Essential trace elements, such as Fe, Cu, I, Zn and Se, are dietary elements that occur in a very small portion in the human body (usually less than 1–10 parts per million) and play significant roles in the development, growth and appropriate function of cells. These elements are involved as cofactors in enzyme catalysis, being essential for a plethora of biological mechanisms and functions, including reproduction(Reference Gajewska, Blazewicz and Laskowska1).
Se is particularly important due to its anti-inflammatory, chemopreventive and antiviral characteristics(Reference Rayman2,Reference McDowell3) . In addition, this element plays an important role in neutralising oxidative stress since it is an integral part of several selenoproteins such as glutathione peroxidase and thioredoxin reductase. The majority of all known selenoproteins are expressed in the thyroid gland(Reference Schomburg4). In fact, it is known that Se composes the enzyme iodothyronine deiodinase, which is important in thyroid hormone production, and it also composes other enzymes such as selenoproteins S, P and W(Reference Vanderlelie and Perkins5).
Se is naturally found in meat, seafood, grains, seeds and nuts(6) and its food content can vary geographically, depending on the natural availability of Se in the soil and the direct addition of this element to the food supply(Reference Reilly7). The RDA of Se for adult men and women is 55 µg (0·7 µmol)/d(8); however, this value can increase in certain conditions, namely in pregnancy, in which Se requirements are 65 µg/d(Reference Oh, Keats and Bhutta9).
In recent years, several studies have shown that low levels of Se may be involved in some adverse pregnancy conditions, such as gestational diabetes, miscarriages and premature birth(Reference Qazi, Angel and Yang10,Reference Askari, Iraj and Salehi-Abargouei11) . Additionally, many authors associate lower levels of Se in placental tissue, blood or toenail with the decreased activity of glutathione peroxidase and, subsequently, with pregnancy-induced hypertension (PIH)(Reference Perkins12,Reference Poston, Igosheva and Mistry13) .
The International Society for the Study of Hypertension in Pregnancy (ISSHP) divides the hypertensive disorders of pregnancy (HDP) into two categories: ‘Hypertension known before pregnancy or present in the first 20 weeks’ and ‘Hypertension arising de novo at or after 20 weeks’(Reference Brown, Magee and Kenny14). Hypertension arising de novo at or after the 20th gestational week – which is the focus of this review – may be classified as: (a) transient gestational hypertension, when hypertension is detected in clinical setting and resolves with repeated blood pressure readings; (b) gestational hypertension (the above referred PIH), if there is not any specific sign of pre-eclampsia (PE) (e.g. proteinuria); (c) De novo PE and (d) PE superimposed on chronic hypertension. In all cases, the ISSHP defines hypertension as a systolic blood pressure ≥ 140 mmHg and/or diastolic blood pressure ≥ 90 mmHg, based on an average of at least two measurements.
There are some clinical papers demonstrating potential benefits of Se supplementation during pregnancy for HDP’s prevention; nevertheless, current evidence does not support yet the use of Se in pregnancy and women of childbearing age due to insufficient data and conflicting results(Reference Hubalewska-Dydejczyk, Duntas and Gilis-Januszewska15). In fact, there is no current comprehensive review that systematises the status of knowledge about this topic. So, this work aimed to systematically review the most recent scientific evidence to understand the relationship between Se levels and HDP. In addition, this review aims to provide a reflection on the potential effectiveness of using Se as a supplement during pregnancy to prevent HDP or as an adjuvant treatment for these conditions.
Methods
Design
This systematic review follows the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines(Reference Page, McKenzie and Bossuyt16).
The protocol of this study was registered in the International Prospective Register of Systematic Reviews (identifier: CRD42022310424), and it is available online in (https://www.crd.york.ac./prospero/display_record.php?IDuk=CRD42022310424).
Data sources, search strategy and eligibility criteria
For this systematic review, PubMed, Scopus and Web of Science databases were searched for relevant articles published in the last 10 years. Table 1 shows the terms used for database search (the same terms were used for the three databases). We chose to restrict our search on publication date since the last specific systematic review we found about this topic was published in 2016(Reference Xu, Guo and Gu17). Additional data sources were used such as reference lists of retrieved publications, grey literature (Portuguese journals and public master’s thesis and dissertations from the University of Porto) and the clinical trials databases clinicaltrials.gov and European Union Clinical Trials Register. The literature search was performed in December 2021, and the last update was conducted on February 2022.
