The paper by Gomez et al.1 provides further evidence that late first-trimester screening of the uterine arteries by Doppler ultrasonography can identify women who will subsequently develop pre-eclampsia and intrauterine growth restriction (PE/IUGR). The authors make a statement unsupported by any discussion that ‘the use of a single uterine Doppler measurement for screening purposes in unselected early pregnancy populations has limited clinical value’. It is this statement I wish to explore in this Opinion. To make my arguments concise, I will confine my remarks to the use of Doppler ultrasonography to predict PE only.
The criteria for a satisfactory screening test with which most of us are familiar are those published by the World Health Organization (WHO), which are themselves based on those produced by Wilson and Junger2. However I prefer to use the ‘requirements for a worthwhile screening program’ by Cuckle and Wald (listed below), for they are concise and more pertinent to a pregnancy-screening program3.
- 1)Disorder: well defined;
- 2)Prevalence: known;
- 3)Natural history: medically important disorder with effective remedy available;
- 4)Financial: cost effective;
- 5)Facilities: available or easily installed;
- 6)Acceptability: procedures following positive result agreed and acceptable to authorities and patients;
- 7)Equity: equal access to screening services;
- 8)Test: simple and safe; and
- 9)Test performance: distributions of test values in affected and unaffected individuals known, extent of overlap small, suitable cut-off defined.
I will not pedantically go through each of these criteria in relation to early uterine artery screening but I will highlight the more important issues as regards the validity of this test as a predictor for PE.
Firstly PE is a condition that has been defined and re-defined ad nauseam based on the three basic signs of hypertension, proteinuria and fluid retention. The first two of these signs are usually quantified to give an assessment of the severity of the condition. However it is only severe PE that should be of any interest for a screening program for it is only this condition that is associated with a significant increase in maternal and perinatal mortality and morbidity4. I also believe that we should discard the various definitions based on maximum sustained blood pressure and g proteinuria/24 h, as this information is frequently mis-reported in the patient's records or not identified during audit.
What is recorded accurately is the gestational age at delivery, which is related to the severity of the PE, the gestational age at onset of the condition and the status of the fetus, all of which are significantly related to maternal and perinatal mortality and morbidity. Although the prevalence of severe PE remains high in some developing countries, in North America and Europe it has fallen to less than one per cent of all pregnancies. In the study of Martin et al.5 the prevalence of PE requiring delivery before 36 weeks' gestation (which I would define as severe) was 0.6% and before 32 weeks (which I define as very severe) was 0.36%. In other words severe PE occurs in about 1 in 200 gestations and very severe PE in about 1 in 300 gestations. Is this a high enough prevalence to justify antenatal screening? The answer must be ‘yes’. Severe PE is a major hazard for mother and fetus and anyway there are many conditions screened for routinely with a lower prevalence than even severe PE. For example Down syndrome screening is practiced routinely in most developed countries, where the prevalence of Down syndrome is 1 in 600 gestations—which is about half the prevalence of very severe PE.
What then about the acceptability, safety and simplicity of the test? Gomez et al.1 performed their uterine artery studies by transvaginal examination, and while this method is regarded as safe it is intrusive as a screening method and in my opinion does not come under the heading of ‘simple’. It is difficult to understand why Gomez et al. used this approach when Martin et al.5 and Hollis et al.6 have shown transabdominal Doppler ultrasonography to be an effective and reproducible method of measuring uterine artery pulsatility index (PI) or resistance index (RI). I do not agree with Gomez et al.1 that a full bladder is required to identify the uterine artery transabdominally in the first trimester. In fact irrespective of whether the bladder is full or empty, both main uterine arteries are usually easily identifiable by transabdominal examination at a low angle of incidence to the beam, thereby ensuring accurate measurement. Thus transabdominal examination of the uterine artery in the first trimester comes under the heading of acceptable, simple and safe.
Now we come to the big question—is the performance good enough? Most studies using uterine artery Doppler ultrasonography as a screening method have accepted a false positive rate of 5%, which is the figure generally used for Down-syndrome screening. This false positive rate is generally favored in pregnancy screening programs as it is not so high as to induce anxiety in a large number of unaffected pregnant women but sufficiently large (hopefully) to identify a significant percentage of the target condition. In fact a false positive in a Down syndrome screening program involves an invasive procedure with a risk of miscarriage; a false positive in a PE screening test only involves increased surveillance and possibly prophylactic low-dose aspirin therapy. Gomez et al.1 on this basis predicted 31% of pregnancies which subsequently developed severe complications. While this may seem to be a low detection rate, a confounding factor was the admittedly low prevalence of severe complications, which might be partly explained by the administration of aspirin to some of the abnormal Doppler study cases. The larger study of Martin et al.5 (again using a 5% false positive rate) demonstrated a 40% sensitivity for the subsequent development of severe PE (i.e. delivery at < 36 weeks' gestation) and a 60% sensitivity for very severe PE (delivery at < 32 weeks' gestation). This performance for severe PE compares favorably with the triple test (which has a similar sensitivity but lower prevalence and odds ratio), which is still the most widely used screening test for Down syndrome used in the UK7. I have never ever heard complaints that the triple test should not be implemented because of its performance, albeit it has now been superseded in performance terms by nuchal translucency and early biochemistry.
