestations appear. REFERENCES
1. Roberts ]M, Taylor RN, Musici T], et al. Preeclampsia: an endothelial cell disorder. AM] OBSTET GYNECOL 1989; 161: 1200-4. 2. Rodgers GM, Taylor RN, Roberts ]M. Preeclampsia is associated with a serum factor that is cytotoxic to human endothelial cells. AM ] OBSTET GYNECOL 1988;159: 908-14. 3. Sibai BM. Immunologic aspects of preeclampsia. Clin Obstet Gynecol 1991;34:27-34. 4. Ari K, Lee F, Miyajima A, Miyatake S, Ari N, Yokota T. Cytokines: coordinators of immune and inflammatory responses. Annu Rev Biochem 1990;59:783-836. 5. Le], Vilcek J. Biology of disease. Tumor necrosis factor and interleukin-l: cytokines with multiple overlapping biological activities. Lab Invest 1987;56:234-48. 6. Beutler B, Cerami A. Ibe biology of cachectin{rNF-a primary mediator of the host response. Annu Rev Immunol 1989;7:625-55. 7. Beutler B, Cerami A. Cachectin: more than a tumor necrosis factor. N Engl] Med 1987;316:379-85. 8. Romero R, Brody DT, Oyarzun E, et al. Infection and labor. III. Interleukin-l: a sign for the onset of parturition. AM] OBSTET GVNECOL 1989;160:1117-23. 9. Romero R, Mazor M, Brandt F, et al. Interleukin-la and interleukin-lj3 in preterm and term human parturition. Am] Reprod Immunol 1992;27:117-23. 10. Dinarello CA, Gelfand ]A, Wolf SM. Anticytokine strategies in the treatment of the systemic inflammatory response syndrome. ]AMA 1993;269: 1829-35. 11. Romero R, Mazor M, Sepulveda W, Avila C, Copeland D, Williams ]. Tumor necrosis factor in pre term and term labor. AM] OBSTET GYNECOL 1992;166:1576-87. 12. Casey ML, Cox SM, Beutler B, Milewich L, Macdonald P. Cachectin/tumor necrosis factor-a formation in human decidua. Potential rule of cytokines in infection induced preterm labor.] Clin Invest 1989;83:430-6. 13. Vince G, Shorter S, Starkey P, et al. Localization of tumor necrosis factor production in cells at the materno/fetal interface in human pregnancy. Clin Exp Immunol 1992; 88: 174-80. 14. Romero R, Manogue KR, Mitchell MD, et al. Infection and labor. IV. Cachectin-tumor necrosis factor in the amniotic fluid of women with intraamniotic infection and preterm labor. AM] OBSTET GYNECOL 1989;161:336-41. 15. Maury CP], Teppo AM. Raised levels of cachectinltumor necrosis factor-a in renal allograft rejection. ] Exp Med 1987; 166: 1132-7. 16. Hunt ]S. Cytokines network in the utero-placental unit. Macrophage as pivotal regulatory cells.] Reprod ImmunoI1989;16:1-25. 17. Hunt]S, Soares MJ, Lei MG, et al. Products oflipopolysaccharide-activated macrophages (tumor necrosis factor-a, transforming growth factor-j3) but not lipopolysaccharide modity DNA synthesis by rat trophoblast cells exhibiting the 80-kDa lipopolysaccharide-binding protein. ] ImmunoI1989;143:1606-13. 18. Heyborne KD, Eitkin SS, McGregor ]A. Tumor necrosis factor-a in midtrimester amniotic fluid is associated with impaired intrauterine fetal growth. AM ] OBSTFT GYNECOL 1992;167:920-5. 19. Vemulapalli S, Chiu P]S, Rivelli M, Foster CJ, Sybertz EJ. Modulation of circulating endothelin levels in hyperten-
1757
sion and endotoxemia in rats. ] Cardiovasc Pharmacol 1991; 18:895-902. 20. Nova A, Sibai BM, Barton ]R, Mercer BM, Mitchell MD. Maternal plasma levels of endothelin is increased in preeclampsia. AM] OBSTET GYNECOL 1991;165:724-7. 21. Hajjar KA, Hajjar DP, Silverstein RL, Nachman RL. Tumor necrosis factor mediated release of platelet-derived growth factor from cultured endothelial cells.] Exp Med 1987; 166:235-45. 22. Taylor RN, Musci T], Rodgers GM, Roberts ]M. Preeclamptic sera stimulate increased platelet-derived growth factor mRNA and protein expression by cultured human endothelial cells. Am] Repron Immllnol 1991 :2'i: 1Wi-f<.
Discussion New Orleans, Louisiana. TNF-a is a 17 kd soluble mediator of cellular immunity. This cytokine is produced by macro phages, T cells, thymocytes, B cells, and natural killer cells in response to a wide variety of stimuli, including infectious agents. TNF -a may also be produced by human decidua in response to bacterial products, and it stimulates prostaglandin production by amnion and decidua. TNF-a and other cytokines appear to participate in the development of labor in patients with microbial invasion of the amniotic cavity.l. 2 For nearly a century investigators have believed in an immune cause for preeclampsia. This study looks at the role of TNF-a as an immune system mediator in the pathophysiologic mechanisms of preeclampsia. The authors describe a study group of severely preeclamptic patients (group A). They construct two control groups - one matched for gestational age but not in labor (group B) and the other selected from laboring patients at term (group C). Neither control group included patients with preeclampsia or clinically identified infection. Plasma and amniotic fluid TNF -a concentrations were determined as described. Comparison by nonparametric techniques was performed. Summarized results suggest that nonlaboring women with severe preeclampsia have higher plasma TNF -a values than do nonlaboring nonpreeclamptic women at the same gestational age. In early labor severe preeclamptics have higher TNF-a values than do laboring control patients at term. At delivery plasma TNF -a was still significantly higher in preeclamptics. By 20 to 24 hours post partum TNF -a values were similarly elevated for both groups (A and C). Amniotic fluid analysis revealed that TNF-a was present in higher concentrations among laboring severely preeclamptic women than among laboring nonpreeclamptic women at term. Significant results for IL-l ~ were absent. From the study presented we observed that (1) elevated plasma TNF-a levels are associated with severe preeclampsia and (2) elevated plasma TNF -a levels are associated with labor. Additionally, we observe that amniotic fluid TNF -a is more often identified in laboring severely preeclamptic patients than in controls. We conclude that TNF-a may be involved in the manifestation of severe preeclampsia. However, it is not shown to be causative. Elevated amniotic fluid TNF-a DR. JOSEPH M. MILLER, JR.,
1758
Kupferminc et aJ.
