Selenium status and adverse fetal and perinatal outcomes

Barrington et al. ⁽Reference Barrington, Lindsay and James³²⁾ observed significantly lower serum Se levels among women who had had a miscarriage in the first trimester of pregnancy than in either pregnant women or non-pregnant healthy volunteers. The authors referred that haemodilution cannot explain the difference in Se levels between these groups, considering that serum albumin and total protein concentrations of the pregnant and miscarried groups were not different. They suggested that low Se status in women with miscarriage could be related to membrane and cell DNA damage caused by loss of the antioxidant activity performed by Se.

Similar results were found by Al-Kunani et al. ⁽Reference Al-Kunani, Knight and Haswell³³⁾, who described significantly lower hair Se concentrations in women with recurrent miscarriages when compared with non-pregnant women who had had successful pregnancies. In disagreement is the study carried out by Zachara et al. ⁽Reference Zachara, Dobrzynski and Trafikowska³⁴⁾, who observed no significant differences in whole-blood and plasma Se levels when comparing women with miscarriage against women with normal pregnancies. Nevertheless, these authors found that women with miscarriage had significantly lower erythrocyte and plasma GPx activities than women with normal pregnancies and non-pregnant women.

Reduced Se concentrations also seem to be related to neural tube defects, as reported in some studies. Güvenç et al. ⁽Reference Güvenç, Karatas and Güvenç³⁵⁾ observed low Se status in both mothers and their infants with neural tube defects and suggested that the occurrence of such defects could be related not only to a deficiency of the mineral, but also to inadequate intakes.

Although mechanisms implicated in the prevention of neural tube defects are still unknown, Martin et al. ⁽Reference Martin, Gibert and Pintos³⁶⁾ suggested that in women with low Se status, low levels of folate could worsen oxidative stress, and thus purine and pyrimidine synthesis could be jeopardised.

Cengiz et al. ⁽Reference Cengiz, Soylemez and Ozturk³⁷⁾ and Zeyrek et al. ⁽Reference Zeyrek, Soran and Cakmak³⁸⁾ assessed multiple micronutrients in mothers and their newborns with neural tube defects and found conflicting results regarding Se status. The first study found statistically significant lower maternal Se levels among cases. The second study, on the other hand, found that Se concentrations did not differ between cases and controls. It has also been reported in the literature that low Se intake could be associated with the occurrence of diaphragmatic hernia. Although this finding is supported by a large population-based case–control study⁽Reference Yang, Shaw and Carmichael³⁹⁾, the authors highlight limitations which include the use of a FFQ for the assessment of nutrient adequacy and the fact that maternal levels of nutrients were not measured. In addition, associations might have been exclusive to a particular nutrient or a consequence of highly correlated nutrients, and because of the elevated number of nutrients investigated, some might have shown statistically significant differences by chance. It is also possible that lack of information from non-participants might have biased the results.

Low Se status could also be related to adverse perinatal outcomes. It has been suggested that newborn Se concentrations may be related to the Se status of their mothers⁽Reference Dobrzynski, Trafikowska and Trafikowska^30, Reference Micetic-Turk, Rossipal and Krachler^40, Reference Iranpour, Zandian and Mohammadizadeh⁴¹⁾, and that pre-term infants born to healthy women have lower Se concentrations than do term newborns⁽Reference Iranpour, Zandian and Mohammadizadeh^41–Reference Nassi, Ponziani and Becatti⁴⁴⁾. Such findings are not surprising since shortened gestations hinder appropriate transport of Se from the mother to the fetus. In fact, Al-Saleh et al. ⁽Reference Al-Saleh, Al-Doush and Ibrahim⁴²⁾ emphasised that pre-term newborns can be at risk of Se deficiency, especially if total parenteral nutrition is needed. Moreover, as a consequence of their lower Se status, pre-term infants might have their antioxidant defences impaired, especially with respect to GPx activity, a Se-dependent enzyme, which may be related to higher incidence of oxygen dependence at 28 d of life⁽Reference Nassi, Ponziani and Becatti^44, Reference Mentro, Smith and Moyer-Mileur⁴⁵⁾. There are conflicting results, however, regarding the influence of gestational age on Se concentrations and erythrocyte GPx activity⁽Reference Dobrzynski, Trafikowska and Trafikowska^30, Reference Makhoul, Sammour and Diamond^46, Reference Bogden, Kemp and Chen⁴⁷⁾.

