Trimester-Specific Reference Ranges for Saturated, Monounsaturated and Polyunsaturated Fatty Acids in Serum of Pregnant Women: A Cohort Study from the ECLIPSES Group

In the course of pregnancy, increasing importance is being placed on maintaining optimal fatty acid (FA) levels and particularly n-3 PUFAs to ensure correct fetal development. However, reference ranges for FA have been reported in only a few studies. Our objective is to provide quantitative reference intervals for SFAs, MUFAs, and PUFAs (n-6 and n-3) in a large population of healthy pregnant women from a developed country. A prospective study of pregnant women (n = 479) was conducted from the first trimester (T1) to the third trimester (T3). A total of 11 fatty acids were analyzed in serum by gas chromatography mass spectrometry and were expressed as absolute (µmol/L) and relative (percentage of total FA) concentration units. Serum concentrations of SFAs, MUFAs, n-6 PUFAs, n-3 PUFAs, various FA ratios, and the EFA index were determined. The reference intervals (2.5/97.5 percentiles) in absolute values from T1 ranged from 1884.32 to 8802.81 µmol/L for SFAs, from 959.91 to 2979.46 µmol/L for MUFAs, from 2325.77 to 7735.74 µmol/L for n-6 PUFAs, and from 129.01 to 495.58 µmol/L for n-3 PUFAs. These intervals mainly include the values of other studies from European populations. However, reference ranges vary according to some maternal factors. The FA levels proposed, obtained from a large sample of pregnant women, will be a useful tool for assessing the degree of adequacy of FAs in pregnant women and will help to carry out dietary interventions based on certain maternal factors.


Introduction
Maternal diet during the periconceptional, pregnancy and lactation period is important for both mother and child [1,2], and specifically, the fatty acid (FA) levels play a crucial role during pregnancy [3][4][5]. Eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA) are specially critical regarding central nervous system, retinal photoreceptors, and immune systems development [6]. Correct concentrations (or status) both in the first trimester (T1) and the third trimester (T3) of pregnancy are of great importance considering their role during the onset of neurogenesis [7,8], and the fetal brain development [8]. Consequently, maternal serum FA levels change in the course of pregnancy, depending on fetal requirements. In this context, it is necessary to monitor FA concentration values throughout pregnancy so that deficiency can be detected. Based on this, reference ranges are the most Low (primary or less) 30

Extraction, Transfer and Storage of Biological Samples
Serum samples were collected both T1 and T3 after fasting into 7.5 mL tubes without anticoagulant and were left without mixing for 30 min at room temperature to enable coagulation. The serum was separated by centrifugation and stored into aliquots of 500 µL at −80 °C inside the BioBank. Samples were thawed at the end of the study and processed simultaneously to minimize inter-batch variation [28]. The medical, socioeconomic, educational level and lifestyle information is shown in Table 1. The socioeconomic level was calculated by using the Catalan classification of occupations (CCO-2011) [31] such as student, employed and unemployed. The education level was classified as primary (low), high school (medium) and university studies or more (high). Physical activity was assessed by using the International Physical Activity Questionnaire [32] and was summed to obtain the total physical activity in metabolic equivalents (METs)/week. Based on total physical activity, participants were classified into 3 levels: low (<600 METs/week), moderate (≥600-2999 METs/week) and high (≥3000 METs/week). BMI was calculated and classified by following WHO criteria [33]: normal weight (BMI = 18.5-24.9 kg/m 2 ), overweight (BMI = 25.0-29.9 kg/m 2 ) and obesity (BMI ≥ 30 kg/m 2 ) at T1. The Spanish diet quality index (SQDI) [34] was estimated from nine food groups (protein foods, dairy foods, cereals, fruits, vegetables, oil, legumes, tubers, sweets). A score was obtained ranging from 0 points (low quality diet) to 18 points (high Nutrients 2021, 13,4037 4 of 18 quality diet). Women were then classified into two categories: low-medium diet quality (score from 0 to 10) and high diet quality (score from 11 to 18). Further information can be found in our previous paper [29].