Table 1. Search terms used in the database search
The primary outcome of this revision was the occurrence of HDP. Additionally, the use of Se supplements was considered as a secondary outcome. Studies were included whenever they referred to: (1) the population of interest, that is, pregnant women; (2) the exposure of interest, that is, measurement of Se levels in any biological sample regardless of Se supplementation at any dose, regimen and type (e.g. Se-only supplements or Se as part of multivitamin/mineral tablets); (3) the outcome of interest, that is, the prevalence or development of HDP and (4) any of the following study types: observational, nonrandomised experimental studies and randomised controlled trials (RCT).
We excluded: animal and in vitro studies; studies in languages other than English, Portuguese, Italian, Spanish or French; unpublished or non-peer-reviewed articles; case reports; narrative reviews/comment articles; systematic reviews and meta-analyses.
Study selection and data extraction
The reference management software EndNote X20 was used to remove duplicate studies and to organise the resulting aggregated database. Titles and abstracts were independently screened by two reviewers (EK and IS). Studies in which the abstract did not mention the outcome of interest were automatically excluded and disagreements between reviewers, which was very low, were resolved through discussion.
The full texts of the potentially eligible studies were assessed for inclusion by two review team members (EK and IS) and discrepancies were resolved by consensus between the whole group. Afterwards, thirty articles were considered to include in the current revision. A flow chart of the study selection process can be found in Fig. 1. After the inclusion phase, relevant information about studies was extracted by three reviewers with an overlap of five studies, for assessment of study quality and evidence synthesis. Data extraction included: reference details (first author name, publication year of study); study details (study type, country); participant details (study population, inclusion and exclusion criteria of participants, number of participants); exposure details (biological sample, pregnancy timepoint and assay method for Se quantification) and main results of the study. Discrepancies were resolved by consensus between the three authors. The corresponding author of one of the included articles was contacted by email to clarify about the study type.
Fig. 1. Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow chart of study selection.
Quality assessment of included studies
Quality assessment of included studies was performed using the National Heart, Lung, and Blood Institute (NHLBI) Study Quality Assessment Tools. Three different tools were used depending on the study type. All authors independently assessed the quality with an overlap of ten articles for calibration of criteria application. Each item of questionnaires was evaluated as ‘Yes’, ‘No’, ‘Not Applicable’, ‘Cannot Determine’ or ‘Not Reported’, according to NHLBI’s instructions. In the end, an overall rating of ‘Good’, ‘Fair’ or ‘Poor’ quality was used based on the total score obtained for each article. Any disagreement between the authors was resolved through discussion. Results of quality assessment of all the studies are presented in online Supplementary Tables 1, 2 and 3.
Results
A total of 263 articles were identified from the three databases mentioned previously, and one dissertation was identified from grey literature. After combining the various databases, 105 duplicates were removed and 158 records were screened afterwards. Among these, 120 articles were excluded based on title and abstract considering the inclusion and exclusion criteria. The remaining thirty-eight articles were sought for full-text retrieval. Four full texts were unavailable and, so, thirty-four articles were screened for eligibility based on full text. Of these, four articles were excluded: two of them did not describe Se levels; one study did not address any of the outcomes of interest and one other referred to assessment of Se levels in pregnant women with gestational diabetes. The dissertation (from grey literature) was also excluded because it was a narrative review. As a result, thirty articles were included in the study. A flow chart of the article search, selection and screening can be found in Fig. 1.
General characteristics of included studies
Table 2 summarises general information about studies included in this review. Almost all the thirty included studies were observational (n 29), and the majority of them were case–control studies (77 %, n 23). Among the case–control studies, five were nested case–control studies(Reference Bommarito, Kim and Meeker18,Reference Ghaemi, Forouhari and Dabbaghmanesh19,Reference Laine, Ray and Bodnar20,Reference McKeating, Fisher and MacDonald21,Reference Mistry, Gill and Kurlak22) and one was a prospective case–control study(Reference Katz, Paz-Tal and Lazer23). A minority of the studies were prospective cohorts (n 3)(Reference Choi, Sun and Yoo24,Reference Holmquist, Brantsæter and Meltzer25,Reference Lewandowska, Sajdak and Lubiński26) , retrospective cohort (n 1)(Reference Liu, Zhang and Guallar27) or cross-sectional (n 2)(Reference Ambad, Jha and Bankar28,Reference McAlpine, McKeating and Vincze29) . Among the included studies, only one was an RCT(Reference Rayman, Bath and Westaway30).
