ORIGINAL ARTICLE: Treatment with Tumor Necrosis Factor Inhibitors and Intravenous Immunoglobulin Improves Live Birth Rates in Women with Recurrent Spontaneous Abortion

Authors


Edward E. Winger, Alan E. Beer Center for Reproductive Immunology & Genetics, Suite 2106, 611 Washington Street, San Francisco, CA 94111, USA.
E-mail: ewinger@sbcglobal.net

Abstract

Problem  The purpose of this study was to investigate whether treatment with tumor necrosis factor (TNF) inhibitors combined with intravenous immunoglobulin (IVIG) increases live birth rates among women with recurrent spontaneous abortion (RSA) concurrently treated with anticoagulants (AC).

Method of study  Seventy-five pregnancies in patients with a history of RSA were retrospectively evaluated. The population was divided into three groups: group I: 21 patients treated with AC (anticoagulants), group II: 37 patients treated with AC and IVIG, and group III: 17 patients treated with AC, IVIG and the TNF inhibitor Etanercept (Enbrel®) or Adalimumab (Humira®). In groups II and III, IVIG was administered at least once during the cycle of conception and/or at least once after a positive pregnancy test. In group III, either Adalimumab or Etanercept was administered by subcutaneous injection according to standard protocols. Statistical analysis of pregnancy outcome was performed using Fisher’s exact test.

Results  Patient populations in the three treatment groups were similar in terms of age, past miscarriages, inherited thrombophilia and autoimmunity. The live birth rate was 19% (4/21) in group I, 54% (20/37) in group II, and 71% (12/17) in group III. There was significant improvement in pregnancy outcome in group II versus group I (P = 0.0127) and in group III versus group I (P = 0.0026). The live birth rate in group III compared to group II was not significantly different (P = 0.3723). Side effects of AC, IVIG and TNF inhibitor treatment were minimal in these patients, and no birth defects were identified in their offspring.

Conclusion  In women with RSA, addition of either IVIG or a TNF inhibitor + IVIG to the AC regimen appears to improve live birth rates compared to the treatment with AC alone. The positive effect of IVIG and TNF inhibitor therapy on pregnancy outcome merits further study in prospective clinical trials.

Introduction

Tumor necrosis factor-α (TNF-α) has been hypothesized to be important in causing fetal loss in chromosomally normal pregnancies.1 Th1 cytokines, interferon-γ (IFN-γ), TNF-α and interleukin-2 (IL-2) are known to increase fetal resorption in mouse models.2 However, the mechanism whereby the implanted embryo is killed in an environment dominated by Th1 cytokines remains unresolved. It is quite possible that death is mediated in part by ischemic events triggered by thrombotic/inflammatory processes within materno–fetal blood vessels. In mice TNF-α is thought to up-regulate a fibrinogen-related prothrombinase, fgl2. Thrombin generated through this pathway has several consequences which compromise the placental circulation through multiple mechanisms. Thrombin activation not only leads to fibrin deposition but also complement C5 and neutrophil activation each of which lead to vascular compromise.3 Clark demonstrated cytokine-boosted abortions in abortion-prone CBA × DBA/2 mice were blocked by antibody to fgl2 prothombinase expressed by cytokine-stimulated vascular endothelial cells and monocytes.4 Del Prete found that Th2 cytokines inhibited Th1-induced tissue factor production by monocytes.5

In addition to changes at the materno–fetal interface, immunologic changes in the systemic immune system are also present. These immunologic changes are characterized by shifts in peripheral blood lymphocyte populations and function, resulting in a decrease in the proportion of cells that produce pro-inflammatory cytokines such as TNFα (Th1 cells) to cells that produce anti-inflammatory cytokines such as IL-10 (Th-2 cells). Raghupathy observed that the Th2 bias exhibited in pregnant women with normal pregnancy was significantly greater than that observed in women with unexplained recurrent pregnancy losses suggesting yet another mechanism moderating immunologic responses to the fetus. He observed that elevated serum levels of Th2 cytokines such as IL-10 were observed in normal pregnancy while significantly higher serum levels of the Th1 cytokine, IFN-γ, were found with unexplained recurrent pregnancy loss. These results suggest that normal pregnancy is associated with a Th2 bias, while women with a history of recurrent pregnancy losses are associated with a Th1 bias.6