Extraction, Transfer and Storage of Biological Samples
Serum samples were collected both T1 and T3 after fasting into 7.5 mL tubes without anticoagulant and were left without mixing for 30 min at room temperature to enable coagulation. The serum was separated by centrifugation and stored into aliquots of 500 µL at −80 • C inside the BioBank. Samples were thawed at the end of the study and processed simultaneously to minimize inter-batch variation [28].

Sample Preparation and GC-MS Conditions
The method starts with a 50 µL serum sample mixed with IS and the derivatization reagent (chloroform and methanolic hydrochloric acid). Then the sample is heated and mixed at 80 • C for 2 h. In this step, the lipidic fraction in blood or serum consisting of free FAs, sterol esters, glycerol esters (mainly triglycerides), and phospholipids were hydrolyzed and methylated into fatty acid methyl esters (FAMEs). Three sequential heating and mixing steps are done every 10 min during a 30-min period. After this, a 100 µL of iso-octane was added and the fatty acid methyl esters were extracted by a liquid-liquid extraction using hexane, and analyzed by gas chromatography mass spectrometry (GC-MS) combination using the 7890A GC coupled to triple quadrupole MS QQQ 7000 Series ® (Agilent Technologies Inc., Santa Clara, CA, USA). Chromatographic analysis was based on David et al. [35] to determine the 36 FAMEs included in the Food Industry FAME Mix (Restek Corporation). Briefly, the FAMEs were separated into a high-polarity column (100 m × 250 µm × 0.25 µm) (HP-88 column, Agilent Technologies) using a temperature program ranging between 140 and 240 • C at a 10 • /min pace using helium as the carrier gas at 1 mL/min. Ionization was carried out by electronic impact (70 eV), and the mass analyzer was operated in Selected Ion Monitoring (SIM) mode. The CG-MS system was controlled by the Agilent MassHunter ® Workstation.
FAs were expressed as absolute (µmol/L) or relative (percentage of total FA) concentration units. The method for calculating the percentage of total fatty acids is based on total FA concentration. To assist in the evaluation of overall nutritional status, reference intervals were determined following the Clinical and Laboratory Standards Institute (CLSI) C28-A3 guidelines [9] and represented the central 95% of the tested population (being the 2.5% and 97.5% confidence intervals the lower and upper limits, respectively). The FA values were analyzed using Agilent MassHunter ® Quantitative Analysis B.06 (Agilent Technologies Inc., Santa Clara, CA, USA). The results were expressed as mean ± standard deviation (SD) for normally distributed data. In this study, z-score analysis was used to detect outlier values in the population data [36]. An absolute z-score value above or below ±3.29 is considered to be an outlier when the sample size is >100 [37]. Correlations between the absolute and relative concentrations of FA in serum were computed using Pearson's correlation coefficient test. All multiple linear regression models were performed using the ENTER method for total FAs, total SFA, total MUFA, total n-6 PUFA, total n-3 PUFA, LA, DHA and AA to evaluate the relation between maternal factors and FA levels. The models were adjusted for maternal factors, such as maternal age, occupation (student, employed, unemployed), educational level (low, medium, high), ethnicity (Spanish, non-Spanish women), parity (no, yes), initial BMI (Kg/m 2 ), gestational weight gain (Kg/m 2 ), maternal smoking status (no, yes), maternal alcohol consumption (no, yes), physical activity in METs/week (score), and SQDI (score). Data was processed using the statistical software package SPSS version 25.0 for Windows and Microsoft Excel 2016 (Microsoft Corporation, Redmond, WA, USA). A p-value < 0.05 was considered to be statistically significant.

Participants' Characteristics
The general characteristics of the pregnant women participating in the ECLIPSES study are given in Table 1. A total of 455/476 women answered the questions about sociodemographic and lifestyle characteristics. Most women had a medium educational level (38.3%) and were employed (87.1%). The maternal age was 30.6 ± 5.01 years old. The participants reported that 15.3% smoked and 14% drank at the beginning of pregnancy. Regarding physical activity, 56.4% of the women had a low level.