levels have been identified in term patients in labor without preeclampsia (this study) or infection.' Additionally, plasma TNF -Ci is frequently found in normal term laboring patients. By 20 to 24 hours post partum normal and severely preeclamptic patients have similar values. If TNF -Ci was causally associated, one would expect preeclampsia to be observed. Perhaps TNF -Ci is a reactor to rather than an initiator of preeclampsia. The authors call for additional study of mild preeclamptics. Longitudinal study would also be indicated. I have two concerns about the assay. First, among uncomplicated laboring patients at term the authors found 63% with detectable amniotic fluid TNF-Ci as opposed to 20% reported by Romero et al.' In nonlaboring patients 31 % of normal and 75% of severely preeclamptic patients had detectable plasma TNF-Ci levels. Although statistical significance exists, why do many normal patients have identifiable TNF -Ci levels without evidence of infection or preeclampsia? Second, infection is associated with the presence of cytokines in amniotic fluid. Although standard infectious morbidity was absent, a negative amniotic fluid culture would add to the strength of the observations. REFERENCES 1. Romero R, Moshe M, Sepulveda W, et al. Tumor necrosis factor in preterm and term labor. AM ] OBSTET GYNECOL 1992; 166: 1576-87. 2. Romero R, Mazor M, Brandt F, et al. Interleukin-la and interleukin-l~ in preterm and term human parturition. Am ] Reprod ImmunoI1992;27:177.
Galveston, Texas. Elevated TNF -Ci in the plasma of women with severe preeclampsia is potentially a very important finding. Elevated levels could explain why these women have high blood pressure. TNF -Ci has (1) an inhibiting effect on nitric oxide' and (2) stimulatory action on endothelin-12 and prostaglandins. 3 Any of these effects could result in elevated blood pressure. What do we know about this cytokine, which is derived from macrophages and leukocytes during pregnancy? TNF-Ci is elevated in the amniotic fluid of women with intraamniotic infection. Although the amniotic fluid was not cultured in this study, intraamniotic infection is unlikely to have played a role in the elevation of the TNF -Ci in the study population, because all these groups had intact membranes at the start of the study. These data strongly suggest that TNF -Ci is elevated in the plasma of both the severely preeclamptic and control women who were in labor (Fig. 1 in article). The amniotic fluid TNF -Ci levels were higher in the severely preeclamptic women than in the control women in labor. No comparisons were made between laboring and nonlaboring women with regard to TNF -Ci levels in amniotic fluid. These data are at odds with Opsjon et al.: who reported a threefold rise in tumor necrosis factor in the amniotic fluid of women in spontaneous labor. They did not find any elevation of tumor necrosis factor in the plasma of women in labor, and in only one of 27 DR. GARLAND D. ANDERSON,
June 1994 Am J Obstet Gynecol
women was it detected in the plasma of women in labor. Would you comment on the different findings of the two studies? It is not known what effect conical effacement of cervical dilatation before labor has on tumor necrosis factor levels. Because cervical effacement and dilatation were not controlled in this study, this may affect the findings. The small sample size, lack of control for cervical dilatation, and inconsistent sampling, particularly with regard to amniotic fluid, leaves the findings of elevated tumor necrosis factor levels in the plasma and amniotic fluid of women with severe preeclampsia suspect. Because some women with severe preeclampsia have normal TNF-Ci levels and some normal women have elevated TNF -Ci levels, preeclampsia occurs independent ofTNF-Ci. IfTNF-Ci is elevated as a result ofpreeclampsia, what is the source of the elevated levels? REFERENCES 1. Yoshizumi M, Perrella M, Burnett]C]r, Lee M-E. Circ Res 1993;73:205-9. 2. Greenberg S, ]ianming X, Wang YE, et al. Tumor necrosis factor-alpha inhibits endothelium-dependent relaxation. ] Appl Physiol 1993;74:2394-403. 3. Romero R, Manogue KR, MitcheUMD, et al. Infection and labor. IV. Cachectin-tumor necrosis factor in the amniotic fluid of women with intraamniotic infection and preterm labor. AM] OBSTET GYNECOL 1989;161:336-41. 4. Opsj(lln SL, Wathen NC, Tingulstad S, et al. Tumor necrosis factor, interleukin-l, and interleukin-6 in normal human pregnancy. AM] OBSTET GYNECOL 1993;169:397-404. DR. ROBERT C. KAUFMAN, Springfield, Illinois. I agree with the comment that this may be a reactive elevation of tumor necrosis factor. Did the authors try to relate the tumor necrosis factor level to the size of the baby, which may be an indication of placental destruction? DR. JAMES N. MARTIN, JR., Jackson, Mississippi. The authors have 21 patients with preeclampsia, and it appears to me that they may be "lumpers" rather than "splitters." I noted that three of the women had HELLP syndrome and two had thrombocytopenia. Because investigators from Chicago (Gleicher's group) have suggested that HELLP syndrome might be an immunologic variant of preeclampsia, I wonder whether the authors have examined the subset of the data for these five patients apart from the larger group of 16 women with preeclampsia but without thrombocytopenia or HELLP syndrome. If the presumption is that this elevated factor is related to increased vascular permeability, could they also relate the clinical findings of the patients with the highest levels of this factor and see any association at all? DR. STEPHEN J. FORTUNATO, New Orleans, Louisiana. I would ask whether the authors have stratified the level of TNF that they've seen for birth weight. A recent publication has shown that at 16 weeks there is an increased level of tumor necrosis factor in those babies destined to be small for gestational age. We've also documented in our laboratory that TNF -Ci is produced both at the message level and at the peptide level from
Volume 170, Number 6 Am J Obstet Gynecol
the amnion and the chorion. This might be another source of TNF. DR. KUPFERMINC (Closing). We found a detection rate of 63% for TNF -a in the amniotic fluid of normal parturients at 2 to 4 cm dilatation, with a median concentration of 60 pg/ml; TNF -a was detected in all preeclamptic patients with a median concentration of 310 pg/ml. It was previously reported by Romero et al. that 20% of normal parturients had detectable TNF-a in the amniotic fluid, whereas OpsjlOn et al. with a bioassay reported a detection rate of 100% and a median concentration of 58 pg/ml in the amniotic fluid of normal parturients. Possible explanations for these differences are as follows: (1) Immunoassays and bioassays may have different sensitivities for TNF-a detection; (2) substantial amounts of TNF-a may be bound to naturally occurring inhibitors or soluble TNF receptors, interfering with detection by some assays; (3) collection of amniotic fluid at different stages of labor may result in different detection rates. Variation between our data and those previously reported by OpsjlOn et al. for plasma values during labor may also be related to these same explanations. In addition, because the half-life of TNF -a in the circulation is only a few minutes, a single blood sample may fail to detect a periodic elevation or a decline in TNF-a. The manufacturer of the assay we performed states that this assay measures both free and bound TNF-a. As further support for the validity of our measurements of TNF -a in the plasma of pregnant patients, we also tested for TNF -a in plasma of 20 nonpregnant reproductive-age patients, where TNF-a is not ordinarily detectable. TNF -a was not detected in any of these patients. Although we cannot ensure that the elevation of TNF -a was not related to infection, none of the patients was infected clinically. It is important to note that IL-I13, which is normally increased compared with TNF -a in response to infection or inflammation, was not increased in the preeclamptic patients compared with control patients. This also argues against the
Kupferminc et al.
