How does smoking reduce the risk of preeclampsia?

Although smoking during pregnancy may lead to many adverse effects such as fetal growth restriction, placental abruption, stillbirth, and preterm labor, smoking is the only environmental exposure known consistently to reduce the risk of preeclampsia and gestational hypertension¹. The manuscript by Wikström et al is a major step forward in understanding this protective effect². Using data from the Swedish Medical Birth Register in a large epidemiological study of more than 600,000 Nordic women, the authors conclude that use of Swedish snuff, a smokeless tobacco, did not reduce the risk of preeclampsia and gestational hypertension, but that tobacco when smoked did. They infer that combustion products of tobacco such as carbon monoxide protect against preeclampsia, but that constituents of tobacco such as nicotine do not. The data strengthen and extend results of a prior smaller study using the Swedish Medical Birth Register, which had reported a similar association³.

Whereas snuff use did not reduce the risk of mild or severe preeclampsia, preeclampsia which began before or after 37 weeks of gestation, or preeclampsia with or without delivery of a small-for-gestational age (SGA) infant or stillbirth, smoking reduced the risk of all categories of preeclampsia except for preeclampsia with an SGA infant or stillbirth. Of considerable interest, using data from women who changed their tobacco habits at gestational weeks 30-32 from that reported at the first antenatal visit (usually before 15 weeks of gestation), Wikström et al found that women who had reported smoking at the first antenatal visit, but no use of tobacco at 30-32 weeks, did not have reduced risk of preeclampsia or gestational hypertension compared to tobacco non-users, whereas women who reported no use of tobacco at the first antenatal visit, but smoking at 30-32 weeks did. This implies that smoking during the second half of pregnancy is necessary for reducing the risk of preeclampsia or gestational hypertension.

Our understanding of the pathogenesis of preeclampsia has undergone a major revolution in the last decade. It is now believed that the clinical phenotype of preeclampsia may be mediated by a circulating anti-angiogenic state largely due to placental overproduction of soluble fms-like tyrosine kinase 1 (sFlt1 or sVEGFR1), an endogenous vascular endothelial growth factor signaling inhibitor, and soluble endoglin (sEng), a transforming growth factor beta signaling inhibitor ⁴. The etiology of these alterations has been the subject of intense debate, with a number of hypotheses advanced including oxidative stress, altered immunological factors, insulin resistance and enhanced angiotensin II signaling. Can we explain the effects of smoking reported by Wikström et al in terms of our new understanding of the pathogenesis of preeclampsia?

First of all, smoking during pregnancy has been associated with lower circulating concentrations of the anti-angiogenic proteins – sFlt1 and sEng and higher concentrations of the pro-angiogenic protein – placental growth factor ^5-6. Since preeclampsia usually begins after the middle of the second trimester of pregnancy coincident with substantial rises in circulating sFlt1 and sEng, it is not entirely surprising why smoking during the latter half of pregnancy has the greatest benefit. Wikström et al report that the protective effect of smoking appears to be less for the more severe forms of preeclampsia – preterm preeclampsia and preeclampsia with an SGA infant or stillbirth. Moreover, while there is a dose response relationship between smoking and risk for the mild forms of preeclampsia (term preeclampsia, preeclampsia without SGA or stillbirth, or mild preeclampsia), there is not for the severe forms (preterm preeclampsia, preeclampsia with SGA or stillbirth, or severe preeclampsia). It should be noted that preeclampsia with SGA in this study is preeclampsia with severe SGA (<2.5 percentile), and it may be associated with much greater alterations of anti-angiogenic proteins than preeclampsia with SGA defined as <10 percentile.

We believe that every woman has a threshold for angiogenic imbalance, which when crossed will lead to preeclampsia. Since alterations in angiogenic proteins are greater in severe forms of preeclampsia, it should take more of a reduction in angiogenic factor concentrations to bring them below threshold and prevent onset of preeclampsia. Thus a given amount of smoking will afford less protection to women destined to develop preterm preeclampsia or preeclampsia with an SGA infant or stillbirth than to women who will develop mild or term preeclampsia. Cnattingius et al ⁷ and Broughton Pipkin ⁸ have both reported that although smoking reduces the incidence of preeclampsia, it worsens pregnancy outcomes in smokers who develop preeclampsia. We suggest that smoking more frequently prevents preeclampsia from developing in women who would otherwise experience mild disease, thus increasing the proportion who experience severe disease and worse outcomes. Recently Chappell et al have reported an association of smoking with increased risk of superimposed preeclampsia in women with chronic hypertension⁹. This finding is of great interest and remains to be explained.

Among tobacco combustion products, nitric oxide, which promotes vascular relaxation, is a possible mediator of protection against preeclampsia. However, smokers have been noted to have decreased, not increased circulating levels of nitric oxide metabolites ¹; therefore, nitric oxide is unlikely to mediate the protective effect. On the other hand, recent data suggest that carbon monoxide (CO) may be the critical mediator. CO is a vascular protective agent in a number of vascular disorders such as ischemia/reperfusion injuries¹⁰. In addition to exogenous sources such as automobile exhaust or smoking, CO is also produced endogenously during cellular metabolism via the hemoxygenase (HO) system. Cudmore et al reported that carbon monoxide and carbon monoxide releasing molecules lower sFlt1 and sEng production in endothelial cells and placental organ cultures, thus potentially providing a molecular explanation for the lower circulating levels of sFlt1 and sEng noted in smokers during pregnancy¹¹. Cigarette smoking may also directly upregulate HO expression in placental trophoblasts, which in turn may stimulate endogenous CO production. It is also likely that carbon monoxide may inhibit placental apoptosis and necrosis directly, thus reducing the release of placentally derived toxic factors¹². Other evidence to support the hypothesis that the HO pathway may play a important role in the pathogenesis of preeclampsia comes from human studies which have demonstrated decreased HO gene expression in placental villous tissue during first trimester among pregnant women destined to develop preeclampsia ¹³ and decreased exhaled CO among women with preeclampsia¹⁴. Whether decreased HO and CO are found only in a subset of women with preeclampsia or in all such women is not known; and whether CO contributes to other adverse outcomes of pregnancy noted among smokers such as fetal growth restriction should be determined.

In summary, Wikström et al suggest that smoking during pregnancy, but not the use of smokeless tobacco, is associated with lower preeclampsia risk. The protective role of smoking is most likely to be explained through the biological effects of CO that is formed during smoking. CO may act by inhibiting placental production of anti-angiogenic proteins such as sFlt1 and by inhibiting placental apoptosis and necrosis. Further research is needed to understand more fully the molecular mechanisms of the CO-mediated vascular protective effect and to harness this knowledge to identify new therapeutic targets for preeclampsia.