Fatty Acid Status in Serum of Pregnant Women
This study determined saturated, mono-and polyunsaturated FAs in 476 maternal serum samples at T1 and T3. A total of 36 FAs were analyzed although only 11 FAs were detected in serum of pregnant women and were summarized in Tables 2 and 3. The FA composition in maternal serum is represented as absolute ( Table 2) and relative (Table 3) FA concentrations (mean ± SD) in both T1 and T3. The absolute total amount of FAs in maternal serum increased during pregnancy (T1, 10073.15 µmol/L and T3, 20,480.82 µmol/L, p < 0.01) and the highest individual FAs were C16:0, C18:1n-9 and C18:2n-6 at both T1 and T3.
The corresponding percentiles of FAs during T1 and T3 were calculated and showed in Tables 2 and 3 and expressed as either an absolute concentration or a percentage of total FAs. Consequently, our reference ranges for relative concentration can be compared to the results expressed as mean ± SD from other European studies [14,15,19,38,39] shown in Table 4. The levels of individual FAs that were distinct from the 2.5 or 97.5 percentiles of the present study are indicated in Table 4.

Influence of Maternal Factors on Fatty Acid Serum Profiles
Multiple linear regression models of the influence of maternal determinants on total FAs, total SFA, total MUFA, total n-6 PUFA, total n-3 PUFA, LA, DHA, and AA are shown in Table 5. It can be observed that women with BMI > 30 kg/m 2 had significantly higher levels of total SFA, total MUFA and AA in both trimesters of pregnancy. Moreover, educational level and ethnicity were significantly associated with higher values of total n-3 PUFA and DHA in T1, and lower values of total SFA, total MUFA and AA in T3. In T1, only low physical activity was associated with lower values of DHA. However, DHA and total n-3 PUFA were significantly higher in T3 at better diet quality and older age.

Influence of Maternal Factors on Fatty Acid Serum Profiles
Multiple linear regression models of the influence of maternal determinants on total FAs, total SFA, total MUFA, total n-6 PUFA, total n-3 PUFA, LA, DHA, and AA are shown in Table 5. It can be observed that women with BMI > 30 kg/m 2 had significantly higher levels of total SFA, total MUFA and AA in both trimesters of pregnancy. Moreover, educational level and ethnicity were significantly associated with higher values of total n-3 PUFA and DHA in T1, and lower values of total SFA, total MUFA and AA in T3. In T1, only low physical activity was associated with lower values of DHA. However, DHA and total n-3 PUFA were significantly higher in T3 at better diet quality and older age.

Influence of Maternal Factors on Fatty Acid Serum Profile
Multiple linear regression models of the influence of FAs, total SFA, total MUFA, total n-6 PUFA, total n-3 PUF in Table 5. It can be observed that women with BMI > 30 levels of total SFA, total MUFA and AA in both trimester cational level and ethnicity were significantly associated PUFA and DHA in T1, and lower values of total SFA, to only low physical activity was associated with lower valu total n-3 PUFA were significantly higher in T3 at better d

Influence of Maternal Factors on Fatty Acid Serum Profiles
Multiple linear regression models of the influence of maternal determinants on total FAs, total SFA, total MUFA, total n-6 PUFA, total n-3 PUFA, LA, DHA, and AA are shown in Table 5. It can be observed that women with BMI > 30 kg/m 2 had significantly higher levels of total SFA, total MUFA and AA in both trimesters of pregnancy. Moreover, educational level and ethnicity were significantly associated with higher values of total n-3 PUFA and DHA in T1, and lower values of total SFA, total MUFA and AA in T3. In T1, only low physical activity was associated with lower values of DHA. However, DHA and total n-3 PUFA were significantly higher in T3 at better diet quality and older age.

Influence of Maternal Factors on Fatty Acid Serum Profiles
Multiple linear regression models of the influence of maternal determinants on total FAs, total SFA, total MUFA, total n-6 PUFA, total n-3 PUFA, LA, DHA, and AA are shown in Table 5. It can be observed that women with BMI > 30 kg/m 2 had significantly higher levels of total SFA, total MUFA and AA in both trimesters of pregnancy. Moreover, educational level and ethnicity were significantly associated with higher values of total n-3 PUFA and DHA in T1, and lower values of total SFA, total MUFA and AA in T3. In T1, only low physical activity was associated with lower values of DHA. However, DHA and total n-3 PUFA were significantly higher in T3 at better diet quality and older age.

