Pregnancy and its management in the Philadelphia negative myeloproliferative diseases

Authors


Dr Claire Harrison, Department of Haematology, Guy's and St Thomas’ Foundation Trust, St Thomas’ Hospital, Lambeth Palace Road, London SE1 7EH, UK.
E-mail: claire.harrison@gstt.sthames.nhs.uk

Summary

The myeloproliferative diseases (MPDs) present several therapeutic challenges in patients of childbearing potential. The most extensive literature exists for patients with essential thrombocythaemia, with over 200 pregnancies reported in retrospective case series. Yet there is conflicting data in relation to predicting pregnancy outcome and optimal management strategy. Pregnancy is less frequently reported for polycythaemia vera and myelofibrosis. There is a need for collaboration to further our knowledge in this field. Here, the literature is reviewed in detail and experience of different therapeutic strategies in pregnancy discussed. There is increasing understanding about the pathogenesis of placental dysfunction in inherited thrombophilia and antiphospholipid antibody syndrome pregnancy outcomes in these conditions parallel those reported for MPDs. Furthermore several large studies have influenced pregnancy management in these conditions and, whilst not directly applicable to MPDs, this data have potential to inform treatment protocols. This data are reviewed and a personal management strategy for pregnancy in MPD proposed.

The myeloproliferative disorders (MPDs) have a peak incidence in the sixth and seventh decades of life but they do occur in younger patients hence posing challenges for management in pregnancy. The disease entities included in the classification of MPDs was expanded in the 2001 World Health Organization classification of haematological malignancies (Vardiman et al, 2002) to include chronic neutrophilic or eosinophilic leukaemia (or hypereosinophilic syndrome), and chronic myelomonocytic leukaemia. The former conditions are both very infrequent, and uncommon in patients of childbearing potential. This review considers the more common MPDs; essential thrombocythaemia (ET), polycythaemia vera (PV), and idiopathic myelofibrosis (MF). These diseases share many clinical features including predisposition to thrombosis, haemorrhage, progression to myelofibrosis and acute myeloid leukaemia.

One of the key factors influencing pregnancy outcome includes the development and maintenance of normal uterine blood flow and adequate placental development. The placental chorionic villi contain fetal capillaries and are bathed in maternal blood within the lacunar system that derives its blood supply from the uterine spiral arteries. This system develops in the first and early second trimester of pregnancy when the embryonic trophoblast initially invades first the endometrial capillaries and then the spiral arteries (Ramsey, 1985). The consequences of trophoblast invasion and normal placentation are conversion from a low volume, high resistance circulation to a high volume, low resistance circulation that can be detected on uterine artery Doppler scanning (Bewley et al, 1991). It has been suggested that abnormal trophoblast invasion of the maternal spiral arteries is an earlier marker of placental dysfunction than, for example, a growth-restricted foetus. Severe pregnancy complications, such as placental abruption, intrauterine growth retardation (IUGR), and pre-eclampsia, are all associated with utero-placental dysfunction. For example, when compared with normal foetuses, those with IUGR have been shown to have a fourfold reduction in utero-placental blood flow (Lunell et al, 1982).

The prothrombotic potential of pregnancy is well defined because of well characterized alterations in haemostatic factors, prothrombotic proteins and physical compromise of venous blood flow. The importance of this is confirmed by thrombosis being consistently identified as the leading cause of maternal morbidity in recent confidential enquiries in maternal mortality (Griesshammer et al, 2003; http://www.cemach.org.uk/publications/WMD2000_2002/content.htm). Thrombotic occlusion of the placental circulation may be a later manifestation of placental dysfunction or an independent mechanism of pregnancy morbidity. Histological review has also demonstrated intervillous and spiral vessel thrombosis in haematologically normal females with this spectrum of serious pregnancy complications. In patients with ET placental thrombosis has been documented in cases complicated by late fetal loss (Pagliaro et al, 1996), pre-term delivery (Pagliaro et al, 1996) and IUGR (Falconer et al, 1987). In MPDs the mechanism of the acquired prothrombotic state differs from other thrombophilias. For MPDs this probably relates to a combination of blood rheology, possibly leucocyte adhesion, platelet activation and platelet-leucocyte aggregates (Falanga et al, 2000; Jensen et al, 2001) amongst other factors.

In the past decade our knowledge and understanding of some specific prothrombotic states and their association with poor pregnancy outcome has expanded. It seems likely that close parallels exist between these conditions and the MPDs in terms of the aetiology of pregnancy-related complications and perhaps some aspects of their management. In antiphospholipid syndrome a relationship between acquired thrombophilia and recurrent pregnancy loss or gestational vascular complications, such as pre-eclampsia, exists and has been better defined. Evidence has also been accruing for a role for inherited thrombophilias, in particular factor V Leiden, prothrombin G20210A mutation, protein S, protein C and anti thrombin III deficiency in such gestational abnormalities (recently reviewed in Brenner, 2003). The stage is thus set for potentially problematic pregnancies in patients with other prothrombotic states, for example the MPDs. Clearly, it is not possible to extrapolate completely from either inherited thrombophilia or antiphospholipid syndrome to the problems in pregnancy associated with MPDs as the pathological mechanisms may differ. But common themes in the aetiology, monitoring, management and prevention are likely.