Dobrzynski et al. ⁽Reference Dobrzynski, Trafikowska and Trafikowska³⁰⁾ have considered a possible relationship between low Se status and premature birth, since pre-term parturient women in their study had lower plasma and blood Se concentrations than did term parturient women and healthy non-pregnant women. In their study, no significant correlations between gestational age and Se status measures were found. The authors also reported that pre-term parturient women had significantly lower antioxidant status when compared with their controls, as indicated by their erythrocyte GPx activity. Nonetheless, this relationship is not supported by other studies⁽Reference Iranpour, Zandian and Mohammadizadeh^41, Reference Bogden, Kemp and Chen⁴⁷⁾.

With regard to the relationship between Se status and birth weight, there is still much disagreement in the literature. Although many studies have pointed to a possible association⁽Reference Kantola, Purkunen and Kroger^11, Reference Iranpour, Zandian and Mohammadizadeh^41, Reference Al-Saleh, Al-Doush and Ibrahim^42, Reference Bogden, Kemp and Chen^47, Reference Kaplec, Cavar and Kasac⁴⁸⁾, significant results varied according to the groups included or excluded from the analyses or the type of statistical tests applied. For instance, Iranpour et al. ⁽Reference Iranpour, Zandian and Mohammadizadeh⁴¹⁾ observed only significant correlations between cord blood Se levels and birth weight when pre-term and term infants were considered together in the analysis; Al-Saleh et al. ⁽Reference Al-Saleh, Al-Doush and Ibrahim⁴²⁾ found only a significant correlation between newborn Se concentrations and birth weight when low-birth-weight infants were included in the analysis; Kaplec et al. ⁽Reference Kaplec, Cavar and Kasac⁴⁸⁾ found a significant correlation between placental Se concentrations and birth weight among newborns whose birth weight was appropriate for gestational age, but not among intra-uterine growth-restricted newborns; Bogden et al. ⁽Reference Bogden, Kemp and Chen⁴⁷⁾ found no significant correlation between birth weight and serum Se as continuous variables, but a significant association was observed when the lowest decile of Se level was compared with the nine highest. In addition, many authors have failed to establish any kind of relationship between Se status and birth weight⁽Reference Dobrzynski, Trafikowska and Trafikowska^30, Reference Makhoul, Sammour and Diamond^46, Reference Odland, Nieboer and Romanova⁴⁹⁾. Bogden et al. ⁽Reference Bogden, Kemp and Chen⁴⁷⁾ suggested that conflicting findings could be partly attributed to different timing in blood collection (i.e. early in pregnancy or at delivery), and that the effect of prematurity on birth weight could mask the association between Se levels and birth weight.

Concerning intra-uterine growth restriction, Zadrozna et al. ⁽Reference Zadrozna, Gawlik and Nowak⁵⁰⁾ found increased Se concentrations in complicated placentas, suggesting a physiological response to maintain GPx activity due to augmented oxidative stress. On the other hand, Kaplec et al. ⁽Reference Kaplec, Cavar and Kasac⁴⁸⁾ observed that Se levels were not predictors of intra-uterine growth restriction. Their findings are consistent with those of Llanos & Ronco⁽Reference Llanos and Ronco⁵¹⁾, who did not report statistically significant differences between normal-weight neonates and neonates with fetal growth restriction.