Influence of Maternal Factors on Fatty Acid Serum Profile
Multiple linear regression models of the influence of FAs, total SFA, total MUFA, total n-6 PUFA, total n-3 PUF in Table 5. It can be observed that women with BMI > 30 levels of total SFA, total MUFA and AA in both trimester cational level and ethnicity were significantly associated PUFA and DHA in T1, and lower values of total SFA, to only low physical activity was associated with lower valu total n-3 PUFA were significantly higher in T3 at better d

Influence of Maternal Factors on Fatty Acid Serum Profiles
Multiple linear regression models of the influence of maternal determinants on total FAs, total SFA, total MUFA, total n-6 PUFA, total n-3 PUFA, LA, DHA, and AA are shown in Table 5. It can be observed that women with BMI > 30 kg/m 2 had significantly higher levels of total SFA, total MUFA and AA in both trimesters of pregnancy. Moreover, educational level and ethnicity were significantly associated with higher values of total n-3 PUFA and DHA in T1, and lower values of total SFA, total MUFA and AA in T3. In T1, only low physical activity was associated with lower values of DHA. However, DHA and total n-3 PUFA were significantly higher in T3 at better diet quality and older age. MUFA, mo ounsaturated fatty acids; n-6 PUFA, omega-6 polyunsaturated fatty acid; n-3 PUFA, omega-3 polyunsaturated fatty aci ( a ) n-6/n-3 ratio = Σtotal n-6 PUFA/Σtotal n-3 PUFA; EFA index, Essential fatty acid index. (-) indicates that the fatty ac is not reported. The (*) and (ⱡ) denote: under 2.5 percentile and over 97.5 percentile of the present study, respectively.

Influence of Maternal Factors on Fatty Acid Serum Profiles
Multiple linear regression models of the influence of maternal determinants o FAs, total SFA, total MUFA, total n-6 PUFA, total n-3 PUFA, LA, DHA, and AA are in Table 5. It can be observed that women with BMI > 30 kg/m 2 had significantly levels of total SFA, total MUFA and AA in both trimesters of pregnancy. Moreove cational level and ethnicity were significantly associated with higher values of to PUFA and DHA in T1, and lower values of total SFA, total MUFA and AA in T3. only low physical activity was associated with lower values of DHA. However, DH total n-3 PUFA were significantly higher in T3 at better diet quality and older age.

Influence of Maternal Factors on Fatty Acid Serum Profile
Multiple linear regression models of the influence of FAs, total SFA, total MUFA, total n-6 PUFA, total n-3 PUF in Table 5. It can be observed that women with BMI > 30 levels of total SFA, total MUFA and AA in both trimester cational level and ethnicity were significantly associated PUFA and DHA in T1, and lower values of total SFA, to only low physical activity was associated with lower valu total n-3 PUFA were significantly higher in T3 at better d

Influence of Maternal Factors on Fatty Acid Serum Profiles
Multiple linear regression models of the influence of maternal determinants o FAs, total SFA, total MUFA, total n-6 PUFA, total n-3 PUFA, LA, DHA, and AA are in Table 5. It can be observed that women with BMI > 30 kg/m 2 had significantly levels of total SFA, total MUFA and AA in both trimesters of pregnancy. Moreove cational level and ethnicity were significantly associated with higher values of to PUFA and DHA in T1, and lower values of total SFA, total MUFA and AA in T3. only low physical activity was associated with lower values of DHA. However, DH total n-3 PUFA were significantly higher in T3 at better diet quality and older age. AA, Arachidonic acid; EPA, Eicosapentaenoic acid; DHA, Docosahexaenoic acid. SFA, saturated fatty acids; MUFA, mon-ounsaturated fatty acids; n-6 PUFA, omega-6 polyunsaturated fatty acid; n-3 PUFA, omega-3 polyunsaturated fatty acid; ( a ) n-6/n-3 ratio = Σtotal n-6 PUFA/Σtotal n-3 PUFA; EFA index, Essential fatty acid index. (-) indicates that the fatty acid is not reported. The (*) and (ⱡ) denote: under 2.5 percentile and over 97.5 percentile of the present study, respectively.