In this paper, we shall review the literature describing pregnancy and its outcome in patients with the major Philadelphia-negative MPDs ET, PV and MF. Using this knowledge and emerging information for patients with the anti-phospholipid syndrome and inherited thrombophilia in pregnancy, therapeutic options will be considered and a management plan proposed.

Essential thrombocythaemia

Essential thrombocythaemia is unique amongst the other Philadelphia negative MPDs as there appears to be a second, earlier peak in its incidence, particularly in women of child-bearing age (McNally et al, 1999). Furthermore, a recent evaluation of the incidence of MPDs in Goteborg, Sweden demonstrated an increase in the annual incidence rate for ET, although this was only statistically significant for males (Johansson et al, 2004). This increase may reflect the wider use of automated platelet counts but the net result is an overall increase in the total ET patient population.

Of the Philadelphia-negative MPDs, the most extensive literature relating to pregnancy exists for ET with over 280 pregnancies reported in 147 patients. There have been a number of reviews that have analysed all these cases (Griesshammer et al, 1996, 1998, 2003; Vantroyen & Vanstraelen, 2002) and demonstrate the increased rate of fetal complications. Overall there is a success (live birth) rate of 50–57%. First trimester loss occurs in 26–36% of ET pregnancies; the anticipated loss in ‘normal’ pregnancies is 15–20% (Hatasaka, 1994; Cook & Pridham, 1995). Late pregnancy loss is also more prevalent in ET, with 5–9·6% of pregnancies ending in this way compared with 0·5% for the normal population (Cook & Pridham, 1995; Martinelli et al, 2000). Furthermore, in ET pregnancies, IUGR occurred in 4–5·1%, preterm delivery occurred in 5·6–8%, and placental abruption was reported in 2·8%. It has been suggested that limiting such analysis to reports of larger patient cohorts from single institutions might avoid the confounding affects of reporting bias. Recently Elliott and Tefferi (2003) restricted their analysis to series comprising at least six patients but interestingly, they reported similar pregnancy outcomes. In all these reports maternal complications are relatively rare and no fatal events have been documented. Thrombosis is usually minor, but major thromboses have also been reported, including sagittal sinus thrombosis, deep vein thrombosis and transient ischaemic attacks (Vantroyen & Vanstraelen, 2002). Importantly, many of these events occurred in the postpartum period, emphasizing the need for robust postpartum prophylaxis. Haemorrhage was generally minor occurring in 4–5%; although three major bleeding events were reported (Bangerter et al, 2000) none of these patients were treated with aspirin but all had acquired von Willebrand's disease. The apparent low risk of maternal complications should also be considered in context as the majority of these patients would be viewed as ‘low risk’ for general ET management, suggesting that pregnancy does indeed increase the risk of thrombosis as would be anticipated.

The ET literature, whilst providing an indication of risks of various pregnancy outcomes, does not facilitate the identification of risk factors for pregnancy-related events. Some authors suggest that, for example, the occurrence of an earlier pregnancy event predicts a future one (Williams et al, 1994) but other authors would dispute this (Bangerter et al, 2000; Wright & Tefferi, 2001). However, given the expanding literature for pregnancy management in both antiphospholipid syndrome and/or inherited thrombophilia favouring active treatment of patients with prior pregnancy events, it would be difficult to justify not offering ET patients with previous pregnancy complications similar therapy. Although the largest series of patients published to date failed to provide support for a beneficial influence of aspirin therapy even in patients with pregnancy losses (Wright & Tefferi, 2001); others have suggested that aspirin from the outset of pregnancy with heparin from the second trimester might improve pregnancy outcome (Pagliaro et al, 1996). Overall, Griesshammer et al (2003) suggested that the success rate in 68 patients treated with aspirin (some in combination with other drugs) was 46 of 68 (68%) live births. In contrast, no therapy in 65 pregnancies was associated with 31 live births and 34 miscarriages. This analysis statistically supported the use of aspirin (P = 0·02, chi-square test). No studies have evaluated the role of heparin throughout a high-risk pregnancy thus it is unclear how successful such a strategy might be. The very nature of the literature and highly varied management strategies make it exceedingly difficult to formulate a clear conclusion.