The occurrence of adverse outcomes among infants born to HIV-infected mothers might be influenced by maternal Se status: low status has been associated with increased risks of fetal and child deaths, as well as intra-partum and breast-feeding transmission of HIV and lower risk of small-for-gestational age infants. Kupka et al. ⁽Reference Kupka, Garland and Msamanga⁵²⁾ have cited the role of Se in antioxidant defences as one of the factors that might be implicated in fetal deaths. Later, the same group conducted a randomised, double-blind, placebo-controlled trial of Se supplementation to HIV-infected pregnant women and found that the intervention reduced the risk of infant death after 6 weeks, although no significant effects were found regarding the risks of low birth weight, pre-term birth and small-for-gestational age⁽Reference Kupka, Mugusi and Aboud⁵³⁾.

The main characteristics of studies assessing the relationship between Se status and the occurrence of adverse fetal outcomes are shown in Table 1, whereas Table 2 summarises the characteristics of studies which investigated the relationship between Se status and adverse perinatal outcomes.

Table 1 Main characteristics and results of studies assessing the relationship between selenium status and adverse fetal outcomes

CS, cross-sectional; CC, case–control; GPx, glutathione peroxidase; NTD, neural tube defect.

Table 2 Main characteristics and results of studies assessing the relationship between selenium status and adverse perinatal outcomes

CS, cross-sectional; GPx, glutathione peroxidase; TPW, term parturient women; PPW, pre-term parturient women; NPW, non-pregnant women; TB, term babies; PB, pre-term babies; L*, longitudinal (including cohort and follow-up studies); ELBW, extremely low birth weight; NCC, nested case–control; IUGR, intra-uterine growth restriction.

Selenium status and adverse maternal outcomes

There is evidence that maternal plasma Se concentrations are reduced in the presence of pre-eclampsia⁽Reference Atamer, Koçyigit and Yokus^54, Reference Mistry, Wilson and Ramsay⁵⁵⁾. Lower Se levels have been reported in the amniotic fluid of pre-eclamptic pregnant women compared with their normotensive controls⁽Reference Dawson, Evans and Nosovitch⁵⁶⁾ and infants born to pre-eclamptic women have shown lower umbilical cord blood Se levels than infants born to normotensive pregnant women⁽Reference Mistry, Wilson and Ramsay⁵⁵⁾. Decreased GPx activity has also been observed in pre-eclamptic pregnancies⁽Reference Mistry, Wilson and Ramsay⁵⁵⁾. Rayman et al. ⁽Reference Rayman, Bode and Redman⁵⁷⁾ suggested that Se deficiency before the onset of pre-eclampsia might be related to the incidence of the disease, although they speculated that the same association might not be found in populations with adequate Se status. Despite the body of evidence pointing to an association between Se status and pre-eclampsia, some studies have failed to find the same results⁽Reference Roy, Ratnam and Karunanithy^58–Reference Mahomed, Williams and Woelk⁶⁰⁾.

Not only can serum Se levels be reduced in the presence of abnormal glucose metabolism and gestational diabetes⁽Reference Tan, Sheng and Qian^6, Reference Bo, Lezo and Menato^61, Reference Kilinc, Guven and Ezer⁶²⁾, but inadequate dietary Se intake has also been related to these conditions. Bo et al. ⁽Reference Bo, Lezo and Menato⁶¹⁾ reported significantly lower Se intake in hyperglycaemic pregnant women, even after adjustment for confounders such as age, gestational age, familial diabetes, pre-pregnancy BMI and total energy intake. However, associations between maternal Se levels and outcomes such as the percentages of caesarean sections, small- or large-for-gestational age newborns, pre-term births and neonatal morbidities have not been found. Taking these results into account, Kilinc et al. ⁽Reference Kilinc, Guven and Ezer⁶²⁾ suggested that Se status is important to glucose-intolerant individuals, and that its deficiency could aggravate diabetes and be related to complications in the gestational period. In addition, Bo et al. ⁽Reference Bo, Lezo and Menato⁶¹⁾ proposed that dietary counselling and antioxidant supplementation might be useful if further research confirms the relationship between micronutrient status and hyperglycaemia.