1759
possibility of infection as the cause of increased TNF-a levels in these patients. As for a correlation between birth weight and TNF-a levels, we did not find any. Furthermore, the five preeclamptic patients who were delivered of small-for-gestational-age newborns had plasma and amniotic fluid TNF-a concentrations similar to those of preeclamptic patients with appropriate-for-gestational-age newborns. In addition, we did not find any difference in TNF-a concentrations between patients with elevated blood pressure (;;:: 160/110 mm Hg) and those who had HELLP syndrome or thrombocytopenia. However, a requirement criterion for entry to our study was blood pressure ;;:: 140/90 mm Hg, so all our patients were hypertensive to some extent. The source of TNF -a is not known but could be intraamniotic macrophages, decidual macrophages, or trophoblast, because all are capable ofTNF-a synthesis. Maternal plasma TNF-a may originate from local release into the systemic circulation or from systemic production by peripheral macrophages and lymphocytes. TNF -a appears to play a physiologic role during normal pregnancy, because it can modulate trophoblast growth. This role may be part of the adaptation of the immune system to normal pregnancy and may account for the detection of TNF -a in the plasma and amniotic fluid of normal patients. Abnormal immune response may cause increased release of TNF -a, which only exerts its pathologic effects at increased levels. It is unclear whether the elevation of TNF -a in maternal blood and amniotic fluid in severe preeclampsia is an effect or a cause. However, because TNF-a is involved in the process of trophoblast invasion into maternal spiral arteries, which is impaired in preeclamptic patients long before clinical manifestations occur, TNF -a may be related to the pathogenesis of preeclampsia. We speculate that exaggeration of the normal immune response in preeclamptic patients causes the increased release of TNF-a.
Chicago, Illinois OBJECTIVE: Our purpose was to investigate whether markers for activation of the immune system are present in patients with preeclampsia by assessing maternal plasma and amniotic fluid for tumor necrosis factor-a and interleukin-1 [3. STUDY DESIGN: Twenty-one patients with severe preeclampsia composed the study group (group A). An antepartum comparison group was composed of healthy nulliparous patients not in labor and matched for gestational age (group B). Another control group consisted of term nulliparous patients in labor with uneventful pregnancies (group C). Maternal plasma samples were collected from all patients at recruitment and from patients in groups A and C immediately after delivery and again 20 to 24 hours post partum. Amniotic fluid was also collected from patients in groups A and C during labor. All samples were collectively assayed for tumor necrosis factor-a and interleukin-1 [3 by specific enzyme-linked immunoassays. RESULTS: Before labor tumor necrosis factor-a was detected more frequently in the plasma of preeclamptic patients than in the plasma of patients in group B (12/16 vs 5/16, P < 0.05) and in higher concentrations (median 35 pg/ml vs median 0 pg/ml, p < 0.05). Although tumor necrosis factor-a was frequently detected in the plasma of patients in group C in early labor (16/20), concentrations were higher in the four preeclamptic patients first sampled in early labor (210 pg/ml vs 65 pg/ml, p < 0.05). Similarly, amniotic fluid levels of tumor necrosis factor-a were increased in preeclamptic patients compared with control patients. At delivery tumor necrosis factor-a was more likely to be identified in the plasma of preeclamptic patients and was found in higher concentrations, but by 20 to 24 hours post partum measurements in the preeclamptic and control patients were similar. There were no differences in the frequency with which interleukin-1 [3 was detected or the concentration of interleukin-1 [3 in any of the samples. CONCLUSIONS: Tumor necrosis factor-a is increased in the plasma and amniotic fluid of patients with severe preeclampsia. These data are suggestive of a role for abnormal immune activation in the pathophysiologic mechanisms of preeclampsia. (AM J OBSTET GVNECOL 1994;170:1752-9.)
Key words: Preeclampsia, cytokine, tumor necrosis factor-a, interleukin-IJ3
In spite of intensive efforts to delineate the pathophysiologic mechanisms of preeclampsia, neither a specific cause nor a pathogenesis has been identified. Recent data suggest that endothelial damage, vasoconstriction, placental ischemia, and enhanced coagulation are caused by an as yet unknown serum factor produced by the trophoblast, 1. 2 and a product of the immune system could be this factor. Abnormal immune system activation by the fetal allograft has been suggested as contributing to the development of preeclampsia." but From the Section of Maternal-Fetal Medicine, Department of Obstetrics and Gynecology, Northwestern University Medical School and Northwestern Memorial Hospital. Presented at the Sixty-first Annual Meeting of The Central Association of Obstetricians and Gynecologists, White Sulphur Springs, West Virginia, October 28-30, 1993. Reprint requests: Michael}. Kupferminc, MD, 333 E. Superior St., Suite 410, Chicago, 1L 60611 Copyright © 1994 by Mosby-Year Book, Inc. 0002-9378/94 $3.00 + 0 6/6/55190
1752
no specific markers of immune activation have been identified in patients with this syndrome. Cytokines are peptide mediators of the immune system. 4 Tumor necrosis factor-a (TNF-a) is a cytokine derived from macrophages and lymphocytes and is involved with such essential biologic functions as immunoregulation, modulation of cell growth and differentiation, coagulation, and endothelial cell function. 4.7 Endothelial cell injury after TNF -a-mediated activation of the immune system may result in the secretion of vasoactive substances, an increase in vascular permeability, and intravascular coagulation,4.7 all of which are pathophysiologic changes observed in preeclampsia. Interleukin-IJ3 (IL-IJ3), another cytokine produced by macrophages, is believed to be one of the most potent regulators of prostaglandin biosynthesis by human decidua." It is secreted with TNF-a in response to inflammation and may serve as a marker for infection. 4 • 8. 9 Given the importance of prostaglandin production in
Kupferminc et al.
