Problems of definition and ascertainment

Problems of definition arise very early in pregnancy, where reliable detection of pregnancy is only possible with biochemical testing using urinary or serum β-human chorionic gonadotrophin (β-hCG), and also in late pregnancy, where the distinction between a late mid-trimester miscarriage and a stillbirth can be difficult (Chard 1991). The ideal research model for determining rates of miscarriage is a prospective longitudinal study of a representative cross-section of the population which is capable of recognizing all conceptions immediately, takes account of termination of pregnancy and follows women through the first 20 weeks of pregnancy (Regan et al 1989, Alberman 1992).

Early histological work from the 1950s (Hertig et al 1959), in which fertilized ova were directly observed in 107 hysterectomy specimens from women who had intercourse around the time of expected ovulation prior to their operation, suggests a postimplantation pregnancy rate of 58% (21 of 36 possible conceptions). In a mathematical model, total pregnancy loss rates have been estimated at 78% (Roberts and Lowe 1975).

Rate of miscarriage

Following critical review of the literature, a reasonably coherent picture is emerging (Figure 23.1). Estimates of early reproductive loss rates in the peri-implantation period of approximately 50–70% still rest largely on the early work of Hertig, although data derived from in-vitro fertilization (IVF) studies support losses of this order of magnitude (Chard 1991).

Figure 23.1 Fetal loss by gestational age.

Adapted from Kline J, Stein Z, Susser M 1989 Conception to birth — epidemiology of prenatal development. In: Kline J (ed) Monographs in Epidemiology and Biostatistics, Vol. 14. Oxford University Press, Oxford.

Postimplantation and biochemical pregnancy loss rates appear to be of the order of 30%, whereas recognized pregnancy losses after clinical recognition of pregnancy remain consistent in most studies as 10–15% (Table 23.1).

Table 23.1 Preclinical and clinical miscarriage rates

Chromosomal abnormalities

Miscarriage is a heterogeneous condition, but the single largest cause of sporadic miscarriage is fetal chromosomal abnormalities, accounting for approximately 50% of all cases (Table 23.2).

It is difficult to establish the precise contribution made by fetal chromosomal abnormalities. The figures in Table 23.2 are probably an underestimate, since miscarriages occurring as a result of fetal chromosomal anomalies are maximal at the earliest and least well-documented stages of pregnancy. Evidence from preimplantation genetic diagnosis studies of embryos created as a result of IVF suggests that at least 65% of all embryos are chromosomally abnormal (Gianaroli et al 2000). However, extrapolation of these results to miscarriage in the general population must be cautious, given the high degree of selectivity of these patients and a different microenvironment at fertilization.

Trisomies are the major fetal chromosomal abnormality in sporadic cases of miscarriage, being found in 30% of all miscarriages and 60% of chromosomally abnormal miscarriages. Trisomies, together with monosomy X (15–25%) and triploidy (12–20%), account for over 90% of all chromosomal abnormalities found in sporadic cases of miscarriage.

Trisomies for all chromosomes have been described, with the exception of chromosomes 1 and Y, although the relative frequencies are vastly different. Chromosome 16 and, to a lesser extent, chromosomes 2, 13, 15, 18, 21 and 22, account for the majority of trisomic abnormalities (Figure 23.2). Most trisomies are believed to be a consequence of non-disjunction during maternal meiosis. Trisomy 16 gives rise to only the most rudimentary embryonic growth with an empty sac (Edmonds 1992), and other trisomies often result in early embryonic demise.

Figure 23.2 Distribution of differing trisomies in sporadic cases of miscarriage.

Monosomy X (45XO) is thought to result from paternal sex chromosome loss (which can be either X or Y). It is usually associated with the presence of a fetus, although focal abnormalities, such as encephalocoele or hygromata, may occur (Edmonds 1992).

Polyploidy results from the addition of complete haploid sets of chromosomes. Most common are triploid micarriages, usually 69XXY or 69XXX, which are thought to result from dispermic fertilization. These pregnancies are also characterized by multiple fetal and placental changes, such as neural tube defects, omphalocoele, hydropic villi and intrachorial haemorrhage. Tetraploid pregnancies are rare and usually do not progress beyond the third week of embryonic life.

Structural chromosome abnormalities have been reported, although minor abnormalities are no less frequent in live births compared with miscarriages. Major structural defects may arise de novo or be inherited. They are rarely a cause of sporadic miscarriage but are associated with recurrent miscarriage.

There appears to be a difference in the gestational age of pregnancy loss between different types of chromosomal abnormality, with trisomic and monosomic pregnancies miscarrying at a modal peak of 9 weeks and triploid pregnancy losses spanning 5–16 weeks of gestation (Alberman 1992).

Fetal malformations other than those caused by chromosomal anomaly

There is little doubt that the risk of miscarriage is increased with fetal malformation (Alberman 1992). However, since there has been no systematic search for malformations in fetal loss and since, in some cases, the malformation is secondary to a chromosomal anomaly, it is very difficult to assess the size of the increased risk. A single published study (Shephard et al 1989) has demonstrated an increased neural tube defect rate (3.6%) and an increased facial cleft rate (2.7%) amongst spontaneous miscarriages. Overall, the proportion of miscarriages as a result of fetal abnormalities without any associated chromosomal anomaly is small.