Table 2. General information about studies included in this systematic review (numbers and percentages)
RCT, randomised controlled trial.
Regarding geographical distribution, the included studies were performed in a total of seventeen countries from all over the world. Most studies were carried out in Asian population (33 %, n 10)(Reference Ghaemi, Forouhari and Dabbaghmanesh19,Reference Katz, Paz-Tal and Lazer23,Reference Choi, Sun and Yoo24,Reference Ambad, Jha and Bankar28,Reference Al-Hilli, Al-Shalah and Hasan31,Reference Farzin and Sajadi32,Reference Haque, Moghal and Sarwar33,Reference Kim, Kim and Lee34,Reference Mazloomi, Khodadadi and Alimohammadi35,Reference Negi, Pande and Karki36) , and the remaining articles are evenly distributed across Europe(Reference Holmquist, Brantsæter and Meltzer25,Reference Lewandowska, Sajdak and Lubiński26,Reference Rayman, Bath and Westaway30,Reference Dahabiyeh, Tooth and Kurlak37,Reference Lewandowska, Więckowska and Sajdak38) , Africa(Reference Elongi Moyene, Scheers and Tandu-Umba39,Reference Enebe, Dim and Ugwu40,Reference Eze, Ododo and Ugwu41,Reference Maduray, Moodley and Soobramoney42,Reference Nnodim, Emmanuel and Hope43,Reference Soobramoney, Maduray and Moodley44) and America(Reference Bommarito, Kim and Meeker18,Reference Laine, Ray and Bodnar20,Reference Liu, Zhang and Guallar27,Reference da Silva, Martins-Costa and Valério45,Reference Rezende, Barbosa and Palei46) . An exception was Australia, which had only two studies(Reference McKeating, Fisher and MacDonald21,Reference McAlpine, McKeating and Vincze29) , one of which a multicentre study(Reference Mistry, Gill and Kurlak22) and one study was conducted in an European-Asian country(Reference Cinemre, Cinemre and Erdogan47). Overall, the most represented countries were Iran, Nigeria and the USA, with three studies each.
In terms of period of exposure assessment, almost half of the studies (44 %, n 14) evaluated Se levels in a biological sample collected in the third trimester. Also, three other studies evaluated Se levels in the third trimester in addition to another previous timepoint evaluation(Reference Ghaemi, Forouhari and Dabbaghmanesh19,Reference Choi, Sun and Yoo24,Reference Rayman, Bath and Westaway30) . Several studies (16 %, n 5) evaluated Se levels at delivery(Reference Katz, Paz-Tal and Lazer23,Reference Kim, Kim and Lee34,Reference Negi, Pande and Karki36,Reference Dahabiyeh, Tooth and Kurlak37,Reference Maduray, Moodley and Soobramoney42) and one study used samples collected within 24–72 h after delivery(Reference Liu, Zhang and Guallar27). Importantly, three studies evaluated Se levels in the first trimester, that is, before the diagnosis of HDP (after 20 weeks) took place(Reference Choi, Sun and Yoo24,Reference Lewandowska, Sajdak and Lubiński26,Reference Lewandowska, Więckowska and Sajdak38) . In three studies, it was not possible to obtain rigorous information about the period of exposure assessment(Reference Ambad, Jha and Bankar28,Reference Enebe, Dim and Ugwu40,Reference Cinemre, Cinemre and Erdogan47) .
Diverse biological samples were collected for analysis and some studies assessed more than one type of biological sample(Reference Katz, Paz-Tal and Lazer23,Reference Rayman, Bath and Westaway30,Reference Maduray, Moodley and Soobramoney42) . As it can be observed in Table 2, the great majority of studies (75 %, n 24) analysed maternal blood samples (serum, plasma or whole blood), followed by maternal urine (n 2)(Reference Bommarito, Kim and Meeker18,Reference Elongi Moyene, Scheers and Tandu-Umba39) , nail (n 2)(Reference Rayman, Bath and Westaway30,Reference Soobramoney, Maduray and Moodley44) and hair samples (n 1)(Reference Maduray, Moodley and Soobramoney42). Only one study used placenta(Reference Laine, Ray and Bodnar20) and two studies used cord blood samples(Reference Katz, Paz-Tal and Lazer23,Reference Negi, Pande and Karki36) .