Volume 170, Number 6 Am J Obstet Gynecol
the pathophysiologic mechanisms of preeclampsia, it is possible to postulate a role for both IL-l J3 and TNF-o: in the development of this disease, We investigated whether activation of the immune system could be detected in patients with preeclampsia by assessing maternal plasma and amniotic fluid for the presence ofTNF-o: and IL-IJ3, Material and methods
Three groups of patients were recruited from those patients delivered at Northwestern Memorial Hospital between July 1992 and April 1993. The study cohort (group A) was composed of 21 patients with severe preeclampsia without underlying chronic hypertension or renal disease. All preeclamptic patients had persistent elevation of blood pressure ~ 140/90 mm Hg in conjunction with proteinuria. Preeclampsia was characterized as severe because of blood pressure ~ 160/110 mm Hg in 12 patients; proteinuria> 5 gm per 24 hours in three patients; platelet count < 100,000/mm2 in two patients; hemolysis, elevated liver enzymes, low platelets (HELLP syndrome) in three patients; and eclampsia in one patient. Nineteen of the preeclamptic patients were nulliparous. Seventeen patients in this group were not in labor but were delivered because of the severity of the disease; the other four patients were seen in early labor (regular uterine contractions with a cervix dilated to <4 cm). None of these pregnancies were complicated by preterm labor, premature rupture of membranes, or clinical chorioamnionitis. An antepartum comparison group (group B) consisted of healthy nulliparous patients not in labor and matched for gestational age with the nonlaboring patients in group A for which a prelabor sample was available (n = 16). None of these patients had preeclampsia or other pregnancy complications. Another control group (group C) consisted of nulliparous patients' in early labor with uneventful pregnancies at 38 to 40 weeks' gestation. The study was approved by the Northwestern University Institutional Review Board, and informed consent was obtained from each patient at the time of recruitment. Maternal venous blood samples were collected on hospital admission from patients in group A and C and during an outpatient prenatal visit for patients in group B. Additional maternal blood samples were obtained from patients in group A and C immediately after delivery and again at 20 to 24 hours post partum. Amniotic fluid specimens were collected during labor at a cervical dilatation of 2 to 4 cm from 17 preeclamptic patients and from 19 patients in group C. For patients in group A a specimen of amniotic fluid was collected from free-flowing fluid at the time of artificial rupture of membranes (n = 9), after placement of an intrauterine pressure catheter (n = 5) or by needle aspiration at
1753
cesarean section (n = 3). For patients in group C the number of samples collected by each method was 13, three, and three, respectively. In four nonlaboring preeclamptic patients amniotic fluid was collected by amniocentesis before induction of labor (n = 3) or during cesarean section for suspected fetal compromise (n = 1).
Sample processing. Blood and amniotic fluid specimens were collected in endotoxin-free tubes. The blood and amniotic fluid samples were centrifuged, respectively, at 1000g and 200g for 10 minutes. The plasma and amniotic fluid supernatants were then stored in polypropylene tubes at -70 0 C and collectively assayed for TNF -0: and IL-1 J3 by a technician blinded to patient group assignment. TNF-o: was measured by an enzyme-linked immunoassay (Human TNF-o:, Endogen, Boston, Mass.), which incorporates a double-antibody sandwich with a mouse anti-TNF-o: antibody as the capture reagent and M-301A mouse immunoglobulin conjugated to horseradish peroxidase as the detecting antibody. The assay is specific for TNF-o: and does not bind human lymphotoxin (TNF-J3), human IL-10:, human IL-2, human IL-3, human IL-6, human IL-7, human IL-8, or interferon gamma. The assay is designed to measure levels of human TNF -0: between 5 and 1715 pg/ml with interassay and intraassay coefficients of variation of 9.3% and 8.3%, respectively. We determined the minimal detectable concentration of TNF -0: in our plasma specimens to be 10 pg/ml. This amount was determined as the amount of TNF -0: that provides absorbance at two SDs above the absorbance of the diluent alone. In addition, we determined the minimal detectable concentration of TNF -0: in amniotic fluid to be 60 pg/ml by means of pooled amniotic fluid as the diluent, which was shown to be negative for TNF -0: with a standard diluent. The matrix effect of the amniotic fluid was determined to be 2 SDs above the difference between the absorbance of the diluent and the absorbance of the amniotic fluid. IL-1J3 was also measured by an enzyme-linked immunoassay (Human IL-1J3, Endogen), which incorporates a double-antibody sandwich with a monoclonal mouse anti-IL-IJ3 antibody as the capture reagent and M-301A mouse immunoglobulin conjugated to acetylcholinesterase as the detecting antibody. The assay is specific for IL-IJ3 and has a cross reactivity of 1.5% with human IL-lo:. It does not cross-react with human IL-2, human IL-3, human IL-6, human IL-7, human IL-8, human TNF -0:, or human interferon gamma. The assay is designed to measure levels of human IL-l J3 between 4 and 540 pg/ml, with interassay and intraassay coefficients of variation of 7.8% and 6.3%, respectively. The minimal detectable concentration of IL-1 J3 in maternal plasma was determined to be 10 pg/ml, and the mini-
1754
June 1994 Am J Obstet Gynecol
Kupferminc et al.
Table I. Maternal clinical characteristics
Maternal age (yr) Gestational age (wk) Race White Black Hispanic Parity 0 ~1
Vaginal delivery Cesarean delivery Length of labor (hr)t
Group A (n = 21)
Group B (n = 16)
Group C (n = 20)
26.6 ± 9.5 34.3 ± 4.5
28.6 ± 3.8 34.6 ± 4.4
27.9 ± 4.7 39.2 ± 1.1*
7 12 2
7 9 0
6 14 0
19 2 17 4
16 0 NA NA NA
20 0 17 3
11.4 ± 6.7
10.6 ± 3.3
Group A, Preeclamptic patients; group B, antepartum control group; group C, intrapartum control group. NA, Not available. *p < 0.0001 compared with group A. tVaginal delivery only.