Placental abnormalities

Histological analysis of placental tissue from sporadic miscarriages has revealed several different patterns (Rushton 1995). These either point towards early fetal or embryonic demise with essentially normal placentation, or they have shown abnormal placental villous development with marked villous hypoplasia, reduced vascularization, enlarged intravillous spaces, often with clots and lacking significant extravillous trophoblastic infiltration, or with acute necrotic changes in the villi and associated clots which can be either patchy or global. Other reported placental changes have included inflammatory changes or a mixture of the above changes.

Amongst sporadic miscarriages, abnormalities of villous development have been reported fairly consistently at 30–40% (Hustin and Jauniaux 1997). However, the causes leading to the majority of these cases of abnormal placentation and therefore subsequent miscarriage remain unknown. Although autoimmune and infective causes can be associated with abnormal villous development, most inherently abnormal embryos probably also have abnormal placentation.

At present, although routine histological analysis of all products of conception is indicated in order to exclude molar and ectopic pregnancies, it is of little benefit in determining the cause of sporadic miscarriage.

Infection

Infection has been cited as a cause of late pregnancy loss and also early pregnancy loss for the past century. However, the precise role of infection as a cause of sporadic miscarriage is poorly and inconsistently reported (Simpson et al 1996). In many cases, whether infection has actually preceded any fetal demise or merely arose afterwards remains satisfactorily unanswered.

Several organisms have been associated with miscarriage (Box 23.1). Listeria, toxoplasmosis, herpes varicella zoster and malaria (Plasmodium falciparum) appear to be the most clinically important pathogens in women with early miscarriage.

Rubella infection, although a cause of first-trimester miscarriage, is now rare. The role of cytomegalovirus is unclear, although primary infection may cause miscarriage. The role of chlamydia infection, whether via an acute primary infection or a resultant chronic endometritis, also remains unclear. An association between herpes simplex virus infection and early pregnancy loss was first reported in the 1970s (Nahmias et al 1971), although later prospective studies have failed to show any association with sporadic or recurrent miscarriage even when primary infection occurred in the first trimester (Stray-Pedersen 1993). It remains unknown whether human immunodeficiency virus (HIV) is an important cause of early pregnancy loss. Some studies have demonstrated an increased rate of early pregnancy loss with HIV (d’Ubaldo et al 1998), while others have not (Bakas et al 1996). Whether any increased risk of miscarriage is the result of the virus, the mother’s general health or her immunocompromised status remains unclear.

Syphilis (Treponema pallidum) and parvovirus B19 are more commonly a cause of late second-trimester miscarriage and stillbirth, whilst group B streptococcus has been implicated in late miscarriage and preterm labour. Bacterial vaginosis (Trichomonas vaginalis) is also associated with late second-trimester miscarriage and preterm labour, but the use of metronidazole or other antibiotics in women with demonstrable bacterial vaginosis infection does not reduce late miscarriage or preterm delivery rates (Carey et al 2000).

Fetal sex, multiple pregnancy, maternal age and parity

Most studies have reported an excess of males in miscarriages and pregnancies complicated by varying degrees of placental dysfunction (Kellokumpu-Lethtinen and Pelliniemi 1984, Edwards et al 2000). However, the sex ratio of most early conceptions still remains unknown.

Multiple pregnancy is associated with an increased risk of fetal loss, either via early resorption, postimplantation loss or second-trimester miscarriage. In early pregnancy, the risk of miscarriage is twice that of singleton pregnancies (Sebire et al 1997a). The rate of late pregnancy loss is also increased, particularly in monochorionic twin pregnancies, where late miscarriage rates can approach 12% (Sebire et al 1997b).

The risk of miscarriage rises with parity; however, the rise is a result of reproductive compensation (Alberman 1987). The risk of miscarriage in first pregnancies is low in young women (Regan et al 1989), but rises significantly after the age of 39 years (Figure 23.3). This rise is not only found in association with trisomic pregnancies (which rise with maternal age) but also in chromosomally normal pregnancies.

Figure 23.3 The risk of miscarriage in first pregnancies with maternal age.

Adapted from Alberman E 1992 Spontaneous abortions: epidemiology. In: Stabile I, Grudzinskas G, Chard T (eds) Spontaneous abortion — diagnosis and treatment. Springer-Verlag, London.

Maternal health

Virtually every maternal medical disorder has been associated with sporadic miscarriage. Women with severe medical disease rarely become pregnant, but if they do, their disease may deteriorate during pregnancy. Various mechanisms, including endocrinological, immunological or infective, have been suggested. Poorly controlled diabetes is associated with an increased risk of miscarriage (Mills et al 1988), whereas well-controlled and subclinical diabetes can rarely be considered a cause. Overall, only a small fraction of all early pregnancy losses can be considered to be attributable to severe maternal disease.