Influence of Maternal Factors on Fatty Acid Serum Profiles
Multiple linear regression models of the influence of maternal determinants on total

Discussion
The results of present study provide a reference range for the vast majority of serum circulating FAs and FAs ratios in 479 healthy pregnant women from the ECLIPSES study [28], assessed both in T1 and T3 of gestation. These results complement the few studies on reference ranges of FAs that exist in pregnant women for different moments of gestation, and add a new measurement with FAs index. Furthermore, this study describes the variation of FA levels in serum according to some maternal factors such as age, obesity, sedentary lifestyle, lower educational level, low diet quality, parity and non-Spanish ethnicity.
In general, any laboratory test is compared with reference ranges in order to assess the degree of compliance with normality levels, both individually in the clinical follow-up of the subject and collectively in epidemiological studies [9]. Adequate levels of FAs in T1 are of great importance considering the part they play in the onset of neurogenesis [7,8]. EPA and DHA are particularly critical for supporting the development of the central nervous system, retinal photoreceptors and immune system [6]. Nevertheless, few investigations have reported reference intervals for serum individualized FA concentrations, mainly due to the technical complexity of FAs evaluation compared to other nutrients or biomarkers. In the scarce literature available, FA reference ranges for serum are reported in non-pregnant adults [10][11][12][13]. Geographical region, sex, age, physio-pathological conditions [13,22] such as pregnancy [17] and dietary pattern have been seen to lead to some discrepancies. Kish-Trier et al. [40] and Mayo Clinic [12] separately provided reference intervals in age groups from the North American population with somewhat different results. Abdelmagid et al. [11] also reported significant differences between Canadian male and female adults in such FAs as LA or DHA. For pregnant women, reference values have only been described in a Brazilian cohort of 225 healthy pregnant women [16] and in a Norwegian cohort of 247 healthy pregnant women [17]. However, the concentration units (relative terms) and type of sample (erythrocyte membrane) evaluated in these two cohorts cannot be compared with our work methodology and our results. In this regard, reference intervals should be developed using a more uniform and systematic process. In any case, our study provides new information on the European Mediterranean region with socioeconomic, cultural and dietary characteristics different from the previous ones.
Depending on concentration units, the reference intervals of individual FAs in serum can be expressed as either absolute concentrations or percentages of total FA. To the best of our knowledge, no study has reported serum reference ranges in both types of concentration units in healthy pregnant women. To make our results comparable with those of previous studies, reference ranges were reported as relative concentrations because very little literature reports absolute values in pregnant women. Correlations between FA concentrations expressed in absolute and relative values were low-moderate (r = 0.1-0.5) with the exception of C12:0, C16:0, Total SFA and C20:5n-3 for which a strong correlation was observed (r > 0.5). Identically to our results, other studies have already noted poor correlations between absolute and relative concentrations [40][41][42][43] in non-pregnant adults. It is postulated that differences between percentages and concentrations depended on individual FA-characteristics [43]. Moreover, the difference in the correlation could be due to the fact that the percentage of individual FA is calculated based on the whole set of FAs detected, however absolute concentration of individual FA is not dependent upon the relative abundance of other FAs. For this reason, we believe that the measurement expressed in absolute terms would facilitate comparisons among studies.
Our relative results were compared with those reported in other European studies ( Table 3). As expected, in the Mediterranean diet pattern Σtotal MUFA levels appear to be higher in the study from Spain [19] and lower in non-Spanish countries such as the study from the Netherlands [14,38]. Surprisingly, similar levels of Σtotal MUFA have also been observed in Germany [15] and in UK [39], which are not characterized by following a Mediterranean pattern diet. DHGLA and AA values were higher in the Netherlands than other countries. Normally, the production of inflammatory mediators such as AA is associated with the course of pregnancy. However, abnormally high levels have been related to pathological pregnancy complications such as preeclampsia, premature labor, fetal growth, among others (especially in women with risk factors for overweight and obesity) [27]. Some of these complications occur during T3, however Otto et al. [14] measured FA during T1. Furthermore, certain maternal factors such as obesity predispose to suffer a greater number of complications and the study of Vlaardingerbroek. et al. [38] did not report the BMI of their pregnant women population. In addition to the above, DHA concentrations from the Netherlands are higher than those reported in the other