mal detectable concentration in amniotic fluid was determined to be 40 pg/ml. These minimal concentrations were determined as described for the TNF-a assay. Statistical analysis. All TNF-a and IL-I ~ results are presented as medians with ranges. Categoric data were analyzed with 1(2 or Fisher's exact test. Continuous data were compared with the Mann-Whitney U test or Student t test. Statistical significance was assumed at a p of <0.05. Results
The maternal clinical characeristics of the three groups of patients are presented in Table I. No significant differences were noted between groups A and B in maternal age, gestational age, race, and parity. Groups A and C differed only in mean gestational age. The length of labor, defined by the interval from the onset of regular uterine contractions until delivery, was similar among those patients delivered vaginally. Maternal plasma concentrations ofTNF-a at recruitment in the preeclamptic patients and comparison groups are displayed in Fig. I. Sixteen plasma samples were obtaine'd from the 17 preeclamptic patients not in labor. TNF -a was detected more frequently in the preeclamptic patients than in the antepartum control patients in group B (12/16 vs 5/16, P < 0.05), and the concentrations of TNF-a were significantly higher (35 pg/ml [0 to 160 pg/ml] vs 0 pg/ml [0 to 60 pg/ml], p < 0.05). TNF-a levels within groups A and B were not correlated with gestational age. The frequency of detection and plasma concentrations of TNF-a were higher in the term control patients in labor (group C) compared with patients in group B (16/20 vs 5/16, P < 0.01; 65 pg/ml versus 0 pg/ml, p < 0.01). However, when the plasma levels of TNF -a from the four preeclamptic patients in labor were compared with those of patients in group C, TNF-a levels were significantly
increased (210 pg/ml [130 to 350 pg/ml] vs 65 pg/ml [0 to 280 pg/ml]; p < 0.05). Maternal plasma TNF-a concentrations immediately after delivery for the preeclamptic patients and patients in group C are depicted in Fig. 2. For those parturients delivered vaginally TNF-a was detected more frequently (15/17 vs 7/16, P < 0.01) and at higher concentrations (130 pg/ml [0 to ll20 pg/ml] vs 0 pg/ml [0 to 490 pg/ml], p < 0.01) in the preeclamptic patients in spite of a similar length of labor. If data from patients delivered by cesarean section (group A: failed attempt at induction of labor [3], suspected fetal compromise [1]; group C: breech presentation [3]) are included, the detection rates (17/20 vs 7/19, P < 0.01) and concentrations ofTNF-a (80 pg/ml vs 0 pg/ml, p < 0.01) were ! still significantly different between the two groups. By 20 to 24 hours post partum the frequency of TNF-a detection (16/21 vs 16/18) and the TNF-a concentrations (30 pglml [0 to 140 pglml] vs 50 pglml [0 to 140 pg/ml]) in the preeclamptic patients and patients in control group C were similar. Amniotic fluid. As shown in Fig. 3, TNF-a was detected more frequently in the amniotic fluid samples obtained from preeclamptic patients during labor than in laboring control patients (17/17 vs 12/19, P < 0.01), and TNF-a concentrations were significantly higher in the preeclamptic patients (310 pg/ml [60 to 490 pg/ml] vs 60 pg/ml [0 to 250 pg/ml], p < 0.0001). The amniotic fluid TNF-a concentrations collected from the four preeclamptic patients not in labor were also significantly elevated (270 pg/ml [180 to 350 pglml]). Within each group of patients the frequency of detection and TNF-a concentrations were similar in free-flowing amniotic fluid compared with amniotic fluid collected by other means. Maternal plasma and amniotic fluid IL-lJi.Plasma IL-l ~ was detected on recruitment in only one patient
Kupferminc et al. 1755
Volume 170, Number 6 Am J Obstet Gynecol
375
1200
•
350 325
1100 1000
300
•• • •
275 250
~
E
CD
B a LL
z
I-
• •
225 200 175
o o o
150 125 100 50 25 0
•
o o o
75
-+ • •
000
co
..i8000
·...0006·. ·...···. ···...·. ·············......~ ...···. ·..~·.. n
Group A
= 16
n
=4
Group B
n
= 16
Group C
n
= 20
Fig. 1. Plasma TNF-a detection rates and concentrations on recruitment were higher in nonlaboring preeclamptic patients (group A) than in antepartum control patients (group B). TNF -a levels were also increased in preeclamptic patients in early labor compared with control patients in early labor (group C). Open circles, Patients not in labor; closed circles, patients in early labor; horizontal bars, median concentrations in each group.
in each of the three groups. Plasma IL-lj3 was detected at delivery in only three of 20 preeclamptic patients and four of 19 patients in group C. IL-lj3 was not detected in any of the plasma samples collected 20 to 24 hours post partum. The frequency of IL-lj3 detection in amniotic fluid obtained in labor and the median concentrations of IL-lj3 were also comparable between patients in group A and group C (7/17 vs 9/19; 0 pg/ml [0 to 660 pg/ml] vs 0 pg/ml [0 to 660 pglml]). Comment
Our data indicate that TNF -a is detected more frequently and at higher concentrations in the plasma of patients with severe preeclampsia than in patients with uncomplicated pregnancies. The increase in plasma TNF -a noted before labor was also found to be independent of gestational age. Although labor appeared to increase the detection rate and concentration of TNF-Ci. for control patients, further elevations in TNF-a concentrations were seen in preeclamptic parturients. Additionally, amniotic fluid levels of TNF-Ci. in early labor were also significantly elevated in patients with preeclampsia. Plasma TNF-Ci. concentrations remained el-
.
900
800
-E ~
01
B
a
LL
z
I-
700 600 500 400
300 200
• •
.....
• •
• • o ......··_······. ······rA/ItI········. . · ·. ·••·••.a:·.989········· . ·. . . . .
100
Group A n = 20
Group C n = 19
Fig. 2. Plasma TNF-a detection rates and concentrations at delivery were higher in preeclamptic patients (group A) than in control patients (group C). Closed circles, Vaginal deliveries; open circles, cesarean births; horizontal bars, median concentrations in each group.
evated in preeclamptic patients at delivery, but by 20 to 24 hours post partum the plasma TNF -Ci. measurements were comparable to those of the control patients. The increase in TNF -Ci. concentrations in the postpartum control patients compared with delivery values may be a secondary effect of the reparative process at the placental site. In contrast to TNF-Ci., no increase in IL-lj3 was detected in plasma or amniotic fluid samples from preeclamptic patients. These data support a role for TNF -Ci., but not IL-lj3, in the pathogenesis of preeclampsia. TNF -Ci. was not consistently found to be either present or absent in all samples from a given patient. This observation suggests a complex or nonlinear mechanism of TNF -a secretion. Because the half-life of TNF-Ci. is only a few minutes in the circulation, 10 a single blood sample may fail to detect a periodic elevation or decline in TNF -Ci. levels. Further, an unknown portion ofTNF-Ci. may be bound to its soluble receptor and may not be detected by the bioreactive or immunoreactive assays. 10 TNF -Ci. and IL-lj3 are not considered normal constituents of amniotic fluid, but both have been detected in amniotic fluid during parturition. II Intraamniotic macrophages, decidual macrophages, and trophoblast
1756 Kupferminc et aI.
••
500
450
•••
400
8
350
~
a a
Ii.
zt-
June 1994
Am J Obstet Gynecol
300
--•
• • 200 • 8 •• 150 • • 100 -_. _-_. _. . . . ._. _. . . . . _. . . _. . ;_. . _. . _. _. . . _. _. . . . . -A-.._. . . .. 50 250
..