Cigarette smoking has been positively correlated with miscarriage, and a review of the effects of nicotine on ovarian, uterine and placental function suggests that cigarette smoking has an adverse effect on trophoblast invasion (Schiverick and Salafia 1999). Cocaine use has also been reported to increase the risk of miscarriage (Ness et al 1999). Alcohol consumption has been shown to be higher in women whose pregnancies ended in miscarriage compared with pregnancies which proceeded beyond 28 weeks of gestation (Harlap and Shapiro 1980, Kline et al 1980), although other studies have not confirmed this observation (Halmesmäki et al 1989, Parazzini et al 1990). There is some evidence to suggest that even moderate maternal alcohol consumption is associated with an increased risk of miscarriage (Windham et al 1997). Many studies have shown a relationship between coffee/caffeine intake and spontaneous miscarriage, although these have not controlled for other confounding variables. Only high levels of caffeine metabolites in maternal serum are associated with miscarriage, so it would appear that moderate consumption of caffeine is unlikely to increase the risk of miscarriage (Klebanoff et al 1999).

Some chemical agents including lead, ethylene oxide, solvents, pesticides, vinyl chloride and anaesthetic gases have been shown to have some association with fetal loss (Cohen et al 1971, Mur et al 1992, Rowland et al 1996, McDonald et al 1988). Although many environmental toxicologists accept these agents as proven, the evidence for low levels of exposure remains far from convincing.

Radiotherapy and chemotherapeutic agents are accepted causes of miscarriage (Zemlickis et al 1992), although they are only administered during pregnancy in seriously ill women. Although significant ionizing radiation for diagnostic purposes can lead to fetal malformation and miscarriage, a dose of more than 25 rads is associated with a 0.1% risk of abnormality (1 rad is the equivalent of eight to 10 abdominal/pelvic X-ray films).

There is a small increase in the risk of spontaneous miscarriage with general anaesthesia and incidental surgery during the first and second trimesters, although this is higher with gynaecological surgery (Duncan et al 1986).

It has been claimed that impaired psychological well-being predisposes to fetal loss (Stray-Pedersen and Stray-Pedersen 1984), although it has been difficult to exclude other confounding variables. Despite problems with recall bias, there are an increased number of negative life events in women with chromosomally normal as compared with chromosomally abnormal miscarriages (Neugebauer et al 1996), but no differences in hormonal markers of stress have been determined.

Threatened miscarriage

This can be defined as uterine bleeding before 24 weeks of pregnancy (the time of presumed fetal viability) with no evidence of any fetal or embryonic demise. It is usually painless. Characteristically, the bleeding is initially bright red, followed by a reducing brown loss. This can affect up to 25% of all pregnancies and is one of the most common indications for emergency early pregnancy referral. Clinical examination reveals a soft, non-tender uterus, usually of the appropriate gestational size, and a closed cervix. Transvaginal or transabdominal ultrasound scan confirms an ongoing pregnancy (Figure 23.4).

Figure 23.4 Ultrasound picture of an ongoing pregnancy with an intrauterine haematoma in a woman who presented with a threatened miscarriage.

The non-specific nature of abdominal pain, vaginal bleeding and pelvic tenderness precludes their use as predictors of ultimate outcome in women with threatened miscarriage. However, continued vomiting (rather than nausea alone) in early pregnancy is associated with an increased chance of live birth (Weigel and Weigel 1989), presumably because it indicates continued placental hormone production.

Inevitable miscarriage

Miscarriage is a process rather than a single event. Of all women presenting with bleeding in early pregnancy, approximately 50% will ultimately miscarry (Stabile et al 1987). Whilst the cervix remains closed, any pain and bleeding may subside and the pregnancy may continue otherwise normally to term. However, once the cervix opens, miscarriage is inevitable.

Usually, women present with crampy abdominal pains and fresh bleeding. Symptoms alone are unreliable and the diagnosis is determined by the confirmation of cervical dilatation at vaginal examination. Examination often reveals a tender, firm uterus, which may be smaller than the expected gestational size, and the cervical os is open. Products of conception may also be felt through the os.

Occasionally, the woman may present with severe shock. This can either be secondary to massive haemorrhage, which will require appropriate emergency resuscitation, blood transfusion and uterine evacuation, or the degree of shock may be out of proportion to the blood loss and is due to the distension of the cervix by products of conception and the resultant sympathetic stimulation. This is termed ‘cervical shock syndrome’ and is often severe. Quick removal of the products from the os, at the bedside, results in relief of the shock. In rare cases, endotoxic shock secondary to sepsis may occur.

Complete and incomplete miscarriage

Inevitable miscarriage is either complete or incomplete depending on whether or not all fetal and placental tissues have been expelled from the uterus. The typical features of incomplete miscarriage are heavy bleeding (which may be intermittent) and abdominal cramps. The finding of a dilated cervix on examination in the presence of continued pain and bleeding is usually diagnostic. Ultrasound scan will confirm the presence of retained products of conception (Figure 23.5).

Figure 23.5 Ultrasound picture of incomplete miscarriage demonstrating retained products of conception within the endometrial cavity.

If the symptoms of incomplete miscarriage resolve spontaneously, complete miscarriage may have occurred. Nevertheless, following abdominal and pelvic examination, ultrasound may need to be perfomed to confirm that the uterus is indeed empty. Occasionally, the cervix may close despite the presence of retained products. The number of complete miscarriages is unknown. Many such early miscarriages may go unreported. Early studies of women with threatened miscarriage showed complete miscarriage rates of less than 1% (Stabile et al 1987), although more recent evidence suggests that approximately 20–30% of miscarriages are complete (Chung et al 1994, Mansur 1992). Similarly, expectant management of early fetal demise has demonstrated that, with time, up to 25% of women go on to have a complete miscarriage (Jurkovic et al 1998).