Finally, low Se status of HIV-infected pregnant women has been related to increased risk of mortality, whereas high Se status has been associated with higher CD4 cell count in the first years of follow-up in the cohort study by Kupka et al. ⁽Reference Kupka, Msamanga and Spiegelman⁶³⁾. However, in the Se supplementation study conducted by the same group and previously cited in the present paper, the intervention had no statistically significant effects on CD4, CD8 and CD3 cell counts, viral load or maternal mortality⁽Reference Kupka, Mugusi and Aboud⁵³⁾.

The main characteristics of studies assessing the relationship between low Se status and adverse maternal outcomes are presented in Table 3.

Table 3 Main characteristics and results of studies assessing the relationship between selenium status and adverse maternal outcomes

CS, cross-sectional; CC, case–control; GD, gestational diabetes; GI, glucose intolerance; GPx, glutathione peroxidase; HPW, healthy pregnant women; L*, longitudinal (including cohort and follow-up studies); NPW, non-pregnant women.

Limitations of available evidence

It must be emphasised that the role of Se status in the aetiology of pregnancy complications is yet to be established. Although a connection between low Se status and the occurrence of adverse fetal outcomes has been suggested, because most available studies are cross-sectional or case–control, their results are not appropriate to the assessment of causal relationships. Thus, low Se status could either be implicated in the causal process of such adverse outcomes or simply indicate a maternal physiological response to increased oxidative stress states. Ideally, high-quality prospective studies assessing the Se status of women through the periconceptional period to pregnancy loss or detection of any adverse outcome should be carried out, but such type of design is not always feasible.

Additionally, some studies have limitations that should be noted and addressed in future research, which comprise a comprehensive description of study populations (including their Se status), selection of subjects, research setting and analytical methods poorly described. The lack of such type of information limits generalisation of study results. Also, most studies fail to present the procedures for defining sample sizes, thus making it difficult to judge their statistical power, especially because many studies have small sample sizes.

Among important confounding variables, which should be taken into account but are not frequently assessed, are maternal Se intake, parity and smoking habits. Maternal nutrient intake assessment as a whole is essential since it can also be hypothesised that Se-deficient women lack other micronutrients, and thus Se deficiency itself might not explain adverse fetal outcomes. Indeed, Bogden et al. ⁽Reference Bogden, Kemp and Chen⁴⁷⁾ suggested that the association between serum Se and birth weight in neonates could simply reflect good maternal nutritional status. Parity could contribute not only to low Se status but also to other micronutrient deficiencies, which could also influence pregnancy outcomes. Smoking should also be included in the analyses since it induces oxidative stress and consequently activates antioxidant defence mechanisms⁽Reference Kantola, Purkunen and Kroger^11, Reference Bogden, Kemp and Chen⁴⁷⁾. In addition, cigarettes might contain Se, which could also interfere in maternal Se status. Bogden et al. ⁽Reference Bogden, Kemp and Chen⁴⁷⁾ highlighted the fact that most studies do not control for this variable, which can hinder possible associations between Se measurements and pregnancy outcomes.

Finally, even though there is evidence suggesting that Se status might be important in relation to the occurrence of gestational morbidities, large representative population studies assessing the relationship between Se intake during pregnancy and maternal and newborn health outcomes are still scarce.

Conclusion

Although Se status might play an important role in adverse pregnancy outcomes, probably because of its participation in the antioxidant defence system as GPx, not enough evidence has been produced so far to form a comprehensive understanding of the role of Se. Further research with bigger sample sizes, prospective designs and well-documented reports are still needed, allowing greater statistical power and more robust analyses. Ideally, maternal Se and antioxidant status should be assessed from periconceptional through postnatal periods, and a large array of maternal and newborn outcomes should be addressed. Moreover, studies evaluating the relationship between dietary and supplementary Se intake before and during pregnancy in relation to adverse outcomes would be of great importance.