Interferon alpha (IFNα), as discussed later in this article, is the cytoreductive agent of choice for those patients requiring control of their platelet counts. Although it does not appear to cross the placenta it is probably excreted in breast milk (Kumar et al, 2000) and hence breast feeding is contra-indicated whilst using IFNα. Some authors have suggested that for ET patients with prior pregnancy events IFNα might improve the outcome in any subsequent pregnancies (Williams et al, 1994; Schmidt et al, 1998) and a recent review suggested a similar benefit (Vantroyen & Vanstraelen, 2002). This is however based upon uncontrolled retrospective analysis of 25 patients in whom IFNα treatment was reported and only continued through pregnancy in 20 of 25 cases. Despite a very small number of patients it appears that IFNα therapy reduced both ET complications during pregnancy and also apparently improved pregnancy outcome (Petit et al, 1992; Thornley & Manoharan, 1994; Williams et al, 1994; Pulik et al, 1996; Shpilberg et al, 1996; Schmidt et al, 1998; Cincotta et al, 2000; Vantroyen & Vanstraelen, 2002). It has also been suggested that treatment with hydroxyurea facilitated successful pregnancy outcome in a patient with a history of two stillbirths in the third trimester (Cinkotai et al, 1994). This data suggest that cytoreductive therapy might potentially protect against fetal loss but remains unproven.

Polycythaemia vera

Polycythaemia vera is unusual in young patients; overall 15% of patients are aged 40 years or below at presentation; furthermore, PV is also a disease with a definite male preponderance (McNally et al, 1999) hence pregnancy is an uncommon occurrence in this patient population. We recently reviewed our experience of pregnancy in this disorder (eight patients and 16 pregnancies) and tried to assess the role of therapy and its possible influence upon pregnancy outcome (Robinson et al, 2004). The previous literature; excluding three reports in which extremely scanty details of maternal status and pregnancy outcome were provided (Lawrence, 1955; Erf, 1956; Diamandopoulous & Hertz, 1963); contains nine reports of 20 pregnancies (Hochman & Stein, 1961; Ruch & Klein, 1964; Centrone et al, 1967; Harris & Conrad, 1967; Ferguson et al, 1983; Crowley et al, 1987; Ruggeri et al, 2001; Subtil et al, 2001; Pata et al, 2004). Of these 20 pregnancies there were only 12 live births; unfortunately three of 12 suffered an early neonatal death. Many of these early reports (12/20 pregnancies) were published in or before 1968; prior to or during the important studies of the Polycythaemia Vera Study Group, which significantly informed management of this disease.

The literature for PV and pregnancy outcome is summarized in Table I. A total of 36 pregnancies have been reported in 18 patients and there was a live birth rate of 21 of 36 (58%), although this includes three episodes of early neonatal death, thus the surviving neonatal rate is 18 of 36. In common with ET, first trimester loss was the frequent complication in eight of 36 (22%), late pregnancy loss and IUGR occurred in seven (19·4%) and preterm delivery in five (13·8%). Maternal morbidity was significant. There was one death due to disseminated intravascular coagulation, deep vein thrombosis, pulmonary emboli, and saggital sinus thrombosis in a patient after elective termination of pregnancy; two postpartum pulmonary emboli, and one large postpartum haemorrhage; and interestingly four patients had pre-eclampsia of variable severity. For our own patients (Robinson et al, 2004) aggressive management following the protocol suggested below appeared to significantly improve the prospects of a live birth (P = 0·0056) and was not associated with significant maternal morbidity. When this analysis was extended to include other patients in the literature, again active individually tailored management appeared beneficial (P = 0·0094). Clearly, these conclusions are confounded by the limited patient numbers and an assumption that prior poor outcome supports intensification of therapy – which remains unproven for ET where there is a more substantial literature.

Table I.  Reported cases of pregnancy in patients with Polycythaemia Vera.
ReferenceNo. of pointsNo. of pregnanciesPrevious thrombosisPrevious haemorrhageTreatment prepregnancyTreatment during pregnancyHigh riskMaternal outcomeLive birth totalPregnancy loss totalFTMStill birth (gestation)IUGRPlacental abruptionLive birth premature delivery <37/40Live birth FTD
  1. FTM first trimester miscarriage, IUGR intra-uterine growth retardation, FTD full term delivery, NND neonatal death, STOP therapeutic termination, PPH, postpartum haemorrhage, PET pre-eclampsia.

  2. *The patient died with evidence of deep vein thrombosis, pulmonary emboli, saggital sinus thrombosis, and disseminated intravascular coagulation.

  3. **Aspirin and postpartum heparin in second pregnancy, low molecular weight heparin and aspirin throughout third pregnancy.

  4. ***Multiple placental infarcts in first, and abnormal Doppler waveforms and severe IUGR in third pregnancy.

  5. ****One singleton and one twin pregnancy (this patient had previously had a twin pregnancy terminating in intra-uterine death at 22 weeks, followed by two normal pregnancies and then the diagnosis of PV).