Early fetal demise (missed abortion/miscarriage)

‘Early embryonic or fetal demise’ is now the preferred term in both the UK and the USA, rather than ‘missed abortion’ or ‘missed miscarriage’. Previously, this has also been termed an ‘anembryonic pregnancy’ (all pregnancies develop from an embryo) or ‘blighted ovum’ (the pregnancy cannot result from an ovum alone). These terms should also be abandoned.

Early fetal demise is where failure of pregnancy is identified before any expulsion of the products of conception occurs. The woman may report a disappearance of the symptoms and signs of early pregnancy, such as nausea and vomiting or breast tenderness. A brown vaginal loss may also be reported. The diagnosis may also be made in otherwise asymptomatic women at routine obstetric/dating ultrasound examination. The diagnosis is made by the lack of fetal heart activity in a pregnancy with a crown–rump length of over 5 mm (Pennell et al 1991), or when two ultrasound examinations 1 or 2 weeks apart have shown no growth and no fetal heart activity. In cases of doubt, a repeat ultrasound is always indicated. Occasionally, a collapsing gestational sac or a failed pregnancy surrounded by clot in utero may be seen (Figure 23.6). Abdominal and pelvic examination often reveals a uterus smaller than expected for the gestational age and a closed cervical os.

Figure 23.6 Ultrasound picture of early fetal demise, showing a collapsing gestation sac.

Sepsis

This occurs when infection complicates miscarriage or termination of pregnancy. Intrauterine sepsis rarely follows incomplete miscarriage, although the incidence is considerably higher at approximately 3.6% following termination of pregnancy (Frank 1985). The most common infecting organisms are Escherichia coli, Bacteroides, streptococci (both anaerobic and occasionally aerobic) and Clostridium welchii. Occasionally, a history of intrauterine instrumentation may be withheld. The woman usually presents with suprapubic pain, malaise, fever and, occasionally, vaginal bleeding. Findings on examination include abdominal rigidity, uterine and adnexal tenderness, and a closed cervical os. Rarely, septicaemia may ensue, leading to bacteraemic endotoxic shock and possibly maternal death.

Trophoblastic tumours

Trophoblastic tumours include complete and partial hydatidiform moles, choriocarcinoma and placental site tumours. These occasionally present as threatened miscarriage, and suggestive features are noted at ultrasound examination. Trophoblastic tumours are discussed in Chapter 43.

Threatened miscarriage

Early studies demonstrated that the presence of fetal heart activity at ultrasound examination in women who present with a history of bleeding in early pregnancy was associated with a high chance (97–98%) of a live birth (Stabile et al 1987). More recent evidence confirms livebirth rates of 90–95% for younger women, but for women over 40 years of age, miscarriage rates of 15–30% are reported even after the identification of fetal heart activity (Deaton et al 1997, Schmidt-Sarosi et al 1998). Nevertheless, for most women, reassurance and continued medical and emotional support are all that is required.

Bed rest and avoiding penetrative intercourse have historically been advised. If the vascular hypothesis for the pathophysiology of miscarriage is accepted (Hustin and Jauniaux 1997), bed rest may improve pressure variation and flow changes, and therefore improve the outcome. However, there is a paucity of clinical evidence to support this hypothesis. Initial presentation is usually in primary care and, currently, over 96% of general practitioners still advise bed rest and the avoidance of intercourse, although many believe that it does not improve the eventual outcome.

Progesterone supplementation in early pregnancy has been prescribed for over 30 years for women presenting with threatened (and also recurrent) miscarriage. The historical rationale was that a progesterone deficiency would lead to miscarriage. Obviously, the converse may also be true; a failed pregnancy may lead to a progesterone deficiency. There is a wealth of published data, mostly from uncontrolled treatment trials. However, several meta-analyses (Daya 1989, Goldstein et al 1989) have been unable to demonstrate a beneficial effect for progesterone treatment. The routine use of progesterone in threatened miscarriage cannot be justified.

Incomplete miscarriage

Surgical evacuation of the uterus after cervical dilatation, if necessary, has remained the cornerstone of management of incomplete miscarriage in the industrialized world since the 1940s (Hertig and Livingstone 1944). Evacuation and curettage have been regarded as essential to ensure that the uterine cavity is empty, otherwise haemorrhage, infection and later complications such as Asherman’s syndrome may result. Early studies showed a maternal mortality rate of 1.6% in women who did not undergo surgical treatment (Russell 1947). In cases of haemodynamic compromise of cervical shock, appropriate resuscitation and urgent surgical evacuation are indicated. Most cases are performed under general anaesthesia, although effective outpatient curettage has been reported (Fawcus et al 1997). Suction evacuation is generally regarded as a safer technique than sharp curettage, with lower rates of perforation, blood loss and subsequent intrauterine adhesion formation (Edmonds 1992, Verkuyl and Crowther 1993).