Se was measured mainly using inductively coupled plasma – mass spectrometry (ICP-MS) (55 % of studies, n 17); however, some studies used different quantification methods, such as atomic absorption spectrometry (n 10) and instrumental neutron activation analysis (n 2). Two studies did not report the method used for Se quantification(Reference Ambad, Jha and Bankar28,Reference Negi, Pande and Karki36) .
Although most studies have evaluated Se levels only in non-supplemented women, in three studies Se levels were measured in supplemented and non-supplemented pregnant women(Reference Ambad, Jha and Bankar28,Reference McAlpine, McKeating and Vincze29,Reference Rayman, Bath and Westaway30) . Furthermore, two studies evaluated Se intake through food frequency questionnaires (FFQ)(Reference Holmquist, Brantsæter and Meltzer25,Reference McAlpine, McKeating and Vincze29) .
Regarding quality assessment, 50 % (n 15) of studies were classified as ‘Good’; 43 % (n 13) were classified as ‘Fair’ and 7 % (n 2) were classified as ‘Poor’. ‘Poor’ classified studies are a case–control study(Reference Maduray, Moodley and Soobramoney42) and a cross-sectional study(Reference Ambad, Jha and Bankar28).
Systematisations of the risk of bias for case–control studies and for observational cohort and cross-sectional studies included in this review are illustrated in Fig. 2 and 3, respectively.
Fig. 2. Risk of bias graph for case–control studies of this review.
Fig. 3. Risk of bias graph for observational cohort and cross-sectional studies of this review.
Data from included studies
Results of data extraction from studies are systematised in Table 3 (case–control studies) and Table 4 (non-case–control studies). These tables also include the overall rating of quality for each study. Overall, the smallest population size was 44 women (34 PE cases and controls and 10 non-pregnant women)(Reference Dahabiyeh, Tooth and Kurlak37) and the largest sample size was 69 972 women (69 972 women for analysis of Se intake and 2572 of them were also assessed for Se status)(Reference Holmquist, Brantsæter and Meltzer25). Twenty-three studies reported PE as an outcome, six studies reported both PE and gestational hypertension as study outcomes(Reference Holmquist, Brantsæter and Meltzer25,Reference Lewandowska, Sajdak and Lubiński26,Reference Rayman, Bath and Westaway30,Reference Lewandowska, Więckowska and Sajdak38,Reference da Silva, Martins-Costa and Valério45,Reference Rezende, Barbosa and Palei46) and one study reported PE and eclampsia(Reference Negi, Pande and Karki36).
Table 3. Characteristics of case–control studies included in this review
AAS, atomic absorption spectrometry; E, eclampsia; GH, gestational hypertension; ICP-MS, inductively coupled plasma – mass spectrometry; IQR, interquartile range; INAA, instrumental neutron activation analysis; PE, pre-eclampsia; PIH, pregnancy-induced hypertension; NT, normotensive.
Table 4. Characteristics of non-case–control studies included in this review
AAS, atomic absorption spectrometry; E, eclampsia; FFQ, food frequency questionnaire; GH, gestational hypertension; ICP-MS, inductively coupled plasma – mass spectrometry; INAA, instrumental neutron activation analysis; IQR, interquartile range; PE, pre-eclampsia; PIH, pregnancy-induced hypertension; NT, normotensive.
Most studies excluded women with risk factors for HDP, such as twin pregnancy, overweight and obesity, history of hypertension and diabetes. Online Supplementary Table 4 describes in detail the inclusion and/or exclusion criteria of participants of each study.
Nineteen studies found a negative association between Se levels and the prevalence or development of HDP, and eleven studies did not find any association. None of the studies reported a positive association.