Tumor necrosis factor-α was amongst a number of pro-inflammatory cytokines, including IL-1, IL-6, granulocyte–macrophage colony-stimulating factor, and IL-8, considered as therapeutic targets in rheumatoid arthritis. In the early 1990s, the prevailing view was that they, together, comprised sufficient redundancy to drive the inflammatory process when any single pro-inflammatory mediator was blocked in isolation. However, TNF-α appears to occupy a crucial signaling position in the inflammatory cascade. Drugs capable of neutralizing this single cytokine are well recognized as effective therapeutic agents in clinical practice. Three TNF-α agents are currently licensed in the United States. These include Remicade® (infliximab), Humira®, (adalimumab) and Enbrel® (etanercept). The first two are monoclonal antibodies directed against TNF-α while the third is a soluble form of the receptor. These drugs differ in their pharmacokinetics. For example, the monoclonal antibodies have significantly longer half-lives making them more convenient. Licensed clinical indications are currently limited to rheumatoid arthritis, however, dermatologic conditions including psoriasis are known to be responsive to these therapies as are inflammatory bowel diseases (Crohn’s disease), and ankylosing spondylitis.7

Concerns about their use, however, relate in part to an increased risk of infectious disease, in particular tuberculosis.8 Infections have been reported including histoplasmosis, and other viral infections caused by herpes zoster, varicella, and cytomegalovirus.9,10 An increased incidence of lymphoma in patients with rheumatoid arthritis treated with anti-TNF-α agents has been noted. However, this increased incidence is believed to be related to the increased incidence in the underlying condition.9 Possible teratogenic effects have been a consideration for use in pregnant women. Embryo-fetal perinatal developmental toxicity studies performed in cynomolgus monkeys using doses several hundred times the recommended human dose have revealed no evidence of teratogenic or other deleterious effect.9 Moreover, a relatively large complement of women incidentally exposed to anti-TNF-α agents have provided significant evidence for the safe use of these agents in pregnancy. No increased risk of malformations has been demonstrated.11,12

Unexplained recurrent pregnancy loss is a vexing clinical problem facing a significant proportion of couples wishing to have children A variety of therapeutic modalities are currently being offered to help prevent this distressing condition. In this study three therapies are utilized; anti-coagulation, intravenous immunoglobulin (IVIG) and anti-TNFα agents.13–15 Treatment with heparin and low-dose aspirin have been shown to reduce the incidence of recurrent miscarriage in patients with both the acquired and inherited forms of thrombophilia possibly through dual protective pathways: inhibition of complement leading to decreased TNF production and increased vascular protection resulting in micro-clot prevention.16 In patients selected for poor prognosis and autoimmunity, IVIG has been shown to be effective.17 IVIG may protect the embryo through several mechanisms: suppression of activated natural killer (NK) cells, deactivation of T cells and polyclonal B cells, control of harmful antibodies through the action of anti-idiotype antibodies and shift from a Th1 to Th2 bias.18 In addition, intravenous gamma globulin may provide a protective CD200 signal, an important defence molecule that may play a variety of roles including promoting activation of T-regulatory cells which are often deficient inpatients prone to pregnancy losses.19–21 Combined with the action of these more commonly used therapies; the addition of anti-TNF-α medication may tip the balance in a positive direction for many patients who have failed with simpler regimens.

This study retrospectively analyzes patients who presented with histories of recurrent spontaneous abortions and who were treated with various combinations of anticoagulation, IVIG and anti-TNF-α agents. The analysis sought to determine whether the addition of anti-TNF-α agents to the standard IVIG and anticoagulant protocols could increase the success rates over IVIG and anticoagulant protocols used alone.