The routine use of syntocinon or ergometrine has shown no benefit in reducing blood loss during the surgical treatment of first-trimester incomplete miscarriage (Beeby and Morgan Hughes 1984). Due to the risk of ascending infection and its sequelae, whenever uterine instrumentation is performed, screening for chlamydia infection is recommended, although the use of prophylactic doxycycline is only likely to be of benefit in areas of relatively high prevalence (Prieto et al 1995).

Although a minor procedure, surgical evacuation is associated with rare but serious morbidity. Complications include tearing or lacerations to the cervix, perforation of the uterus, which may also lead to bowel perforation, bladder perforation, damage to the broad ligament, infection, Asherman’s syndrome (intrauterine adhesions) and haemorrhage (Ratnam and Prasad 1990). The incidence of serious morbidity has been estimated at 2.1% (Lawson et al 1994, Royal College of General Practitioners/Royal College of Obstetricians and Gynaecologists 1985), of which the most common problem is infection. This can lead to later sequelae including secondary infertility, ectopic pregnancy and Asherman’s syndrome. The incidence of mortality associated with surgical evacuation of the uterus has been estimated at 0.5 per 100,000 (Lawson et al 1994).

Recent studies demonstrated the efficacy of expectant management or ‘observation alone’ in women with incomplete miscarriage (Nielsen and Hahlin 1995, Hurd et al 1997). In cases where there was no haemodynamic compromise or maternal anaemia, spontaneous resolution occurred within 3 days in up to 80% of cases with minimal retained products (15–50 ml). Therefore, in women with minimal intrauterine tissue (after ectopic pregnancy has been excluded), expectant management is safe. However, large volumes of retained products are associated with an increase in complications, primarily infection and prolonged bleeding (Hurd et al 1997). There is no evidence that future fertility is impaired following expectant management (Kaplan et al 1996).

Herbal remedies have been used in the past to encourage the uterus to expel its contents, but an effective non-surgical alternative to termination of pregnancy and miscarriage had to await the development of the antiprogesterone mifepristone and the prostaglandin analogues gemeprost and misoprostol. Their efficacy has been demonstrated in the treatment of incomplete miscarriage (Henshaw et al 1993, Ashok et al 1998), leading to complete miscarriage rates of approximately 95%.

The use of medical treatment or the adoption of an expectant approach in appropriately selected cases may have many medical and economic benefits. However, many women continue to express a preference for surgical treatment (Hamilton-Fairley and Donaghy 1997), citing fears regarding pain, bleeding and the length of time to resolution in the non-surgical group.

Early fetal demise

The treatment options for early fetal demise are essentially the same as for incomplete miscarriage. Surgical treatment should be preceded by cervical ripening agents (mifepristone or prostaglandins) in order to reduce the risks of cervical trauma or uterine perforation associated with forced cervical dilatation. Gemeprost, mifepristone, and oral or vaginal misoprostol seem to be equally effective (Gupta and Johnson 1992, Platz-Christiansen et al 1995, Ayres de Campos et al 2000). Expectant management is feasible, although less effective than in cases of incomplete miscarriage, with only 25% proceeding to complete miscarriage (Jurkovic et al 1998). The efficacy of medical treatment is also less when compared with incomplete miscarriage, although complete miscarriage rates of up to 90% have been reported with higher doses of mifepristone and misoprostol and a longer surveillance period (El-Refaey et al 1992).

Parental chromosomal abnormalities

Parental chromosomal abnormalities are the most important genetic anomalies currently detectable amongst couples with recurrent miscarriage. Most studies report an incidence of 3–5% (Stray-Pedersen and Stray-Pedersen 1984, Clifford et al 1994, Stephenson 1996, Li 1998), compared with an incidence of 0.5% in the general population.

Balanced or reciprocal translocations are the most frequently detected parental chromosomal anomaly in couples with recurrent miscarriage. The male:female ratio is approximately 1:2. A portion of one chromosome is exchanged with a portion of another, resulting in two abnormal chromosomes but, overall, a normal chromosomal complement. Translocations have been reported for all chromosomes in many different combinations. At gametogenesis, there is a 50% chance of a chromosomally abnormal gamete being produced. However, there appears to be a lower fertilization or implantation rate with abnormal gametes, since chorionic villus sampling and amniocentesis have demonstrated a 40% and 11% risk, respectively, of a chromosomally unbalanced fetus (Mikkelson 1985).

Robertsonian translocations are less common, occurring in approximately 1% of couples with recurrent miscarriage. Here, two chromosomes adhere to each other at either the centromere or the short arms of the chromosome. This leads to a total chromosome count of 45 but a normal chromosome complement overall. The risks of recurrence and miscarriage vary in different Robertsonian translocations. The most common translocation (involving chromosomes 14 and 21) results in 10–15% of pregnancies with trisomy 21 (Boué and Gallano 1984). If homologous chromosomes are involved, the risk of a chromosomally abnormal conceptus is 100%.

Pericentric chromosomal inversions occur in approximately 1% of the general population, and are not considered to be of clinical significance for recurrent miscarriage.

Peripheral blood karyotyping should be performed in both partners of all couples with recurrent miscarriage to determine any parental chromosome abnormality. Specialized genetic counselling is essential and offers the couple a prognosis for future pregnancies; this may, in some cases, not be as bleak as anticipated. There should also be the opportunity to discuss preimplantation genetic diagnosis and antenatal fetal karyotyping, including chorionic villus sampling and amniocentesis. In cases of homologus Robertsonian translocations, gamete donation and/or effective contraception should be discussed.