Negative association between Se and hypertensive disorders of pregnancy
All the nineteen studies that found a negative association between Se levels and the prevalence or development of HDP referred to PE; however, three of them also referred to gestational hypertension/PIH(Reference Lewandowska, Sajdak and Lubiński26,Reference Rayman, Bath and Westaway30,Reference Lewandowska, Więckowska and Sajdak38) and one to eclampsia(Reference Negi, Pande and Karki36). Of the nineteen studies, sixteen refer to Se levels quantified in serum/plasma/blood, and regarding quality, eight were classified as ‘Good’(Reference Bommarito, Kim and Meeker18,Reference Ghaemi, Forouhari and Dabbaghmanesh19,Reference McKeating, Fisher and MacDonald21,Reference Lewandowska, Sajdak and Lubiński26,Reference Rayman, Bath and Westaway30,Reference Lewandowska, Więckowska and Sajdak38,Reference Enebe, Dim and Ugwu40,Reference Eze, Ododo and Ugwu41) , nine were classified as ‘Fair’(Reference Katz, Paz-Tal and Lazer23,Reference Al-Hilli, Al-Shalah and Hasan31,Reference Farzin and Sajadi32,Reference Haque, Moghal and Sarwar33,Reference Mazloomi, Khodadadi and Alimohammadi35,Reference Negi, Pande and Karki36,Reference Dahabiyeh, Tooth and Kurlak37,Reference Nnodim, Emmanuel and Hope43,Reference Soobramoney, Maduray and Moodley44) and two studies were classified as ‘Poor’(Reference Ambad, Jha and Bankar28,Reference Maduray, Moodley and Soobramoney42) (Tables 3 and 4).
In addition to data showing negative associations between Se and HDP, which details can be found in Tables 3 and 4, some other important conclusions were found. For example, Eze et al.(Reference Eze, Ododo and Ugwu41) observed that maternal serum Se deficiency worsened with the increasing severity of PE (P < 0·001).
In line with this, Haque et al.(Reference Haque, Moghal and Sarwar33) reported that maternal serum Se concentrations for mild and severe PE were significantly different (24·63 (sd 0·65) µg/l and 21·71 (sd 1·35) µg/l, respectively) (P < 0·05); Negi et al.(Reference Negi, Pande and Karki36) reported that cord blood Se concentrations were lower in the hypertension cases (P < 0·005), and eclampsia cases tended to present lower values of Se in comparison with the PE cases and Soobramoney et al.(Reference Soobramoney, Maduray and Moodley44) observed that, despite maternal nail Se levels were similar between PE and normotensive (NT) women (P > 0·05), lower Se levels were significantly related to the severity of PE (P < 0·05).
Importantly, Lewandowska et al. (Reference Lewandowska, Sajdak and Lubiński26) found that lower Se levels in serum were found in pregnant women who subsequently developed gestational hypertension or PE, compared with matched controls who remained normotensive (P < 0·05). Concurring with this, Lewandowska et al. (Reference Lewandowska, Więckowska and Sajdak38) also reported that an increase in maternal serum Se concentrations of 1 µg/l reduced the risk of gestational hypertension/PE and isolated gestational hypertension by 5 % (P = 0·011) and 6 % (P = 0·004), respectively. In the same line of evidence, Bommarito et al. (Reference Bommarito, Kim and Meeker18) reported that an interquartile range increase in urinary Se was associated with a reduction in the risk of PE (hazard ratio [HR]: 0·28, 95 % CI (0·08, 0·94)). Lastly, Rayman et al.(Reference Rayman, Bath and Westaway30), the only RCT included in this review, used samples originated from the SPRINT (Selenium in PRegnancy INTervention) study(Reference Rayman, Searle and Kelly48), in which primiparous women from UK were randomised to treatment with Se (60 mg/d) or placebo from 12 to 14 weeks of gestation until delivery. After exclusion of non-compliers with Se treatment, Se supplementation was found to significantly reduce the OR for gestational hypertension/PE (OR 0·30, 95 % CI (0·09, 1·00), P = 0·049).