Crowley et al (1987)11NoNoAspirin +  dipyrimadoleAspirin +  dipyrimadoleNoDeath*011 STOP00000
Centrone et al (1967)13NoNoNilNilNoAlive12200001
Ferguson et al (1983)12NoNoVenesectionNilNoAlive PET20000002 PET
Ruch and Klein (1964)12Superficial thrombo -phlebitisNoVenesection and 32PNoneNoAlive PET1101 (35/40) PET1001
Subtil et al (2001)13NoNoVenesectionAspirin, heparin** venesectionNoAlive PE postpartum1202 (24/40 and 28/40)***201 (32/40)0
Hochman and Stein (1961)14Yes, CVANoVenesection TBI prior to third pregnancyNilYesAlive PET2202 (5 + 7  months) PET001 (7 months, PET), 1 (8 months)0
Harris and Conrad (1967)22****NoNoNilNilNoAlive PPH2****0000002
Ruggeri et al (2001)12NoNoVenesection +  aspirinHeparin 3/52 PostpartumNoAlive, PE 24/7 postpartum11100001
Pata et al (2004)11NoNoHydroxyureaHydroxyurea 9/40 then nilNoAlive10000001
Robinson et al (2004)816 (one twin)Yes (one patient)NoVenesection, aspirin, interferon, hydroxyureaVaried: Venesection, Aspirin, Interferon LMWH, vitamin C + ENoAlive PET in 19642401 (34/40, IUGR), 1 (36/40), 1 26/40) (NND)7
Total18361 CVA
1 Thrombo -phlebitis
None  1 Yes1 Death
4 PET
2 PE
1 PPH
21
3 Neonatal deaths
158770615

Myelofibrosis

Myelofibrosis is the MPD that is least common in patients of child bearing age (McNally et al, 1999) and, combined with its worse prognosis, probably explains the paucity of published information for pregnancy in this condition. Indeed the literature comprises only four pregnancies in two patients (Taylor et al, 1992; Gotic et al, 2001) and is summarized in Table II. It is difficult to interpret such limited data but the parallels with pregnancy outcome in ET and PV are apparent and the pathophysiology is likely to be similar.

Table II.  Reported cases of pregnancy in patients with Idiopathic myelofibrosis.
ReferencePatientPregnancyPrevious thrombosisPrevious haemorrhageTreatment prepregnancyTreatment during pregnancyLive birthMaternal outcomeFTMStill birth (gestation)IUGRPlacental abruptionLive birth premature delivery <37 of 40
Taylor et al (1992)11NoNoSupportive onlySupportive1No complications0000Elective induction 36 of 40
Gotic et al (2001)11NoNoNoneNone0No complications030 (placental infarctions)000
2Placental infarctionsNoNoNone0No complications027 (placental infarctions)000
3Placental infarctionsNoInterferon αInterferon α1No complications0000Elective delivery 34 of 40 because of placental insufficiency
Total24    2 02002

Options for the management of MPD in pregnancy

As already discussed, there is only limited published data that is informative specifically in relation to the management of MPDs in pregnancy. The recommendations presented here are based on current knowledge of PV, ET and the management of both inherited thrombophilia and the antiphospholipid syndrome, which all appear to have placental dysfunction in common. Further audit or other analysis is required to substantiate this or any management strategy for MPDs in pregnancy. A further research agenda would include the further value of screening for other inherited or acquired thrombophilia and clonal analysis using, for example, X-chromosome inactivation patterns, in MPD patients and whether this has any impact in outcome of pregnancy or would dictate a change in its management.

These patients should ideally be under joint care of a consultant obstetrician experienced in the care of patients with high-risk pregnancies and a haematologist. A preconception-planning meeting should be held to discuss a plan of management for pregnancy. During the pregnancy the patient should be monitored regularly (Fig 1). In addition to standard fetal assessment by ultrasound, it is recommended that uterine artery Doppler studies should be performed at 20 and 24 weeks (as discussed below). Fetal growth monitoring is advised at least 4-weekly, or more frequently according to the progress of the pregnancy.

Figure 1.

Summary of Pregnancy Management and LMWH doses, FBC, full blood count; wks, weeks; wkly, weekly; BP, blood pressure; USS, ultra sound scan; RI, resistance index; LMWH, low molecular weight heparin; PCV, packed cell volume.

Therapeutic strategies

Therapeutic strategies for MPDs in pregnancy are arguably, but most likely to be, influenced by the patients’ disease status and prior obstetric history. If any of the following factors are present then the pregnancy is likely to be at high risk of complication to the mother and/or fetus:

  • 1previous venous or arterial thrombosis in mother (whether pregnant or not);
  • 2previous haemorrhage attributed to MPD (whether pregnant or not);
  • 3previous pregnancy complication that may have been caused by MPD; e.g.
    • (i) ≥3 first trimester or≥1 second or third trimester pregnancy loss,
    • (ii) birth weight <5th centile for gestation,
    • (iii) intrauterine death or still birth (with no obvious other cause, or evidence of placental dysfunction and growth restricted fetus),
    • (iv) significant ante-partum haemorrhage,
    • (v) postpartum haemorrhage (requiring red cell transfusion),
    • (vi) severe pre-eclampsia (necessitating preterm delivery <37 weeks), or
    • (vii) development of any such complication in the index pregnancy,
  • 4platelet count rising to >1,500 × 109/l.