Materials and methods

Seventy-five pregnancies in patients with a history of recurrent spontaneous abortion (RSA; 3 ≥ miscarriages) were retrospectively evaluated. The population was divided into three groups: group I: 21 patients treated with anticoagulants (AC), group II: 37 patients treated with AC and IVIG, and group III: 17 patients treated with AC, IVIG and the TNF-α inhibitor Etanercept (Enbrel®; Amgen, Thousand Oaks, CA, USA) or Adalimumab (Humira®; Abbot Laboratories, North Chicago, IL, USA) (Tables I, II and III). Patient groups were formed in part through self-selection and in part through differences in initial work-up and history. Patients with an NK Assay 50:1 cytotoxicity level over 15% and/or CD56 number over 12% drawn either before conception or after a positive pregnancy test were offered IVIG therapy, however many of these patients still refused this treatment due to cost or safety concerns. Patient refusal to the use of IVIG created much of the heparin-only treatment group. Other patients with a more difficult history i.e. a repeat (≥3) IVF failure in addition to recurrent (≥3) miscarriage, and patients with an elevated TNF ratio over 30 and /or positive endometrial biopsy result with ≥ 4 CD57 cells/hpf were offered the more aggressive TNF-α therapy protocol. However, again, not all patients who were offered this TNFα therapy actually used the treatment, resulting in more equal patient stratification between the three treatment groups.

Table I.   Group I (Hep) Patient Characteristics, Treatments and Outcomes
Patient historyTreatmentsOutcomes
PatientAgeNo. of lossesHeparinPregnancyLoss (weeks)Delivery (weeks)Weight (g)
 1419HeparinMiscarriage8  
 2244LovenoxDelivered 393430
 3324LovenoxMiscarriage8  
 4363HeparinDelivered 383686
 5357LovenoxMiscarriage9  
 6393LovenoxDelivered 404451
 7453ClexaneMiscarriage4  
 8403HeparinDelivered 383204
 9363LovenoxMiscarriage8  
10353HeparinMiscarriage8  
11353HeparinMiscarriage4  
12345LovenoxMiscarriage8  
13473LovenoxMiscarriage7  
14433HeparinMiscarriage10  
15405LovenoxMiscarriage8  
16407LovenoxMiscarriage8  
173210HeparinMiscarriage5  
18303LovenoxMiscarriage8  
19375LovenoxMiscarriage4  
20354LovenoxMiscarriage5  
21366FragminMiscarriage8  
Table II.   Group II (Hep + IVIG) Patient Group Characteristics, Treatments and Outcomes
Patient historyTreatmentsOutcome
PatientAgeNo. of losses (total)IVIGHeparinPregnancyLoss (weeks)Delivery (weeks)Weight (g)
 1374IVIGLovenoxMiscarriage8  
 2364IVIGLovenoxDelivered 393572
 3356IVIGLovenoxMiscarriage9  
 4426IVIGClexaneDelivered 402948
 5343IVIGHeparinDelivered 37Not given
 6414IVIGLovenoxMiscarriage5  
 7343IVIGLovenoxDelivered 372523
2012
 8324IVIGHeparinDelivered 403629
 9417IVIGHeparinMiscarriage4  
10396IVIGLovenoxDelivered 382835
11455IVIGClexaneMiscarriage6  
12404IVIGLovenoxDelivered 352268
13403IVIGLovenoxMiscarriage10  
14413IVIGLovenoxMiscarriage6  
15374IVIGLovenoxDelivered 362580
16413IVIGLovenoxDelivered 393827
17364IVIGHeparinMiscarriage7  
18323IVIGLovenoxMiscarriage9  
19383IVIGLovenoxDelivered 403629
20368IVIGLovenoxMiscarriage5  
21433IVIGLovenoxMiscarriage9  
22386IVIGLovenoxDelivered 383204
23375IVIGHeparinMiscarriage8  
24313IVIGHeparinDelivered 38Not given
25355IVIGArixtraMiscarriage7  
26333IVIGLovenoxDelivered 24Loss
27444IVIGLovenoxDelivered 383402
28364IVIGLovenoxDelivered 393572
29316IVIGLovenoxMiscarriage6393260
30295IVIGLovenoxMiscarriage8  
31353IVIGClexaneDelivered 382411
32433IVIGLovenoxMiscarriage8  
33415IVIGLovenoxMiscarriage5  
34413IVIGLovenoxDelivered 393062
35325IVIGRegular heparin switched to LovenoxDelivered 394082
36345IVIGLovenoxDelivered 383857
37353IVIGLovenoxDelivered 311842
Table III.   Group III (Hep + IVIG + Humira) Patient Characteristics, Treatments and Outcomes
Patient historyTreatmentsOutcomes
PatientAgeNo. of lossTNF therapyIVIGHeparinPregnancyLoss (weeks)Delivery (weeks)Weight (g)
 1424HumiraIVIGClexaneDelivered Not givenNot given
 2285HumiraIVIGLovenoxDelivered 383655
 3406EnbrelIVIGLovenoxMiscarriage8  
 4383HumiraIVIGHeparinMiscarriage7  
 5484HumiraIVIGClexaneDelivered 393402
 6398HumiraIVIGLovenoxMiscarriage16  
 7445HumiraIVIGLovenoxMiscarriage8  
 8345EnbrelIVIGLovenoxDelivered 383006
 9305EnbrelIVIGLovenoxDelivered 372467
10395HumiraIVIGLovenoxDelivered 403572
11314HumiraIVIGLovenoxDelivered 393742
12438HumiraIVIGHeparinMiscarriage8  
13263HumiraIVIGLovenoxDelivered 373175
14433HumiraIVIGLovenoxDelivered 383119
15234HumiraIVIGLovenoxDelivered 402864
16395HumiraIVIGLovenoxDelivered 393175
17344HumiraIVIGArixtraDelivered 362693