0 n
=
Group A 4 n 17
=
••••••• Group C n = 19
Fig. 3. Amniotic fluid TNF-a detection rates and concentrations were higher in preeclamptic patients (group A) than in control patients (group C). OPen circles, patients not in labor; closed circles, patients in early labor; horizontal bars, median concentrations in each group.
are all possible sources, because all are capable of cytokine synthesis. 9. 11·1' Among women with negative amniotic fluid cultures, TNF-a was detected in the amniotic fluid of 20% of patients in labor at term and in 18.7% of patients in pre term labor. II In comparable groups of patients IL-l~ was detected in the amniotic fluid of 61.5% and in 47.6% of patients, respectively.9 TNF-a and IL-l ~ are present in higher concentrations in patients with preterm labor and microbial invasion of the amniotic fluid cavity,9. II and it has been postulated that these cytokines play a role in the pathophysiologic mechanisms of preterm labor in this setting. 8. 9. II. 14 Although we did not perform cultures of the amniotic fluid, increased TNF-a levels in the amniotic fluid secondary to infection is unlikely, because detection rates and levels of IL-l ~ were similar to those reported for patients with preterm labor and lower than those for term parturients with negative cultures of the amniotic fluid. 9 The increased detection of TNF-a in our amniotic fluid samples may be related to the underlying disease process or to the stage of labor at collection. Immunologic stimuli occurring as a part of the adaptation of the immune system to pregnancy may account for the detection of TNF-a in plasma and amniotic fluid in some normal pregnant patients. On
the other hand, the increased detection rate and concentrations of TNF-a in preeclamptic patients may follow abnormal activation of the immune system. Indeed, one postulated mechanism for the development of preeclampsia involves abnormal activation of the immunologic system against the fetal allograft. 3 Alloantigenic fetal tissue may be exposed to the maternal immune system as a result of impaired immunologic suppression, causing a maternal immune response similar to graft rejection, manifested in part by activation of macrophages and lymphocytes with secretion ofTNF-a. Consistent with this premise is the observation that TNF-a has been shown to be an important mediator in transplantation rejection. 15 An altered immune response to the fetal allograft may also play a role in the impaired trophoblastic invasion into maternal spiral arterioles that has been reported in preeclampsia by 16 to 20 weeks' gestation. I Recent observations suggest that TNF-a is capable of limiting trophoblast invasion in the uterus, as well as inhibiting deoxyribonucleic acid synthesis and trophoblastic cellular proliferation in vitro!6. 17 Furthermore, TNF-a is also increased in midtrimester amniotic fluid of pregnancies resulting in fetal growth retardation,18 which has similar pathophysiologic characteristics to preeclampsia. Involvement of the cellular branch of the immune system and its products, the cytokines, in preeclampsia remains poorly understood, but the presence of and biologic capabilities of TNF-a strongly suggests that a role in the pathogenesis of preeclampsia is plausible. TNF-a is involved in many pathologic processes such as infection, tumor cytotoxicity, and endotoxic shock. 5-7 TNF-a may cause tissue damage either directly through activation of protease, collagenase, or phospholipase A2 enzymes or through oxygen radicals. 5 Direct damage to vascular endothelial cells, reduced regional blood flow, occlusion of vessels, and increased endothelial permeability are characteristic effects of pathologically secreted TNF_a. 5-7. 16 TNF-a also enhances procoagulant activity, by inducing endothelial cell tissue factor, and suppresses protein C activation, an antithrombotic mechanism that functions on the surface of quiescent endothelial cells. 5. 7 Finally, TNF-a increases the release of vasoactive substances such as endothelin,19 which is increased in preeclamptic patients,20 and platelet-derived growth factor, a potent constrictor of vascular smooth muscle 21 that is released from endothelial cells after exposure to serum from preeclamptic patients in vitro. 22 Our preliminary data support the involvement of abnormal immune activation in the pathogenesis of preeclampsia. TNF-a may play an active role in mediating the endothelial damage, ischemia, and enhanced coagulation seen in preeclampsia. The low detection rates and concentrations of IL-l~ imply different pat-
Kupferminc et al.
Volume 170, Number 6 Am J Obstet Gynecol
terns of secretion for the two cytokines in the setting of preeclampsia, as opposed to contemporary elevation in response to infection or inflammation. It remains to be determined whether increased TNF-a levels are present in patients with mild preeclampsia and are detectable before clinical manifestations appear. REFERENCES
1. Roberts ]M, Taylor RN, Musici T], et al. Preeclampsia: an endothelial cell disorder. AM] OBSTET GYNECOL 1989; 161: 1200-4. 2. Rodgers GM, Taylor RN, Roberts ]M. Preeclampsia is associated with a serum factor that is cytotoxic to human endothelial cells. AM ] OBSTET GYNECOL 1988;159: 908-14. 3. Sibai BM. Immunologic aspects of preeclampsia. Clin Obstet Gynecol 1991;34:27-34. 4. Ari K, Lee F, Miyajima A, Miyatake S, Ari N, Yokota T. Cytokines: coordinators of immune and inflammatory responses. Annu Rev Biochem 1990;59:783-836. 5. Le], Vilcek J. Biology of disease. Tumor necrosis factor and interleukin-l: cytokines with multiple overlapping biological activities. Lab Invest 1987;56:234-48. 6. Beutler B, Cerami A. Ibe biology of cachectin{rNF-a primary mediator of the host response. Annu Rev Immunol 1989;7:625-55. 7. Beutler B, Cerami A. Cachectin: more than a tumor necrosis factor. N Engl] Med 1987;316:379-85. 8. Romero R, Brody DT, Oyarzun E, et al. Infection and labor. III. Interleukin-l: a sign for the onset of parturition. AM] OBSTET GVNECOL 1989;160:1117-23. 9. Romero R, Mazor M, Brandt F, et al. Interleukin-la and interleukin-lj3 in preterm and term human parturition. Am] Reprod Immunol 1992;27:117-23. 10. Dinarello CA, Gelfand ]A, Wolf SM. Anticytokine strategies in the treatment of the systemic inflammatory response syndrome. ]AMA 1993;269: 1829-35. 