The most controversial areas listed here include those events in the index or prior pregnancy for reasons previously discussed. Selecting a target platelet threshold for cytoreductive treatment is also contentious; other authors have suggested 1000–1500 × 109/l (Griesshammer et al, 2003; Barbui et al, 2004). The risks to the unusual patient who has an acquired von Willebrand's syndrome secondary to ET and their subsequent management in pregnancy are unclear; it is likely, however, that the expected natural decline of the platelet count by 350–450 × 109/l during pregnancy would reduce the severity of any acquired von Willebrand's syndrome (Wright & Tefferi, 2001).

Therapeutic options include antithrombotic treatment, venesection and cytoreductive agents; although the expected natural fall of the platelet count and haematocrit (Hct) during pregnancy may anyway obviate or reduce the need for the latter. The Hct could be controlled with either careful venesection or cytoreductive therapy. The target Hct for a non-pregnant female has yet to be determined, but a reasonable level would probably be in the middle of the gestation appropriate range (Bain, 1995). There is currently no evidence for maintaining the Hct less than this in pregnancy although this has been controversial (Spivak, 2002; McMullin et al, 2003). Cytoreduction should preferably be avoided in pregnancy, particularly in the first trimester. None of the cytoreductive agents currently used in treatment of MPDs has a product licence for use in pregnancy. The cytoreductive agent of choice, as discussed already, is probably IFNα.

Interferon α

The interferons are naturally occurring protein-like macromolecules with anti-viral and cytokine-like actions. IFNα seems to be important in maintaining human pregnancy (Chard, 1991) whether by virtue of its actions upon cell proliferation, differentiation or in maternal recognition of pregnancy is unclear. There are no reports of teratogenic effects in animals at standard doses but, when given at 20–50-fold the recommended human dose, an increase in abortions was documented in rhesus monkeys. Conversely its administration to sheep prior to conception resulted in an increased number of pregnant animals and increased embryonic survival (Briggs et al, 2002). There is a significant literature documenting the apparently safe use of IFNα in patients with a variety of haematological conditions, including chronic myeloid leukaemia, hairy cell leukaemia and ET, as well as for hepatitis C. No adverse effects in the admittedly small numbers of human pregnancies exposed to this drug have been reported to date but it should still only be used with caution (Briggs et al, 2002). However, some evidence suggests that IFNα may decrease fertility (Griesshammer et al, 1998) and hence it may be best avoided in women with difficulty conceiving. Data from a number of uncontrolled retrospective analyses have led authors to suggest IFNα might have improved the chance of a successful pregnancy outcome (reviewed in Vantroyen & Vanstraelen, 2002). Clearly, this is controversial and many patients have had successful pregnancy outcomes with minimal therapy. One case of multiple fetal abnormalities (ambiguous genitalia, hemivertebrae D6 to D10 with 11 pairs of ribs and secondaryscoliosis) and intra-uterine growth retardation has been reported in an ET patient taking hydroxyurea and aspirin at conception and IFNα after the fourth month (Perez-Encinas et al, 1994) and one case of intra-uterine growth retardation associated with placental infarction (Pardini et al, 1993). Interestingly the abnormalities described by Perez-Encinas et al, 1994) are perhaps more in keeping with those described in animals treated with hydroxyurea (see below) and perhaps relate to the latter agent rather than IFNα.

Hydroxyurea (or hydroxycarbamide)

Hydroxyurea is also known, and now marketed, as hydroxycarbamide. Its therapeutic actions were first reported when Rosenthal et al (1928) documented leucopenia, megaloblastic anaemia and death following administration to animals. This is the only drug that has been shown in a randomized controlled trial to reduce complications in high risk patients with ET (Cortelazzo et al, 1995). It has a number of therapeutic actions but a primary effect is the inhibition of the enzyme ribonucleotide reductase and is a small molecule capable of crossing the placenta. Embryotoxicity of hydroxyurea has been documented in many animal species including rats, rabbits and rhesus monkeys (Butcher et al, 1973; Wilson et al, 1975; Theisen, 1979; Aliverti et al, 1980; Iwama et al, 1983; Asano & Okaniwa, 1987). Fetal abnormalities included partially ossified cranial bones, absent orbital sockets, hydrocephaly, missing lumbar vertebrae in rabbits, and similar abnormalities were found in rhesus monkeys where, in addition, growth retardation and increased spontaneous abortion were documented (Theisen, 1979). These toxicities were however observed at doses of 10- to 100-fold those used in humans.