Anticoagulant Therapy

For all treatment groups (I, II, and III) AC therapy was started preconception or at a positive pregnancy test. AC therapy consisted of standard heparin 5000 IU b.i.d, Lovenox® 30 or 40 mg q.d. or b.i.d., (Sanofi-aventis, Paris, France) Clexane® 20 mg or 40 mg q.d. or b.i.d. (Sanofi-aventis, Paris, France) Arixtra® 2.5 mg q.d. or b.i.d (Glaxo SmithKline, Brentford, UK).

IVIG Therapy

In groups II and III, IVIG (400 mg/kg body weight) was administered at least once during the cycle of conception and/or at least once after a positive pregnancy test.

TNF-α Therapy (Enbrel® or Humira®)

In group III, either Adalimumab 40 mg was injected by subcutaneous injection every 1–2 weeks or Etanercept 25 mg was injected by subcutaneous injection every 84 hr. For patients receiving a TNF-α inhibitor, medication was generally started 30 days before starting a cycle of conception and continued until cardiac activity was demonstrated by ultrasound.

Miscarriage

Was defined as a failed pregnancy that has reached a minimum beta-HCG level of 25 and/or demonstrated a visible uterine pregnancy sac via ultrasound. Ectopic pregnancies and elective terminations of viable pregnancies were excluded from this study.

Preconception testing parameters

All patients included in this study completed immunologic and thrombophilia testing. Testing was performed at the Rosalind Franklin Clinical Laboratory in Chicago. In most cases, blood was drawn 1–6 months preconception.

NK Cytotoxicity

The NK cytotoxicity was assessed by flow cytometry where labeled K562 target cells were incubated with isolated patient mononuclear cells. Target cell killing was assessed by propidium iodide uptake. NK cytotoxicity was tested at an effector to target ratio of 50:1. Cytotoxicity was regarded as increased when target cell killing exceeded 15 percent.

Acquired Thrombophilia: Antiphospholipid Antibodies

Antibodies of three immunoglobulin classes (IgM, IgG, IgA) directed against six phospholipid antigens (cardiolipin, phosphatidylethanolamine, phosphatidylserine, phosphatidylglycerol, phosphatidylinositol, phosphatiditic acid) were tested.

Inherited Thrombophilia

Genetic tests to identify carriage of alleles associated with thrombophilia were performed. A patient who carried one or more of the following mutations was regarded as positive: Factor V Leiden R506Q mutation heterozygous or homozygous; prothrombin G20210A mutation heterozygous or homozygous; PAI-1 4G/5G heterozygous or homozygous; MTHFR C677T homozygous; MTHFR C677T/ A1298C compound heterozygous.

Antinuclear Antibodies

Both antinuclear antibodies and DNA histones were tested.