Lack of association between Se levels and hypertensive disorders of pregnancy
Eleven studies did not find any association between Se levels and HDP. Of these studies, nine refer to Se levels quantified in serum/plasma/blood and, regarding quality, seven were classified as ‘Good’(Reference Mistry, Gill and Kurlak22,Reference Choi, Sun and Yoo24,Reference Holmquist, Brantsæter and Meltzer25,Reference Liu, Zhang and Guallar27,Reference Kim, Kim and Lee34,Reference Elongi Moyene, Scheers and Tandu-Umba39,Reference da Silva, Martins-Costa and Valério45) and four were classified as having a ‘Fair’ quality(Reference Laine, Ray and Bodnar20,Reference McAlpine, McKeating and Vincze29,Reference Rezende, Barbosa and Palei46,Reference Cinemre, Cinemre and Erdogan47) . All the eight studies referred to PE; however, four of them also referred to gestational hypertension/PIH(Reference Holmquist, Brantsæter and Meltzer25,Reference McAlpine, McKeating and Vincze29,Reference da Silva, Martins-Costa and Valério45,Reference Rezende, Barbosa and Palei46) .
Details of the studies reporting a lack of association between Se and HDP can be found in Tables 3 and 4. Importantly, Da Silva et al.(Reference da Silva, Martins-Costa and Valério45) observed that maternal serum Se levels did not differ significantly between PE, normotensive or hypertensive (chronic or gestational) pregnant women. Similarly, Rezende et al.(Reference Rezende, Barbosa and Palei46) compared plasma levels of Se among non-pregnant, healthy pregnant, gestational hypertensive and PE women. Se levels were equivalent among pregnant groups (all P > 0·05). However, non-pregnant women had considerably higher Se levels compared with other groups. On the other hand, despite Elongi et al.(Reference Elongi Moyene, Scheers and Tandu-Umba39) reported significantly higher urinary Se concentrations in PE women when compared with normotensive women (P < 0·001), the differences were less marked and statistical significance was lost after adjustment for urine dilution.
Regarding supplemental Se and Se intake, Holmquist et al.(Reference Holmquist, Brantsæter and Meltzer25) did not find significant associations between dietary or supplemental Se intake and gestational hypertension, mild-PE or severe PE. Also, McAlpine et al.(Reference McAlpine, McKeating and Vincze29) aimed to investigate the effects of supplements on micronutrient status and birth outcomes. Dietary data were self-reported by participants using a FFQ. Supplement use had no significant influence on the incidence of hypertensive disorders (P > 0·05). Despite this, no detailed data about the type of supplement, duration of supplementation and dose taken by supplemented women were described by the authors.
Discussion
This systematic review aimed to evaluate and sum up the available evidence of the last decade about the association between Se levels and HDP. Among the thirty included studies, a majority of them, 61 % (n 19) of the ‘good’ or ‘fair’ studies, reported a negative association between Se and HDP, that is, lower Se levels were associated with the occurrence of HDP or with an increased risk of developing HDP. On the order hand, some studies, 39 % (n 11) of the ‘good’ or ‘fair’ studies, reported a lack of association.
Women with HDP, present a dysfunctional maternal endothelium, with increased production of reactive oxygen species, and an excessive systemic inflammation(Reference Mistry, Wilson and Ramsay49). This can explain an increment in Se demand and thus the lower Se levels in women with HDP when compared with normotensive controls, as observed in the majority of studies of this review. In fact, some authors suggest that concentrations of Se decrease proportionally with the severity of previously mentioned phenomena(Reference Hesse-Bähr, Dreher and Köhrle50).
Low Se availability may thus be a consequence of HDP but, on the other hand, it can be seen as a cause of those conditions. Specifically, Se deprivation may impair correct function of antioxidant Se-dependent enzymes – glutathione peroxidases and thioredoxin reductases – compromising antioxidant defence pathways, which could lead to endothelial dysfunction observed in the pathophysiology of HDP(Reference Mistry, Wilson and Ramsay49,Reference Roman, Jitaru and Barbante51) . Additionally, Se deficiency could contribute to HDP by disturbing thyroid function and increasing TSH production(Reference Ventura, Melo and Carrilho52) which has been suggested as a risk factor for gestational hypertension(Reference Lai, Zhan and Liu53).
Three studies of this review(Reference Choi, Sun and Yoo24,Reference Lewandowska, Sajdak and Lubiński26,Reference Lewandowska, Więckowska and Sajdak38) measured Se levels in the first trimester of pregnancy, that is, before the diagnosis of HDP (after 20 weeks) took place, and two of them(Reference Lewandowska, Sajdak and Lubiński26,Reference Lewandowska, Więckowska and Sajdak38) (‘good’ studies) found a negative association between Se and HDP. Lewandowska et al.(Reference Lewandowska, Sajdak and Lubiński26) additionally found that the risk of developing HDP increased along with decreasing levels of Se in the first trimester. These findings suggest that adequate Se levels from an early stage of pregnancy, or even in the preconceptional period, have an important and protective role against HDP. On the other hand, our revision included six studies that evaluated Se levels in blood/serum or in placenta at delivery or hours after delivery. Of these, the majority revealed a negative association between Se levels and PE, which corroborates that Se deficiency could also be seen as a consequence of HDP.