11. Romero R, Mazor M, Sepulveda W, Avila C, Copeland D, Williams ]. Tumor necrosis factor in pre term and term labor. AM] OBSTET GYNECOL 1992;166:1576-87. 12. Casey ML, Cox SM, Beutler B, Milewich L, Macdonald P. Cachectin/tumor necrosis factor-a formation in human decidua. Potential rule of cytokines in infection induced preterm labor.] Clin Invest 1989;83:430-6. 13. Vince G, Shorter S, Starkey P, et al. Localization of tumor necrosis factor production in cells at the materno/fetal interface in human pregnancy. Clin Exp Immunol 1992; 88: 174-80. 14. Romero R, Manogue KR, Mitchell MD, et al. Infection and labor. IV. Cachectin-tumor necrosis factor in the amniotic fluid of women with intraamniotic infection and preterm labor. AM] OBSTET GYNECOL 1989;161:336-41. 15. Maury CP], Teppo AM. Raised levels of cachectinltumor necrosis factor-a in renal allograft rejection. ] Exp Med 1987; 166: 1132-7. 16. Hunt ]S. Cytokines network in the utero-placental unit. Macrophage as pivotal regulatory cells.] Reprod ImmunoI1989;16:1-25. 17. Hunt]S, Soares MJ, Lei MG, et al. Products oflipopolysaccharide-activated macrophages (tumor necrosis factor-a, transforming growth factor-j3) but not lipopolysaccharide modity DNA synthesis by rat trophoblast cells exhibiting the 80-kDa lipopolysaccharide-binding protein. ] ImmunoI1989;143:1606-13. 18. Heyborne KD, Eitkin SS, McGregor ]A. Tumor necrosis factor-a in midtrimester amniotic fluid is associated with impaired intrauterine fetal growth. AM ] OBSTFT GYNECOL 1992;167:920-5. 19. Vemulapalli S, Chiu P]S, Rivelli M, Foster CJ, Sybertz EJ. Modulation of circulating endothelin levels in hyperten-
1757
sion and endotoxemia in rats. ] Cardiovasc Pharmacol 1991; 18:895-902. 20. Nova A, Sibai BM, Barton ]R, Mercer BM, Mitchell MD. Maternal plasma levels of endothelin is increased in preeclampsia. AM] OBSTET GYNECOL 1991;165:724-7. 21. Hajjar KA, Hajjar DP, Silverstein RL, Nachman RL. Tumor necrosis factor mediated release of platelet-derived growth factor from cultured endothelial cells.] Exp Med 1987; 166:235-45. 22. Taylor RN, Musci T], Rodgers GM, Roberts ]M. Preeclamptic sera stimulate increased platelet-derived growth factor mRNA and protein expression by cultured human endothelial cells. Am] Repron Immllnol 1991 :2'i: 1Wi-f<.
Discussion New Orleans, Louisiana. TNF-a is a 17 kd soluble mediator of cellular immunity. This cytokine is produced by macro phages, T cells, thymocytes, B cells, and natural killer cells in response to a wide variety of stimuli, including infectious agents. TNF -a may also be produced by human decidua in response to bacterial products, and it stimulates prostaglandin production by amnion and decidua. TNF-a and other cytokines appear to participate in the development of labor in patients with microbial invasion of the amniotic cavity.l. 2 For nearly a century investigators have believed in an immune cause for preeclampsia. This study looks at the role of TNF-a as an immune system mediator in the pathophysiologic mechanisms of preeclampsia. The authors describe a study group of severely preeclamptic patients (group A). They construct two control groups - one matched for gestational age but not in labor (group B) and the other selected from laboring patients at term (group C). Neither control group included patients with preeclampsia or clinically identified infection. Plasma and amniotic fluid TNF -a concentrations were determined as described. Comparison by nonparametric techniques was performed. Summarized results suggest that nonlaboring women with severe preeclampsia have higher plasma TNF -a values than do nonlaboring nonpreeclamptic women at the same gestational age. In early labor severe preeclamptics have higher TNF-a values than do laboring control patients at term. At delivery plasma TNF -a was still significantly higher in preeclamptics. By 20 to 24 hours post partum TNF -a values were similarly elevated for both groups (A and C). Amniotic fluid analysis revealed that TNF-a was present in higher concentrations among laboring severely preeclamptic women than among laboring nonpreeclamptic women at term. Significant results for IL-l ~ were absent. From the study presented we observed that (1) elevated plasma TNF-a levels are associated with severe preeclampsia and (2) elevated plasma TNF -a levels are associated with labor. Additionally, we observe that amniotic fluid TNF -a is more often identified in laboring severely preeclamptic patients than in controls. We conclude that TNF-a may be involved in the manifestation of severe preeclampsia. However, it is not shown to be causative. Elevated amniotic fluid TNF-a DR. JOSEPH M. MILLER, JR.,
1758
Kupferminc et aJ.
levels have been identified in term patients in labor without preeclampsia (this study) or infection.' Additionally, plasma TNF -Ci is frequently found in normal term laboring patients. By 20 to 24 hours post partum normal and severely preeclamptic patients have similar values. If TNF -Ci was causally associated, one would expect preeclampsia to be observed. Perhaps TNF -Ci is a reactor to rather than an initiator of preeclampsia. The authors call for additional study of mild preeclamptics. Longitudinal study would also be indicated. I have two concerns about the assay. First, among uncomplicated laboring patients at term the authors found 63% with detectable amniotic fluid TNF-Ci as opposed to 20% reported by Romero et al.' In nonlaboring patients 31 % of normal and 75% of severely preeclamptic patients had detectable plasma TNF-Ci levels. Although statistical significance exists, why do many normal patients have identifiable TNF -Ci levels without evidence of infection or preeclampsia? Second, infection is associated with the presence of cytokines in amniotic fluid. Although standard infectious morbidity was absent, a negative amniotic fluid culture would add to the strength of the observations. REFERENCES 1. Romero R, Moshe M, Sepulveda W, et al. Tumor necrosis factor in preterm and term labor. AM ] OBSTET GYNECOL 1992; 166: 1576-87. 2. Romero R, Mazor M, Brandt F, et al. Interleukin-la and interleukin-l~ in preterm and term human parturition. Am ] Reprod ImmunoI1992;27:177.