Results

Patient populations of all three treatment groups were relatively homogeneous for age, past miscarriages, inherited thrombophilia and autoimmunity. Average maternal age at conception for group I was 36.8 ± 5.2 years, group II, 37.2 ± 4.1 years and for group III was 36.5 ± 7.0 years (Fig. 1). Average number of prior miscarriages for group I was 4.6 ± 2.1 for group II was 4.3 ± 1.3 and for group III was 4.8 ± 1.5 (Fig. 2). Preconception test results were also quite similar, however group III had a somewhat higher incidence of NK cytotoxicity and frequency of antinuclear and antiphospholipid antibodies, suggesting a higher incidence of autoimmunity (Fig. 3). However, live birth rates were greater within group III. Group I, using AC and baby aspirin, had a live birth rate of 19% (4/21). Group II, using AC, baby aspirin and IVIG had a live birth rate of 54% (20/37), group III using anticoagulants, baby aspirin, IVIG with TNF-α inhibitors had a live birth rate of 71% (12/17). Statistical analysis of pregnancy outcome was performed using Fisher’s exact test. There was significant improvement in pregnancy outcome in group II versus group I (P = 0.0127) and in group III versus group I (P = 0.0026) (Fig. 4). Delivery outcomes were also quite similar for the three different treatment groups. Average gestational age at delivery for group I was 38.8 ± 1.0 weeks, for group II was 37.2 ± 3.6 weeks (one set of twins), for group III was 38.3 ± 1.3 weeks (Fig. 5). The average delivery birthweight was 3693 ± 542 g for group I, 3080 ± 659 g for group II and 3170 ± 403 g for group III (Fig. 6). One patient in group II delivered a baby with Down Syndrome, possibly related to her advanced maternal age of 42. Side effects of AC, IVIG and TNF-inhibitor treatment were minimal, and no other birth defects were identified in any of the other offspring observed in this study.

Figure 1.

 Average age at conception: group I Hep: 36.8 ± 5.2 years; group II Hep + IVIG: 37.2 ± 4.1 years; group III Hep + IVIG + Humira: 36.5 ± 7.0 years.

Figure 2.

 # Prior miscarriages for treatment groups: group I Hep: 4.6 ± 2.1 mc; Group II Hep + IVIG: 4.3 ± 1.3 mc; Group III Hep + IVIG + Humira: 4.8 ± 1.5 mc.

Figure 3.

 Preconception test results: group I. Hep: 55% (6/11) NK 50:1+ 29% (6/21) ANA+, 38% (8/21) APA+, 25% (5/20) + inherited thrombophilia; group II. Hep + IVIG: 29% (5/17) NK 50:1+, 16% (6/37) ANA+, 49% (18/37) APA+, 24% (9/37) inherited thrombophilia+; group III Hep + IVIG + Humira: 70% (7/10) NK 50:1+, 35% (6/17) ANA+, 59% (10/17) APA+, 29% (5/17) inherited thrombophilia+.

Figure 4.

 Live birth rates for treatment groups with RSA: group I: Hep 19% (4/21); group II: Hep + IVIG: 54% (20/37) P = 0.0127; group III: Hep + IVIG + Humira 71% (12/17) P = 0.0026.

Figure 5.

 Average gestational age at delivery: group I: Hep: 38.8 ± 1.0 weeks; group II: Hep + IVIG 37.2 ± 3.6 weeks (one set of twins); group III: Hep + IVIG + Humira 38.3 ± 1.3 weeks.

Figure 6.

 Average delivery weights for treatment groups: group I Hep: 3693 ± 542 g; group II. Hep + IVIG: 3080 ± 659 g (one set of twins); group III. Hep + IVIG + Humira: 3170 ± 403 g.

Discussion

The group receiving AC, IVIG, and anti-TNF-α therapy, the triple therapy group, had the highest frequency of laboratory abnormalities in each of the measured categories: NK cytotoxicity, antinuclear antibodies, antiphospholipid antibodies, and inherited thrombophilias. Any selection bias, therefore, present amongst these groups is not likely to improve results in the triple therapy group.