A comprehensive search of the medical literature revealed at total of 55 pregnancies where hydroxyurea use had been documented (Doney et al, 1979; Charache et al, 1987; Patel et al, 1991; Delmer et al, 1992; Tertian et al, 1992; Fitzgerald & McCann, 1993; Jackson et al, 1993; Szanto & Kovacs, 1993; Cinkotai et al, 1994; Fernandez, 1994; Perez-Encinas et al, 1994; Dell'Isola et al, 1997; Byrd et al, 1999; Diav-Citrin et al, 1999; Baykal et al, 2000; Celiloglu et al, 2000; Thauvin-Robinet et al, 2001; Fadilah et al, 2002; Koh et al, 2002; Pata et al, 2004). Overall, eight pregnancies were therapeutically terminated, and one was terminated for unstated medical reasons. Of the remaining 46 pregnancies there were 42 live births (including 10 premature, one IUGR, one pre-eclampsia, and one IUGR with multiple fetal abnormalities) and four intra-uterine deaths. Only one of these infants had major abnormalities documented (Perez-Encinas et al, 1994), perhaps more likely to be due to hydroxyurea exposure although this patient was also exposed to IFNα. The minor abnormalities reported were seen at a frequency expected in the normal population (Thauvin-Robinet et al, 2001). Although none of these infants had been followed for a period extending beyond a year, it is important to highlight the potential for long-term toxicity including secondary malignancies as well as other effects in these offspring. The largest series of 31 patients (Thauvin-Robinet et al, 2001) reported a significant number of pregnancies complicated by IUGR, intra-uterine death and premature infants but also highlighted that these complications are common in pregnancies associated with the haematological conditions the hydroxyurea was used to treat. In this series karyotypic analysis was reported in seven patients; six were normal and an inherited inversion of chromosome 9 was demonstrated in the seventh.

Hydroxyurea thus appears to be less toxic than might be anticipated in human pregnancies. Whilst conception with concurrent hydroxyurea should be avoided, when it occurs the short-term risks appear relatively limited. More data should be collected in relation to these pregnancies and the long-term health of infants exposed in utero. The manufacturer (Bristol Meyers Squibb, New York, NY, USA) and other authors (Halsey & Roberts, 2003) suggest that a period of 3–6 months should elapse between the last dose of hydroxyurea and planned conception. For high risk ET patients requiring cytoreductive therapy, the strategy would then be a planned switch to IFNα, allowing a 3–6 months ‘wash-out’ period.

Anagrelide

Anagrelide is not recommended during pregnancy as the compound is small enough to cross the placenta and induce thrombocytopenia in the fetus. A normal pregnancy and spontaneous first trimester abortion have however occurred after exposure to anagrelide in early pregnancy (Wright & Tefferi, 2001).

Plateletpheresis

A limited number of pregnancies in patients with ET who were treated with plateletpheresis have been reported (Falconer et al, 1987; Mercer et al, 1988; Beard et al, 1991; Beressi et al, 1995; Koh et al, 2002). In most, but not all, patients this successfully reduced the platelet count (Beressi et al, 1995) although it did not always result in a successful outcome (Koh et al, 2002). However, this is a cumbersome technique, which demands central venous access, frequent procedures (twice weekly) and has only a transient effect. In this clinical context, it is safe for the fetus. In general, plateletpheresis has a limited role in the management of an acute, life-threatening event where very rapid control of the platelet count is required. Plateletpheresis could be considered in patients at high risk of thrombosis who are either intolerant of, or fail to respond adequately to, IFNα. However, it may be that hydroxyurea, particularly given after the first trimester, is also a reasonable, more convenient albeit difficult choice that needs to be fully and carefully discussed with the patient.

Aspirin

Low dose aspirin is safe in pregnancy, as demonstrated by the Collaborative Low-dose Aspirin Study in Pregnancy (CLASP) Collaborative Group (1994). In 32 randomized controlled trials, including nearly 30 000 women, low dose aspirin resulted in a 15% decrease in pre-eclampsia and a similar reduction in fetal death (Knight et al, 2000; Duley et al, 2001). Furthermore, the widespread use of low dose aspirin in non-pregnant patients with PV and possibly also other MPDs is supported by the recently reported European Collaboration on Low-dose Aspirin in Polycythaemia Vera (ECLAP) study (Landolfi et al, 2004). Based on data from small and retrospective studies, the use of low dose aspirin seems advantageous in pregnancy in MPD patients (Griesshammer et al, 1996, 1998, 2003; Barbui et al, 2004). The advantage of aspirin probably outweighs the recent preliminary suggestion that prenatal use of non-steroidal anti-inflammatory drugs, including aspirin, might significantly increase the risk of miscarriage (Li et al, 2003).