Another major consideration is the cost implications of early-pregnancy Doppler ultrasonography for PE/IUGR. Without going into any cost-effectiveness calculations I feel that the balance of the argument falls very strongly in favor of performing the test. When you consider that at the very same examination, screening for Down syndrome and monochorionic twins will be performed, it would seem almost wanton not to perform a 2-min test with the same equipment that would provide valuable and potentially life-saving information on a woman's risks of PE/IUGR. Surely standard concepts of the value of single screening tests go out of the window when they are used as part of a group of tests to triage women for more intensive surveillance. I would suggest that an additional criterion for an acceptable screening test would be ‘screening tests which can be “piggy-backed” onto existing tests, which do not involve significant extra costs, which do not have a high false positive rate and which will identify a cohort of at-risk patients who would benefit from increased surveillance’.
In terms of performance, uterine artery screening at 20–24 weeks' gestation is superior to first-trimester screening, and appears to fulfill all the requirements for a worthwhile screening test. In the study by Papageorghiou et al.8, the sensitivity for the prediction of severe PE and very severe PE was 77% and 93%, respectively, at 23 weeks' gestation according to the definitions outlined above. One would find it hard not to make a case for uterine artery Doppler screening at this stage, especially as the test can be added to the routine second-trimester scan at virtually no extra cost. Sadly in the UK, the government advisory body established to set standards in clinical practice (National Institute for Clinical Excellence (NICE)) does not recommend uterine artery screening, although the reasons why they prefer poorly performing indicators such as nulliparity, age over 40, family or previous history of PE and BMI > 35 are not adequately explained9.
An objection frequently raised to uterine artery screening is that as there is no effective preventative measure for PE, then there is no point in doing the test. I strongly disagree with this view. There is powerful evidence to show that increased surveillance in high-risk cases significantly improves outcome. For example the perinatal mortality in diabetic pregnancies managed in specialist tertiary centers is usually lower than the overall perinatal mortality for low-risk women because the intensive surveillance they receive is hugely beneficial. The study by Harrington et al.10 also illustrates this point. In an open study they randomly allocated 216 women with high RI and notches in the uterine arteries at a mean gestational age of 20 weeks into either intensive monitoring plus aspirin 100 mg or routine antenatal care. Although the incidence of PE was similar in both groups, there was a significant reduction of complications in the monitored group. With present knowledge it is unlikely that aspirin contributed much to this reduction, which was due, I believe, almost entirely to increased surveillance and change of lifestyle. Whatever effects aspirin or increased surveillance may have in preventing or delaying PE/IUGR they are more likely to be effective when started in the first trimester before the secondary wave of trophoblast invasion (i.e. along the spiral arterioles) has begun—hence the importance of early Doppler ultrasound screening.
Is there an alternative? If you have lived as long as I have you would have seen a lot of biochemical screening tests come and go. No one discusses fibronectin, β-hCG, AFP, inhibin A, activin A or even fetal DNA any more. The latest biochemical ‘kid on the block’ is circulating vascular endothelial growth factor (VEGF) and its soluble receptor sFLT-1. The receptor in particular appears to be elevated early in pregnancy in women who subsequently develop PE11. Whether the overlap between affected and unaffected individuals will be small enough to be the basis for a screening test remains to be seen. It should be borne in mind that all these markers, including Doppler, are merely reflections of the disease process and in no way represent the fundamental etiological defect which, it has recently been proposed12, may be due to defective epigenetic modifications of the LH/hCG receptor during spermatogenesis leading to haploinsufficiency of the receptor. This means that the placenta fails to react normally to hCG because of inadequate levels of its receptor. If this hypothesis is true then the basic defect behind PE begins at conception and the events we are studying are well down the time line in the disease process.
However, because we hope that eventually a biochemical marker will prove to be a useful early predictor of PE, this is no excuse for paralysis on implementing uterine artery screening. The technique is effective at defining a population of women who will develop severe/very severe PE/IUGR. It can be introduced at minimal cost by being ‘piggy-backed’ onto standard ultrasound screening examinations in the first and second trimesters. It will cause minimal disruption to current obstetric practice. There is no excuse for delay.