Galveston, Texas. Elevated TNF -Ci in the plasma of women with severe preeclampsia is potentially a very important finding. Elevated levels could explain why these women have high blood pressure. TNF -Ci has (1) an inhibiting effect on nitric oxide' and (2) stimulatory action on endothelin-12 and prostaglandins. 3 Any of these effects could result in elevated blood pressure. What do we know about this cytokine, which is derived from macrophages and leukocytes during pregnancy? TNF-Ci is elevated in the amniotic fluid of women with intraamniotic infection. Although the amniotic fluid was not cultured in this study, intraamniotic infection is unlikely to have played a role in the elevation of the TNF -Ci in the study population, because all these groups had intact membranes at the start of the study. These data strongly suggest that TNF -Ci is elevated in the plasma of both the severely preeclamptic and control women who were in labor (Fig. 1 in article). The amniotic fluid TNF -Ci levels were higher in the severely preeclamptic women than in the control women in labor. No comparisons were made between laboring and nonlaboring women with regard to TNF -Ci levels in amniotic fluid. These data are at odds with Opsjon et al.: who reported a threefold rise in tumor necrosis factor in the amniotic fluid of women in spontaneous labor. They did not find any elevation of tumor necrosis factor in the plasma of women in labor, and in only one of 27 DR. GARLAND D. ANDERSON,
June 1994 Am J Obstet Gynecol
women was it detected in the plasma of women in labor. Would you comment on the different findings of the two studies? It is not known what effect conical effacement of cervical dilatation before labor has on tumor necrosis factor levels. Because cervical effacement and dilatation were not controlled in this study, this may affect the findings. The small sample size, lack of control for cervical dilatation, and inconsistent sampling, particularly with regard to amniotic fluid, leaves the findings of elevated tumor necrosis factor levels in the plasma and amniotic fluid of women with severe preeclampsia suspect. Because some women with severe preeclampsia have normal TNF-Ci levels and some normal women have elevated TNF -Ci levels, preeclampsia occurs independent ofTNF-Ci. IfTNF-Ci is elevated as a result ofpreeclampsia, what is the source of the elevated levels? REFERENCES 1. Yoshizumi M, Perrella M, Burnett]C]r, Lee M-E. Circ Res 1993;73:205-9. 2. Greenberg S, ]ianming X, Wang YE, et al. Tumor necrosis factor-alpha inhibits endothelium-dependent relaxation. ] Appl Physiol 1993;74:2394-403. 3. Romero R, Manogue KR, MitcheUMD, et al. Infection and labor. IV. Cachectin-tumor necrosis factor in the amniotic fluid of women with intraamniotic infection and preterm labor. AM] OBSTET GYNECOL 1989;161:336-41. 4. Opsj(lln SL, Wathen NC, Tingulstad S, et al. Tumor necrosis factor, interleukin-l, and interleukin-6 in normal human pregnancy. AM] OBSTET GYNECOL 1993;169:397-404. DR. ROBERT C. KAUFMAN, Springfield, Illinois. I agree with the comment that this may be a reactive elevation of tumor necrosis factor. Did the authors try to relate the tumor necrosis factor level to the size of the baby, which may be an indication of placental destruction? DR. JAMES N. MARTIN, JR., Jackson, Mississippi. The authors have 21 patients with preeclampsia, and it appears to me that they may be "lumpers" rather than "splitters." I noted that three of the women had HELLP syndrome and two had thrombocytopenia. Because investigators from Chicago (Gleicher's group) have suggested that HELLP syndrome might be an immunologic variant of preeclampsia, I wonder whether the authors have examined the subset of the data for these five patients apart from the larger group of 16 women with preeclampsia but without thrombocytopenia or HELLP syndrome. If the presumption is that this elevated factor is related to increased vascular permeability, could they also relate the clinical findings of the patients with the highest levels of this factor and see any association at all? DR. STEPHEN J. FORTUNATO, New Orleans, Louisiana. I would ask whether the authors have stratified the level of TNF that they've seen for birth weight. A recent publication has shown that at 16 weeks there is an increased level of tumor necrosis factor in those babies destined to be small for gestational age. We've also documented in our laboratory that TNF -Ci is produced both at the message level and at the peptide level from
Volume 170, Number 6 Am J Obstet Gynecol
the amnion and the chorion. This might be another source of TNF. DR. KUPFERMINC (Closing). We found a detection rate of 63% for TNF -a in the amniotic fluid of normal parturients at 2 to 4 cm dilatation, with a median concentration of 60 pg/ml; TNF -a was detected in all preeclamptic patients with a median concentration of 310 pg/ml. It was previously reported by Romero et al. that 20% of normal parturients had detectable TNF-a in the amniotic fluid, whereas OpsjlOn et al. with a bioassay reported a detection rate of 100% and a median concentration of 58 pg/ml in the amniotic fluid of normal parturients. Possible explanations for these differences are as follows: (1) Immunoassays and bioassays may have different sensitivities for TNF-a detection; (2) substantial amounts of TNF-a may be bound to naturally occurring inhibitors or soluble TNF receptors, interfering with detection by some assays; (3) collection of amniotic fluid at different stages of labor may result in different detection rates. Variation between our data and those previously reported by OpsjlOn et al. for plasma values during labor may also be related to these same explanations. In addition, because the half-life of TNF -a in the circulation is only a few minutes, a single blood sample may fail to detect a periodic elevation or a decline in TNF-a. The manufacturer of the assay we performed states that this assay measures both free and bound TNF-a. As further support for the validity of our measurements of TNF -a in the plasma of pregnant patients, we also tested for TNF -a in plasma of 20 nonpregnant reproductive-age patients, where TNF-a is not ordinarily detectable. TNF -a was not detected in any of these patients. Although we cannot ensure that the elevation of TNF -a was not related to infection, none of the patients was infected clinically. It is important to note that IL-I13, which is normally increased compared with TNF -a in response to infection or inflammation, was not increased in the preeclamptic patients compared with control patients. This also argues against the
Kupferminc et al.
1759
possibility of infection as the cause of increased TNF-a levels in these patients. As for a correlation between birth weight and TNF-a levels, we did not find any. Furthermore, the five preeclamptic patients who were delivered of small-for-gestational-age newborns had plasma and amniotic fluid TNF-a concentrations similar to those of preeclamptic patients with appropriate-for-gestational-age newborns. In addition, we did not find any difference in TNF-a concentrations between patients with elevated blood pressure (;;:: 160/110 mm Hg) and those who had HELLP syndrome or thrombocytopenia. However, a requirement criterion for entry to our study was blood pressure ;;:: 140/90 mm Hg, so all our patients were hypertensive to some extent. The source of TNF -a is not known but could be intraamniotic macrophages, decidual macrophages, or trophoblast, because all are capable ofTNF-a synthesis. Maternal plasma TNF-a may originate from local release into the systemic circulation or from systemic production by peripheral macrophages and lymphocytes. TNF -a appears to play a physiologic role during normal pregnancy, because it can modulate trophoblast growth. This role may be part of the adaptation of the immune system to normal pregnancy and may account for the detection of TNF -a in the plasma and amniotic fluid of normal patients. Abnormal immune response may cause increased release of TNF -a, which only exerts its pathologic effects at increased levels. It is unclear whether the elevation of TNF -a in maternal blood and amniotic fluid in severe preeclampsia is an effect or a cause. However, because TNF-a is involved in the process of trophoblast invasion into maternal spiral arteries, which is impaired in preeclamptic patients long before clinical manifestations occur, TNF -a may be related to the pathogenesis of preeclampsia. We speculate that exaggeration of the normal immune response in preeclamptic patients causes the increased release of TNF-a.