Therefore, in the absence of clear contraindications, all patients should probably be on aspirin (initially 75 mg o.d.) throughout the pregnancy and for at least 6 weeks after delivery. For patients with on-going symptoms the aspirin resistance should be confirmed, for example, by use of the PFA-100 (Sysmex, Milton Keynes, UK). The aspirin dose can then be either increased or an alternative antiplatelet drug, such as clopidogrel, substituted. The successful use of clopidogrel in pregnancy has been reported (Klinzing et al, 2001) in an ET patient whose previous history included myocardial infarction and coronary artery bypass grafting; no sequelae were reported for the fetus. However, this is an isolated report and the use of clopidogrel in pregnancy cannot be recommended on this basis.

However, aspirin alone may not be the optimum therapy. A recent study in patients with either factor V Leiden, prothrombin P20210A mutation, or protein S deficiency and one fetal loss demonstrated that aspirin was apparently inferior to low molecular weight heparin (LMWH) (enoxaparin 40 mg) in terms of live birth rate, and birth weight (Gris et al, 2004). This data clearly cannot be extrapolated directly to MPD patients where the acquired prothrombotic state relates to a combination of blood rheology, possibly leucocyte adhesion, platelet activation and platelet-leucocyte aggregates (Falanga et al, 2000; Jensen et al, 2001) and in whom aspirin in the non-pregnant state has been shown to protect against life-threatening thrombosis (Landolfi et al, 2004). Nonetheless, for MPD patients with prior fetal loss this would suggest that LMWH should be considered in addition to aspirin.

Thrombo-embolic deterrent stockings

The British Society for Haematology guidelines (Walker et al, 2001) give a grade C recommendation (evidence level IV) that all women with previous venous thromboembolism or a thrombophilia should be encouraged to wear Thrombo-embolic deterrent stockings (TEDS) throughout their pregnancy and for 6–12 weeks after delivery. The use of TEDS is also recommended for pregnant women travelling by air (Royal College of Obstetricians and Gynaecologists, 2001).

Heparin

Low molecular weight heparin (LMWH) has been extensively and successfully used not only in pregnancies at high risk of thrombosis (Hunt et al, 1997, 2003) but also in reducing fetal morbidly (Rai et al, 1997; Gris et al, 2004). LMWH is probably safer than unfractionated heparin for it has a lower risk of heparin-induced thrombocytopenia and osteoporosis whilst likely to be equally efficacious, as supported by a relatively recent meta-analysis (Sanson et al, 1999). Recently, a problem with the therapeutic range of unfractionated heparin in pregnancy has been reported, as the response (activated partial thromboplastin time) to heparin is attenuated due to pregnancy-induced changes of heparin binding proteins and factor VIII; whether this translates into excess bleeding is currently unclear (Chunilal et al, 2002). Interestingly, although LMWH and conventional unfractionated heparin appear to be equally effective clinically, in vitro data suggested LMWH may be less potent at stimulating trophoblast differentiation and mobility (Quenby et al, 2004). However, LMWH remains the heparin of choice for this clinical indication.

LMWH has been used anecdotally in women with MPD and fetal morbidity and/or previous thrombosis. The use of unmonitored intermediate dose LMWH is widely used, e.g. enoxaparin 40 mg once daily increased to 40 mg twice daily from 16 weeks, dropping to 40 mg daily for 6 weeks postpartum. If another risk factor occurs, e.g. immobility, hyper emesis, ovarian hyperstimulation, or surgery then the use of increased dose subcutaneous LMWH during pregnancy should be considered. A suggested dosing schedule for LMWH is summarized in Fig 1.

Uterine-artery Doppler scanning

Uterine artery Doppler scanning at 20 and 24 weeks will reveal whether the woman has a high resistance index and/or bilateral notching. Bilateral persistent notching in particular indicates there is increased resistance to flow and thus possible placental dysfunction or impaired trophoblast migration. There are acceptable levels of inter-observer variation (Farrell et al, 1998). Thus, it selects all the women at risk of IUGR and pre-eclampsia, although many of these women have a normal pregnancy (reviewed in Axt-Fliedner et al, 2004). It has been estimated that the positive predictive value of this test is merely 20% and the value of screening patients at low risk has yet to be established. This technique has the advantage that it potentially allows targeting of these high-risk women for more intensive monitoring and consideration for adding or escalating the dose of LMWH or vitamin supplementation as below. For example, in five randomized controlled trials including women with abnormal uterine artery Doppler ultrasounds the reduction in relative risk of pre-eclampsia with aspirin use approached 45% (Coomarasamy et al, 2001).