The mechanisms whereby patients receiving combination therapy have enhanced benefit remain unclear. However, these benefits may be mediated by proteins of the coagulation cascade. The hemochorial placenta comprises an extensive hemiallogeneic surface interacting with maternal blood which must accommodate the progressive needs of the growing fetus. Remodeling involving deposition of fibrin, migration of extravillous trophoblast, placental anchorage and conversion of vessels all involve enzymes and signaling molecules of the coagulation cascade. Placental development during the first trimester exists precariously balanced on a knife edge between vascular growth accompanied by the risk of hemorrhage and the counter-risk of excessive thrombosis.

A number of unique features of the hemochorial placenta involve the enzymes and signaling moleculesassociated with the coagulation cascade. Unlike most blood-bathed surfaces in the body, syncytiotrophoblast constitutively express tissue factor as well as the pro-coagulant phospholipid, phosphatidylserine. Conversion of high resistance vessels into the high capacitance vessels, essential for adequate nutrient and blood gas transfer in the hemochorial placenta, enhances the risk of hemorrhage.22 Moreover, unlike all other vascular beds, the maternal endothelium is replaced by trophoblast which expresses thromboregulatory proteins.23 Thrombin and other coagulation proteases provide signals mediated through protease-activated receptors (PARs) transforming extravillous trophoblast into an invasive phenotype.24

Conditions within the developing placenta are thus ripe for coagulation and fibrin deposition. Histologic examination of the term placenta reveals extensive areas of fibrin deposition, incorporation of ghost villi suggesting a constant process of remodeling, growth, and involution. It is, therefore, not surprising that otherwise silent inherited and acquired thrombophilias may have profound effects on the survival of the embryo. The major natural system for inhibition of thrombin generation is the protein C anticoagulant system. It comprises protein C, S, activated protein C (APC), thrombomodulin (TM), and endothelial cell protein C receptor (EPCR).25–27 The protein C system acting through thrombomodulin, EPCR and thrombin have opposite effects. Protein C acts on members of the PARs to augment trophoblast growth as well as control excessive fibrin deposition while thrombin generation acts on the same system to inhibit trophoblast growth.24

Crosstalk between inflammatory and thrombogenic pathways is known to occur.28 It is well recognized that the damaged endothelium is a common site for both the organization of pro-coagulant systems as well as pro-inflammatory cytokines. Cytokines modulate thrombomodulin expression and the activation of protein C altering the balance of pro-coagulant and anticoagulant activities. Cytokines such as TNF-α are known to down-regulate protein C, thrombomodulin and EPCR.29

Moreover, EPCR is structurally similar to the CD1 family of molecules suggesting a role in natural immunity. Blocking APC-binding to EPCR appears to inhibit host response to bacterial challenge thereby enhancing pro-coagulant and inflammatory responses.30 The pro-inflammatory, Th1 cytokines, TNF-α and IFN-γ also act via thrombin to trigger inflammation and fibrin deposition. Clark et al. demonstrated pro-inflammatory cytokine-driven down-regulation of CD200 (OX-2) in a murine model. CD200 counteracts the prothrombinase Fgl2 at the feto–maternal interface.31

Anticoagulation was a therapeutic component in each of the three patient groups. Assuming that anticoagulation was provided at full therapeutic dose, the added beneficial effect of IVIG demonstrated in the study suggests involvement of additional pathways. Many different mechanisms for the immunomodulatory effects of IVIG have been postulated. CD200 may be a component.32 The addition of anti-TNF-α agents may directly inhibit pro-inflammatory input into the system.

We suggest direct pathway inhibition of coagulation may be insufficient to inhibit cross-reacting inflammatory pathways. We demonstrate additional benefit in cohorts treated with combination therapies.

Conclusion

In women with RSA, addition of IVIG or a TNF-α inhibitor with IVIG to the AC regimen appears to improve live birth rates compared to treatment with AC or IVIG with AC alone. Pregnancies appear to be healthy and the babies normal. We propose that TNF therapy may offer an additional treatment option for some women with immunologic RSA who have failed with AC or IVIG therapy. The combined effect of IVIG and TNF inhibitor therapy on pregnancy outcome merits further study in prospective clinical trials.

Ancillary