Antiphospholipid antibodies and pregnancy loss [Mar 2002; 97-2]

Antiphospholipid antibodies and pregnancy loss
Systematic review Results Comment

What's worse than being asked a question to which you don't know the answer (or even understand the question)? It could be having only half of the answer, but not knowing which half. Such a position could arise with treatments for antiphospholipid antibodies in pregnancy.

Antiphospholipid antibodies are antibodies directed against several phospholipids in the body. There is an association between antiphospholipid antibodies in the circulation and pregnancy loss, and between 3% and 7% of pregnant women have the antibodies. In low risk pregnancies, the antibodies are associated with a nine-fold increase in pregnancy loss, while in high risk pregnancies with at least three previous losses, they are associated with a 90% risk of further pregnancy loss.

The mechanism of pregnancy loss is thought to be through thrombosis of placental vessels, though this is a complicated, and perhaps controversial area. The important question is whether there are effective treatments. A new systematic review [1] begins to answer that.

Systematic review

The searching strategy for this review was impressive, using the usual electronic databases, plus the Cochrane controlled trials register, plus hand-searching of specialist journals and abstracts of relevant symposia. Trials sought were those that sought to prevent pregnancy loss in pregnant women with a history of at least one previous loss and serological evidence of antiphospholipid antibodies. The primary outcome sought from the trials was pregnancy loss, though many others, including birth weight, prematurity and even issues around maternal bone mineral density were looked for.

Results

There were 10 randomised or quasi-randomised trials included with 627 women, and their design and results are given in detail in the review. In most trials women had at least two and often three previous miscarriages, and both treatments and the results of serology tests are described. Trials were not large, though, with 90 women being the biggest, and some were very small. Trial design was generally adequate, most being properly randomised, some blind, and with intention to treat analysis and 100% follow up in all. The biggest difficulty was the plethora of different treatments used, from aspirin alone, to heparin plus aspirin, prednisolone plus aspirin or intravenous immunoglobulin.

In direct comparisons, aspirin alone was no better than placebo or usual care, with the important qualification that these were small trials with only 70 women in total, and with what appeared to be a lowish rate of loss without treatment. Heparin and aspirin was better than aspirin in two trials with 140 patients, with a relative risk of loss of 0.45 (95% CI 0.29 to 0.70) and number needed to treat of 3.2 (2.1 to 6.3). A trial of high dose heparin versus low dose heparin, both with aspirin, showed no difference, and with rates of pregnancy loss consistent with the comparisons of heparin plus aspirin with aspirin.

A number of the trials had treatment arms that included usual care, aspirin, or heparin plus aspirin. The results in those women are shown in Figure 1 and Table 1. Heparin plus aspirin resulted in lower rates of pregnancy loss, though with wide variations in these small numbers.

Figure 1: Abacus plot of single trial arms for pregnancy loss with usual care, aspirin, and heparin plus aspirin

Table 1: Numbers and percentages of pregnancy losses with different treatments

Intervention	Trial arms	Losses/total pregnancies	Percent (95%CI)
None, usual care	4	30/80	37 (27 to 48)
Aspirin alone	6	50/129	39 (30 to 47)
Low MW heparin plus aspirin	6	36/155	23 (17 to 30)

One reasonably solid conclusion from other outcomes was that use of prednisolone was associated with higher rates of premature birth and admission to neonatal intensive care units. There was no evidence that it helped prevent pregnancy loss. For other treatments there were few secondary events like premature delivery, and no indication that treatment made them occur more frequently.

Comment

When a young woman asks whether any treatment is effective for recurrent pregnancy loss because of antiphospholipid antibodies, the immediate reaction for most of us will be a big question mark in a bubble coming out of our head. Last year it would have been almost impossible to answer. This year, because we now have a systematic review, it's just difficult.

In two direct comparisons, heparin and aspirin beat aspirin alone, with a NNT that we usually regard as being impressive. Looking at all trials where heparin plus aspirin, aspirin alone, or usual care have been used, heparin plus aspirin again looks a better bet. We have to acknowledge that this way of looking at available information is nothing like as conclusive as direct comparisons, but where information is scarce and numbers low, it is useful to bolster confidence in the conclusion. Prednisolone is not an option.

A more generic lesson is that we need additional confidence when numbers are low and quality uncertain, and when it is hard to judge validity becuase the territory is unfamiliar. In Figure 1, with both aspirin and usual care some trials reported no pregnancy losses while others reported more than 80% losses. Of course they were small, some treatment arms with information on as few as six women.

Clever people have looked at the uncertainties that can arise because of trial quality and trial size. The lessons for this example would be to be cautious. The trouble is that if you are a prospective mother or their professional advisor, you don't have a decade or so to wait for more or better trials to be performed. You have to make a decision now. There may be confidence in numbers and quality, but in their absence, or, as here, when they are in short supply, wisdom and experience will have to make up the deficit.

References:

M Empson et al. Recurrent pregnancy loss with antiphospholipid antibody: a systematic review of therapeutic trials. Obstetrics & Gynecology 2002 99: 135-144).

previous or next story in this issue