Antioxidants

There is a significant body of evidence supporting the conclusion that pre-eclampsia partly results from widespread endothelial dysfunction. The potential role of an imbalance in free-radicals and reactive oxidative species (such as nitric oxide or super oxide anion) in the endothelial manifestations of pre-eclampsia has attracted considerable attention (reviewed in Bilodeau & Hubel, 2003). This has led to interest in the use of anti-oxidants in pregnancies at high risk of pre-eclampsia. A small but randomized study of supplementation with antioxidant vitamins C (1000 mg/d) and E (400 IU/d) in patients at high risk of pre-eclampsia by virtue of either previous history or abnormal two-stage uterine-artery Doppler scans suggested beneficial reduction in the incidence of pre-eclampsia of over 50% (odds ratio 0·24) and evidence of reduced endothelial activation, as judged by plasminogen activator inhibitor levels (Chappell et al, 1999, 2002). The benefit of such supplementation may not relate to oxidative balance. A larger multi centre study is currently underway (Bilodeau & Hubel, 2003). There is no clear evidence for the role of antioxidant supplementation in MPD patients (only one patient was treated in this way in our institution with successful outcome). However, its targeted use in MPD patients in whom there is evidence suggestive of likely poor fetal outcome is possibly warranted albeit unproven; further data should be collected.

Delivery

It is important to discuss the implications of the use of thromboprophylaxis for epidural or spinal anaesthesia prior to labour or caesarean section with the patient and, ideally, an obstetric anaesthetist. During labour dehydration should be avoided, attention should be given to dosing of LMWH (as below) and the use of TEDS should be considered. To minimize or avoid the risk of epidural haematoma, some anaesthetists prefer aspirin to be stopped 2 weeks prior to delivery. Regional anaesthetic techniques should not be used until 12 h after the previous prophylactic dose of LMWH. Following treatment doses of LMWH, regional anaesthesia should not be employed for at least 24 h. LMWH should not be given for at least 4 h after the epidural catheter has been removed and the cannula should not be removed within 10–12 h of the most recent injection (Horlocker & Wedel, 1998; Checketts & Wildsmith, 1999). For delivery by elective caesarean section, the woman should receive a prophylactic dose of LMWH on the day prior to delivery. On the day of delivery, the morning dose should be omitted and the operation performed that morning. The prophylactic dose of LMWH should be given by 3 h postoperatively (over 4 h after removal of the epidural catheter, if appropriate).

Postpartum therapy

The time of greatest risk for venous thromboembolism associated with pregnancy is the immediate puerperium. The prothrombotic changes of pregnancy do not revert completely to normal until 6 weeks after delivery. Hence puerperal thromboprophylaxis, usually in the form of aspirin and LMWH, should be continued for 6 weeks. The first dose should be given as soon as possible after delivery provided there is no bleeding and no epidural catheter (as above). Breast-feeding is safe with both heparin and warfarin (providing baby receives adequate vitamin K). Breast-feeding is, however, contra-indicated with the cytoreductive agents (IFNα, anagrelide and hydroxyurea). During the early postpartum period blood counts may also rise rapidly, thus on-going haematological monitoring is important.

Paternity for MPD patients

It is unusual for an increased risk of congenital malformations to be associated with paternal exposure to drugs or chemicals unless point mutations occur. Even if dominant lethal mutations occur as a result of drug exposure this can be difficult to distinguish from the background risk of spontaneous abortion. The commoner adverse reproductive effects of drugs in men are loss of libido, infertility or abnormal sperm counts and morphology. Provided the damage is not severe there is a chance of a return to normal spermatogenesis and once this has occurred there is no published evidence to suggest that there is an increased risk of fetal toxicity in future pregnancies. To date, there have been no reports of hydroxyurea-treated males having fathered a child with genetic abnormalilties. But it has not been established whether other risks exist for the fetus. The summary of product characteristics for hydroxyurea (available at http://www.medicines.org.uk) suggests that those patients receiving hydroxyurea should withdraw the drug, substituting another if required, and aim to have had their last dose for at least 3 months before unprotected intercourse. This suggestion is most likely to be based upon the observation that mature sperm are finally released from Sertoli cells approximately 74 d after their initial development from spermatogonia (Bustos-Obregon et al, 1975). Both anagrelide and IFNα are acceptable therapies as alternatives to hydroxyurea.

Conclusions

Pregnancy is an uncommon clinical occurrence in the MPD patient population, however, there is a need to be aware of the options for patients who are considering pregnancy. There are many lessons and much practical data to consider from the expanding literature in relation to other prothrombotic states and pregnancy, particularly antiphospholipid syndrome and the inherited thrombophilias. Whilst it is clear that therapeutic strategies are not necessarily directly transferable to the MPD population, some of whom will have successful uneventful pregnancy without additional intervention. For those patients with previous pregnancy complications, particularly those of a serious nature, therapeutic options include aspirin, LMWH and IFNα. The answers to several key questions are still lacking: who needs treatment, when will they need it and what is the optimum therapy? There is a need for international collaboration to address these issues and provide the best care for patients.

Ancillary