It is important to note that over 35% of couples with a significant parental chromosomal abnormality have already achieved a successful pregnancy in addition to their miscarriages (Clifford et al 1994).

Recurrent aneuploidy

Some couples have an increased risk of miscarriage because they produce recurrent aneuploid fetuses. This may be the result of an increased tendency to non-disjunction, either inherited or induced environmentally. Chromosomal abnormalities are seen with a higher frequency amongst preimplantation embryos created during IVF from couples with a history of recurrent miscarriage (Vidal et al 2000). Presumably, these are the result of errors in gametogenesis, and much current research is focused on the role of sperm chromosome abnormalities in recurrent miscarriage.

Other genetic factors

The karyotype of many recurrent miscarriage conceptions may be euploid (46XX or 46XY), and in the absence of any other identifiable cause or association, these unexplained recurrent early pregnancy losses may be the result of molecular mutations or single gene defects. Chromosome analysis itself is a very crude tool to determine genetic abnormalities. So far, there has only been one report of a specific single gene locus abnormality associated with recurrent miscarriage (Pegoraro et al 1997), but much work remains to be done. The impact of other genetic factors on recurrent miscarriage is confirmed by the increased incidence amongst consanguinous couples (Hendrick 1988).

Uterine anomalies

The reported incidence of uterine anomalies in women with recurrent miscarriage varies widely from as low as 1% up to 27% (Regan 1997). Described anomalies vary from uterus didelphis to subseptate uteri, and some studies even include submucosal fibroids. The incidence of such anomalies in the fertile population has been reported to be 1.8–3.6% (Ashton et al 1988).

The significance of such anomalies, particularly intrauterine septae, arcuate uterus and submucosal fibroids, in women with recurrent miscarriage is unclear. Historically, uterine anomalies were associated with second-trimester miscarriages, but later reports have also shown an increase in early miscarriages. The exact aetiology is unknown, although implantation over a septum or similar defect may result in decreased vascularity of the placenta.

Hysterosalpingography (HSG), laparoscopy and hysteroscopy, magnetic resonance imaging, computed tomography and three-dimensional ultrasound have all been used in the diagnosis and evaluation of uterine anomalies. Although HSG has traditionally been used as the primary screening test, three-dimensional ultrasound shows promise as a less invasive screening tool (Jurkovic et al 1995). In most cases, further evaluation at laparoscopy and hysteroscopy is indicated (Figure 23.7).

Figure 23.7 (A) Hysterosalpingogram demonstrating a septate uterus. (B) Three-dimensional ultrasound demonstrating a septate uterus.

Treatment of uterine anomalies in women with recurrent miscarriage remains controversial. Many reports demonstrate a benefit of hysteroscopic or conventional metroplasty in women with septate, subseptate or bicornuate uterus who have suffered mid-trimester miscarriage compared with matched controls (Candiani et al 1990, Ayhan et al 1992, Pabuccu et al 1995, Heinonen 1997). However, the likelihood of a live birth in untreated women is as high as 66% (Heinonen 1997), and a similar percentage of women presenting with recurrent miscarriage have already achieved a live birth (Clifford et al 1994). Open pelvic surgery, in particular, has been associated with subsequent infertility (Bennett 1987). The value of corrective uterine surgery in women with these anomalies and a history of early pregnancy loss is uncertain (Ben-Rafael et al 1991). Similarly, the value of transcervical myomectomy in women with submucous fibroids remains unknown, although there have been initial successful reports (Egwuatu 1989).

Cervical incompetence

Cervical incompetence is usually associated with miscarriage occurring after 12–14 weeks or premature labour. It usually presents as silent dilatation of the cervix without painful contractions. The membranes bulge through the cervical canal and eventually rupture spontaneously.

The diagnosis of cervical incompetence poses a difficult problem (Medical Research Council/Royal College of Obstetricians and Gynaecologists Working Party on Cervical Cerclage 1993). It is usually based on a previous history of mid-trimester miscarriage in the absence of painful uterine contractions. Vaginal ultrasound may be useful to detect early features of cervical incompetence (shortening or funnelling) (Althuisius et al 2000), but neither ultrasound nor HSG has been found to be useful in the diagnosis of cervical incompetence before pregnancy.

Treatment of cervical incompetence is cervical cerclage in pregnancy, usually after 12–14 weeks when the toll of first-trimester miscarriages has occurred and after screening for chromosomal abnormalities has been performed. It has traditionally involved the insertion of a MacDonald suture, where a tape is inserted round the exposed vaginal cervix (MacDonald 1957), or a Shirodkar suture, where the vaginal mucosa is incised and the bladder reflected, allowing insertion of the suture at a higher level of the cervical canal (Shirodkar 1960). Modifications of this procedure have included burying the suture under the vaginal mucosa at the end of the procedure. In the authors’ experience, this is rarely accompanied by significant vaginal discharge. The MacDonald suture is more widely used and less traumatic to the cervix than the Shirodkar suture, although the exposed tape remains a possible focus for later infection. More recently, emergency cervical cerclage in women who present acutely in the mid-trimester with silent cervical dilatation and/or herniation of the membranes through the cervix has been reported with livebirth rates of 50–60% (Wong et al 1993, Aarts et al 1995). Nevertheless, there are high rates of premature delivery and infection, including chorioamnionitis, in over 30% of cases.