Like the included study of Rayman et al.(Reference Rayman, Bath and Westaway30), some studies from literature combine the outcomes of PE and gestational hypertension/PIH in their analysis since the distinction between the two is not always completely obvious and consensual(Reference Barton, O’Brien and Bergauer54,Reference Davis, Mackenzie and Brown55) . PE was the most reported HDP in the included studies of this review. To date, few other reviews have been carried out about this theme. For example, a negative association between Se and PE was also reported in the year 2016 by Xu et al. (Reference Xu, Guo and Gu17). The researchers included in their meta-analysis thirteen observational studies and three RCT (in two of them Se supplementation began in the first trimester with a duration of 5–6 months, while the other one began on late pregnancy with a duration of 6–8 weeks). In the meta-analyses, they observed lower Se blood levels in women with PE, when compared with normotensive controls, and also concluded that Se supplementation in pregnant women was associated with a lower risk of developing PE (the relative risk for PE was 0·28). On the other hand, Rumbold et al.(Reference Rumbold, Duley and Crowther56) conducted a systematic review with meta-analysis about antioxidants for preventing PE, but only one study about Se was included – Han et al.(Reference Han and Zhou57) – in which fifty-two women were supplemented with 100 µg/d of Se for 6–8 weeks during late pregnancy, but no differences between the Se-treated and the control groups in preventing PE were found. This study was also the only included study referring to Se in the systematic review conducted by Salles et al. (Reference Salles, Galvao and Silva58).
Our review shows strong evidence in favour of a negative association between Se and HDP. Nevertheless, some of the included studies were not in agreement with this. The disparity among some available evidence could be due to different study types of the included studies, to differences in sample size or in characteristics of the study population, as well as due to the heterogeneity in biological samples used, gestational age of sample collection and Se quantification method. In fact, the proportion of case–control studies was higher in the group reporting a negative association (84 % case–control) when compared with the proportion of case–control studies in the group that did not find associations (64 % case–control). Also, the mean sample size of studies finding a negative association was smaller (mean n 179) when compared with the mean sample size of studies that did not find significant associations (mean n 551). Regarding timepoint of quantification, the proportion of studies quantifying Se during pregnancy (before delivery) was higher in the group reporting a negative association (74 %, n 14) when compared with the group that did not find associations (64 %, n 7). Finally, the proportion of studies using inductively coupled plasma – MS to measure Se was lower in the group reporting a negative association (36 %, n 7) when compared with the group that did not find associations (81 %, n 9).
Despite this heterogeneity among studies, the proportion of studies using blood (plasma/serum/whole blood) samples was similar in the group that found negative associations (84 %, n 16) when compared with the group that did not report associations (81 %, n 9).
Still regarding biological samples used, most of the included studies used maternal serum, plasma, or whole blood for Se quantification. Plasma/serum and erythrocyte are the samples most commonly used for measurement of Se levels, and they reflect short-term and long-term Se status, respectively(Reference Thomson and Caballero59). On the other hand, blood Se concentration is generally considered a useful measure of both Se status. Some authors consider erythrocytes as the more precise and stable biomarker for quantification of Se concentration(Reference Salles, Galvao and Silva58,Reference Chandra, Tripathi and Mishra60,Reference Chen, Myers and Wei61,Reference Faupel-Badger, Hsieh and Troisi62) ; however, only one study of this review(Reference Liu, Zhang and Guallar27) measured concentrations of Se in erythrocytes. Urine is also commonly used because of the convenience and low discomfort associated with the collection. Two of the included studies used maternal urine, but only in one of them Se concentrations in urine were adjusted for dilution and converted to daily amounts. In fact, daily urinary excretion of Se is closely associated with plasma Se concentrations(Reference Thomson and Caballero59). In general, we considered that a great majority of the included studies used adequate biological samples for Se quantification.