studies. Possible explanations for increased DHA in the Netherlands might be the high intake of n-3 supplements or the higher consumption of seafood per capita [17]. Finally, the differences in the measurement time window, sample size and FA assay protocol could explain the difference found with these authors [14,15,19,38,39]. Nevertheless, it is important to note that absolute concentrations are more useful and facilitate comparisons between studies and the recommendation for future studies is to report FA levels as absolute units. Further, absolute values consist in an individual measure that assure an economic analysis compared with relative values that require the measurement of all circulating FAs in serum. In this regard, absolute analysis could be included in routine clinical practice of laboratories and not be limited to research studies.
Regarding the evolution of FA levels between T1 and T3, we conclude in an article from the ECLIPSES study, that serum FA profile changed during the gestational period: Σtotal SFA, Σtotal MUFA and Σtotal n-6 PUFA increased during pregnancy, whereas essential FAs such AA and EPA decrease and DHA remains unchanged [44]. It is well-known that the composition in serum of essential FAs such n-6 PUFA and n-3 PUFA is closely linked to dietary FA consumption [18,19]. However, Volk et al. [45] showed that dietary and plasma SFA and MUFA are not related. In addition, some authors reported likewise to our results that FAs intake and the dietary pattern remained similar throughout pregnancy [14,29]. In this sense, diet would not explain the variations observed in serum FA composition and levels, including the lack of correlation between absolute and relative concentrations in our results, which could be related to other factors in combination with physiological changes in plasma volume during pregnancy.
We found that the women with obesity, sedentary lifestyle, lower educational level and those of non-Spanish ethnicity had higher levels of total SFA, total MUFA and AA, and lower levels of total n-3 PUFA and DHA. Factors such as obesity and sedentary lifestyle, could favor the presence of proinflammatory FAs such as AA. Ma et al. showed that overweight, smoking and alcohol drinking can increased the endogenous synthesis of SFA [46]. Other studies showed that variations in FA concentrations could be affected by the genetically conditions, such as genotype expression of fatty acid desaturase [25,47]. Overall, it is important to note that lifestyle, sociodemographic and genetic factors can affect normal FA reference intervals in pregnancy. Therefore, serum FA measure could be considered as an integration of dietary intake and individual biological response [46]. Consequently, absolute percentiles of FAs for 95% of the population could assist the evaluation of FA status and the identification of women at risk of either under exposure (≤2.5% percentile) or over exposure (≥97.5% percentile) to an individual FA.
One of the strengths of this study is that it is the first one to define individual SFA, MUFA, and PUFA reference ranges in absolute and relative units in a large sample of healthy women at the beginning and end of pregnancy. The representativeness of the sample size emphasizes the validity of the proposed reference values in the present study. All data were collected by the same trained researchers, and all FA analyses were performed in the same research laboratory. Furthermore, FAs were analyzed by GC-MS which is the more recently clinically validated method for individual FA quantitation [40] and can be applied for fully automated FA profiling in serum samples during dietary studies [35]. Further, our results would facilitate future research into the role of high/low concentrations of individual FAs in obstetric and fetal pathologies. Nevertheless, it does have several limitations. First, FAs were not analyzed in the second trimester of pregnancy and it would have been interesting to monitor the whole gestational period. Second, we were unable to assess all FAs in our sample. One of these was ALA, one of the main n-3 PUFAs, which may explain why the n-3 PUFA relative levels in our study were lower than those in other countries. Third, it is difficult to establish a comparison between countries because the FA profile analyzed in total serum/plasma lipids can vary from different populations, included in the same country; further the measurement of FAs in all circulating lipids differ from FA profile performed in circulating triglycerides, phospholipids or esters of cholesterol.

Conclusions
The present study provides the serum reference interval for FA levels in a large sample of pregnant women from a Mediterranean region. The data are reported in the first and third trimesters of pregnancy, expressed both as absolute values of their serum concentrations, and as relative values in terms of the percentage of total FA. The percentiles of reference ranges proposed will be a useful tool for assessing the degree of adequacy of FAs in pregnant women, in both individual monitoring of pregnancy, and populationbased epidemiological studies. However, every region must provide their own reference ranges in accordance with the characteristics of their population. Further research is needed to identify the personal and environmental factors that contribute to a healthy diet during pregnancy.