In some women, particularly where the diagnosis is unsure or in those who have other risk factors, regular ultrasound monitoring of the cervix can be performed. If there is funnelling or shortening of the cervix, emergency cerclage can be performed (Althuisius et al 2000). Ultrasound may also be useful in women who have already had cervical cerclage in order to warn of problems (Dijkstra et al 2000). Transabdominal cervical cerclage may be useful in a highly selected group of women with anatomical defects in the cervix and previous mid-trimester miscarriages or preterm labours following failed vaginal cervical cerclage (Gibb and Salaria 1995). Laparoscopic transabdominal cervical cerclage has also been reported (Scibetta et al 1998), but neither of these procedures have been assessed in appropriate trials.

Systemic endocrine disease

As discussed previously, diabetic women with good metabolic control are no more likely to miscarry than non-diabetic women. However, diabetic women with high glycosylated haemoglobin A_1c levels in the first trimester are at significantly higher risk of both miscarriage and fetal malformation (Hanson et al 1990). As the risk of miscarriage is only increased in women with poorly controlled diabetes mellitus, there is no value in screening for occult disease in asymptomatic women.

Although thyroid autoantibodies are associated with an increased risk of miscarriage, this is secondary to a generalized autoimmune abnormality rather than a specific thyroid dysfunction (Singh et al 1995). Screening asymptomatic women with a thyroid function test is unhelpful as they are usually normal (Clifford et al 1994, Rushworth et al 2000).

Luteal-phase deficiency

A functional corpus luteum is essential for the implantation and maintenance of early pregnancy, primarily through the production of progesterone, which is responsible for the conversion of a proliferative endometrium into a secretory endometrium available for embryo implantation. Disorders or removal of the corpus luteum can result in infertility and early pregnancy loss (Miller et al 1969, Csapo et al 1973). Although much controversy exists about luteal function, the success of oocyte donation with oestradiol and progesterone support in the absence of any ovarian activity would suggest that its effects are primarily progesterone determined.

The prevalence of a luteal-phase defect is reported to occur in 23–60% of women with recurrent miscarriage (Li and Cooke 1991), although it is difficult to determine the exact proportion of women with luteal-phase problems because of difficulties in diagnosis. The diagnosis is usually determined by low luteal-phase progesterone level and/or a non-secretory endometrium, by biopsy, in non-fertile, non-pregnant cycles (Jordan et al 1994, Serle et al 1994). Between 30% and 50% of cases of luteal-phase defect, as determined by endometrial histology, are found in the presence of normal circulating progesterone levels, suggesting that a primary endometrial defect is as common as deficient progesterone production (Li et al 1991). There is no reliable way of demonstrating a luteal-phase defect in conception cycles or early pregnancy as low levels of progesterone in early pregnancy are a reflection that the pregnancy has already failed because the trophoblast cannot produce sufficient progesterone. The use of colour-flow pulsed Doppler ultrasound may offer a non-invasive tool for diagnosis, and has been used in the diagnosis of luteal-phase defect, although only at a preliminary stage (Glock and Brumsted 1995).

There have been many reports of successful pregnancies following treatment with progesterone after the diagnosis of a luteal-phase defect. These studies have, for the most part, been uncontrolled. The only meta-analysis of controlled trials of progesterone treatment in women with recurrent miscarriage has not demonstrated any benefit (Daya 1989).

Treatment with hCG should stimulate progesterone production from the corpus luteum, and has been used in the treatment of luteal-phase defect. Although one small study demonstrated a beneficial effect of hCG treatment in women with oligomenorrhoea (Quenby and Farquharson 1994), meta-analysis has concluded that there is insufficient evidence at present concerning the effectiveness of hCG to recommend its use for women with unexplained recurrent miscarriage (Prendiville 1995).

Polycystic ovary syndrome and hypersecretion of luteinizing hormone

The prevalence of polycystic ovarian morphology in women with recurrent miscarriage is 40% (Rai et al 2000), and there have been numerous reports of an increased risk of miscarriage in women with hypersecretion of luteinizing hormone (LH), hyperandrogenaemia and, more recently, hyperprolactinaemia; all the classic endocrinopathies of polycystic ovary syndrome (PCOS) (Stanger and Yovich 1985, Howles et al 1987, Homburg et al 1988, Regan et al 1990, Bussen et al 1999). Nevertheless, a degree of controversy exists.

Several reports have not confirmed the relationship between hypersecretion of LH and an increased risk of miscarriage in women with recurrent miscarriage (Tulppala et al 1993, Liddell et al 1997). Whether this is due to the radioimmunoassays or a genetic variant of LH is unclear (Tulppala et al 1998). Also, suppression of high endogenous LH secretion with gonadotrophin-releasing hormone analogues in a prospective randomized placebo-controlled trial did not improve the livebirth rate (Clifford et al 1996).

The association between raised androgens and recurrent miscarriage has similarly not been confirmed with later studies (Liddell et al 1997, Rai et al 2000), although these are not universal findings and some studies have demonstrated retarded endometrial development and increased risk of miscarriage in association with raised testosterone levels (Tulppala et al 1993, Okon et al 1998).