Additionally, this review included three studies of ‘good’ quality suggesting that an increase in Se levels is protective against HDP (of these, only one was an RCT of Se supplementation), but two other studies of ‘good’ or ‘fair’ quality did not find evidence supporting that Se supplementation reduces the risk of HDP. So, the gathered evidence in favour of Se supplementation to prevent HDP is not sufficient.
In fact, despite the benefits of Se as a protector against oxidative damage are well known, Se supplementation is not yet preconised in guidelines for prevention and/or management of HDP, because there is a lack of knowledge about the complex interactions between Se and other different micronutrients, as well as about individual response to different doses of Se(Reference Hubalewska-Dydejczyk, Duntas and Gilis-Januszewska63). Currently, for example, when Ca intake is likely to be low (<600 mg/d) in cases of pregnant women with risk factors for PE, the ISSHP recommends supplementation with Ca (1·2–2·5 g/d) in addition to aspirin. In this context, would Se supplementation interfere in any way with the already preconised Ca supplementation? Considering the relevance of Se for thyroid function, would Se supplementation be compatible with the already preconised iodine supplementation?
Our study has important strengths and some limitations. Strengths of our review include the use of a search methodology in line with the current recommendations for systematic reviews, with a comprehensive and sensitive search strategy, employing multiple databases, searching for grey literature, paired selection, data extraction by multiple reviewers and the use of quality assessment tools. The use of a standardised tool such as the NHLBI Study Quality Assessment Tools allowed a better and systematised characterisation of each article. Also, all the studies retrieved in our search were written in English, and for that reason, no studies were excluded by language. Another strength is that our revision did not exclude studies based on the type of biological sample, on the type of HDP and, importantly, on the timepoint for Se quantification which enabled us to suggest about a putative role of Se deficiency in the pathogenesis but also as a consequence of HDP. Due to huge variations and heterogeneity of the included articles, a quantitative meta-analysis was not performed.
In contrast, this review has potential limitations: the exclusion of articles based on publication date could have excluded some relevant studies. Nevertheless, the last available systematic review about this theme was published in 2016 and it included papers until 2014, which ensures an overlap of 4 years with our search. Another limitation is that our search retrieved only one RCT while the majority of the included studies were observational case–controls, which do not allow to draw conclusions about causality.
Conclusion
In conclusion, this review provides an important amount of quality evidence suggesting that low Se levels associate with the occurrence of HDP, corroborating the most recent meta-analyses about this topic(Reference Xu, Guo and Gu17). In addition, this review emphasises that low Se could be a player in the pathogenesis of HDP, but it could also be a consequence of those diseases. Nevertheless, the gathered information is not enough to underlie a recommendation for Se supplementation in pregnancy to protect against HDP.
In fact, several questions remain unanswered and require urgent attention from the scientific community: (a) what would be the best time to begin Se supplementation during or before pregnancy; (b) what would be the adequate daily dose of Se for an effective protection against HDP; (c) what would be the adequate duration of supplementation; (d) who would be the target population: all pregnant women or only women at high risk for HDP; (e) what would be the health risks for the mother and the child and (f) what micronutrient interactions should be considered.
Despite a recent increase and investment in the scientific research about this theme (studies published in the last 3 years account for 47 % (n 14) of the included studies), this review emphasises the need for further well-designed, large and high-quality, double blind, placebo-controlled randomised trials that may find answers to the abovementioned questions and provide blunt evidence regarding the benefits of Se supplementation during pregnancy to prevent HDP.
Acknowledgements
The authors especially thank Professor Margaret P. Rayman (University of Surrey) for help in clarifying the study type of her article.
The present study was supported by National Funds through FCT – Fundação para a Ciência e a Tecnologia, I.P., within CINTESIS, R&D Unit (reference UIDB/4255/2020).
The authors’ contributions were as follows: E. K. and I. S. contributed to the conception and design of the study; E. K., I. B. and I. S. contributed to the acquisition, analysis and interpretation of data; I. S. drafted the manuscript; and E. K. and I. B. performed critical revision of the manuscript for important content. All authors critically revised the manuscript, provided their final approval and agreed to be accountable for all aspects of the work, ensuring its integrity and accuracy.
None of the authors has any conflict of interest to declare.
Supplementary material
For supplementary material referred to in this article, please visit https://doi.org/10.1017/S0007114522003671