Hyperprolactinaemia is a common finding in women with PCOS, and has also been associated with recurrent miscarriage (Hirahara et al 1998, Bussen et al 1999). Although one study has demonstrated a beneficial effect of bromocriptine treatment, this has not been corroborated elsewhere (Hirahara et al 1998).

In summary, PCOS is over-represented in women with recurrent miscarriage but the exact mechanism or individual endocrinopathy by which miscarriage is mediated remains unclear. Until this is determined, any corrective treatment is unlikely to be effective.

Autoimmune disease

Sporadic and recurrent miscarriage are recognized complications of systemic lupus erythematosus, the most common autoimmune disorder in women of reproductive age. Since the 1980s, it has been recognized that recurrent miscarriage is associated with an increase in the detection of many autoantibodies, even in asymptomatic women. The two most common groups of autoantibodies detected have been the antiphospholipid antibodies (aPL), which in conjunction with a history of recurrent miscarriage is now known as the ‘antiphospholipid syndrome’ (APS) (Harris 1987), and thyroid autoantibodies (Bussen and Steck 1995). However, with the exception of APS, the mechanisms by which a generalized increase in the autoantibody pool leads to miscarriage remain unclear. Although intravenous immunoglobulin (IVIG) therapy has been used in women with recurrent miscarriage associated with non-specific autoimmunity, at present the collective evidence indicates that IVIG does not have a therapeutic effect that is clinically meaningful (Daya et al 1999).

Antiphospholipid syndrome

APS refers to the relationship between aPL, namely lupus anticoagulant (LA) and anticardiolipin antibodies (aCL), and recurrent miscarriage, thrombosis or thrombocytopenia (Harris 1987). Since this original description, it has become apparent that the three defining clinical features are too limiting, and aPL are implicated in a wide range of clinical conditions (Box 23.3).

In women with recurrent miscarriage, a previous personal or family history of thrombosis, cardiovascular disease, epilepsy and migraine is strongly predictive of a positive aPL status (Regan 1997). The overall prevalence of aPL in women with recurrent miscarriage is approximately 15% (Li 1998).

Women with APS without treatment have a miscarriage risk of 85–90%, the majority of which occur in the first trimester after establishment of a fetal heart (Rai et al 1995a,b). Pregnancy loss associated with APS is initially attributed to defective embryonic implantation and later to thrombosis of the uteroplacental vasculature and placental infarction (Rai and Regan 1998).

Diagnosis of APS and detection of aPL are subject to widespread interlaboratory variation and the fluctuating nature of the antibodies themselves. Testing for LA should follow internationally agreed guidelines (Lupus Anticoagulant Working Party on behalf of the BSCH Homeostasis and Thrombosis Task Force 1991), and the test of choice is the dilute Russell’s viper venom time which is more sensitive than the activated partial thromboplastin time and the kaolin clotting time. Testing for aCL is by standardized enzyme-linked immunosorbent assay. Testing must be repeated on at least two occasions 8 weeks apart in order to make a diagnosis of APS. Testing for aPL other than LA and aCL is of no clinical use in women with recurrent miscarriage.

Although a variety of treatments have been described for APS, including corticosteroids (Lubbe et al 1983), low-dose aspirin (Silver et al 1993), heparin alone (Rosove et al 1990) and IVIG therapy (Carreras et al 1988), treatment with subcutaneous heparin and low-dose aspirin until 34 weeks of gestation has been shown to be the most effective treatment for women with APS and recurrent miscarriage (Rai et al 1997). In cases where there are additional medical complications, such as thrombocytopenia, IVIG may be appropriate (Cowchock 1998, Piette et al 2000).

Other thrombophilic abnormalities

While APS is an acquired autoimmune thrombophilic state, recurrent miscarriage is also associated with other inherited and acquired causes of thrombophilia (Blumenfeld and Brenner 1999). Most studies are retrospective and interpretation must be cautious because of problems with acquisition and ascertainment.

Activated protein C resistance, usually but not always inherited via factor V Leiden mutation, is an important cause of acquired venous thrombosis and thrombophilia (Bertina et al 1994). It is also associated with recurrent fetal loss (Brenner et al 1997). Successful treatment with low-dose aspirin and heparin has been reported, although there are no prospective controlled studies to date.

Hyperhomocysteinaemia, as a result of congenital enzyme deficiencies or vitamin B6, B9 and B12 deficiencies, is associated with thrombosis and premature vascular disease (Boers et al 1985). It has also been reported in association with recurrent pregnancy loss (Wouters et al 1993). Thromboprophylaxis and vitamin supplementation have been reported (Aubard et al 2000), but no prospective data are yet available.

Other inherited thrombophilias, including protein S and protein C deficiency and antithrombin III deficiency, are associated with recurrent miscarriage as well as late pregnancy complications (Girling and de Swiet 1998). These need to be managed with appropriate haematological experience because of the increased risk of thromboembolism during pregnancy.

Screening women who have a history of recurrent miscarriage has shown increased incidence of thrombin generation — a global marker of a prothrombotic state — even when not pregnant (Vincent et al 1998). Further research is needed to determine whether thromboprophylaxis is appropriate for these women.