Peri-conception maternal lipid profiles predict pregnancy outcomes

Author’s Accepted Manuscript Peri-Conception Maternal Lipid Profiles Predict Pregnancy Outcomes Enitan Ogundipe, Mark R. Johnson, Yiqun Wang, Michael A. Crawford www.elsevier.com/locate/plefa

PII: DOI: Reference:

S0952-3278(16)30039-4 http://dx.doi.org/10.1016/j.plefa.2016.08.012 YPLEF1772

To appear in: Prostaglandins Leukotrienes and Essential Fatty Acids Received date: 18 April 2016 Revised date: 21 August 2016 Accepted date: 22 August 2016 Cite this article as: Enitan Ogundipe, Mark R. Johnson, Yiqun Wang and Michael A. Crawford, Peri-Conception Maternal Lipid Profiles Predict Pregnancy Outcomes, Prostaglandins Leukotrienes and Essential Fatty Acids, http://dx.doi.org/10.1016/j.plefa.2016.08.012 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting galley proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Peri-Conception Maternal Lipid Profiles Predict Pregnancy Outcomes Enitan Ogundipea, Mark R. Johnsonb, Yiqun Wangc, Michael A. Crawfordb, a

Neonatal unit, Chelsea and Westminster Hospital, London, Division of Medicine,

Imperial college, London, UK b

Division of Obstetrics and Gynaecology, Department of Surgery and Cancer, Imperial

College London, Chelsea and Westminster Hospital Campus, London, UK c

Division of Medicine, Imperial College London, Chelsea and Westminster Hospital

Campus, London, UK

*Correspondence should be addressed to: Michael A Crawford, Visiting Professor Division of Obstetrics and Gynaecology, Department of Surgery and Cancer, Faculty of Medicine, Imperial College London, Chelsea and Westminster Hospital, 369 Fulham Road, London SW10 9NH, United Kingdom; E-mail: [email protected], Tel. +44 (0) 2033157899, Fax. +44 (0) 2033153090

Sources of Funding: The Mother and Child Foundation, Letten Foundation, Waterloo Foundation and Vifor Pharma, Switzerland.

Abbreviations: AA, Arachidonic acid; DHA, Docosahexaenoic acid; ROC, Receiver operating characteristic; EFA, Essential fatty acids; ALSPAC; The Avon Longitudinal Study of Parents and Children; EPA, Eicosapentaenoic acid; GDM, Gestational diabetes mellitus; PET, Pre-eclamptic toxaemia; LBW, Low birth weight; HIV, Human immunodeficiency virus; NHC, Normal healthy controls; GLA, Gamma-linolenic acid; BHT, Butylated hydroxytoluene; FAME, Fatty acid methyl esters; AUC, Area under the curve; MUFA, Mono-unsaturated fatty acids; SAG, sn-1-stearoyl-2-arachidonoylglycerol; LC-PUFA, Long-chain polyunsaturated fatty acids; RBC, Red blood cells; BMI, Body mass

index;

BW,

Birth

weight;

LC,

Long

chain.

1

Abstract In this study, healthy women and those at high-risk of adverse pregnancy outcomes (pre-eclampsia, fetal growth restriction, gestational diabetes) were selected to assess the effect of fatty acid supplementation. The purpose of this paper is to report two novel findings (i) at recruitment the receiver operating characteristic (ROC) for erythrocyte oleic acid predicted spontaneous delivery at 34 weeks gestation (ROC=0·926 n=296) for all women entering the study). Further analysis revealed oleic and all monounsaturated fatty acids were similarly predictive with or without a supplement during the pregnancy. (ii) At delivery, we observed a biomagnification of saturated fatty acids from mother to fetus with the reverse for monounsaturates. The major conclusions are (i) the status of the mother in the months prior to conception is a stronger predictor of preterm delivery than the events during the pregnancy. (ii) Saturated fats may be playing an important function in supporting fetal membrane growth.

Key words: pregnancy outcomes, birthweight, docosahexaenoic acid, oleic acids, saturated fatty acids, monounsaturated fatty acids

1. Introduction “We know now that if children under two do not receive sufficient nutrition they will be sentenced to a lifetime of mental and physical limitations” [Josette Sheerhan, report to the Board of UN World Food Programme 2010] There is robust evidence for the essentiality of long chain essential fatty acids for brain development and function [1]. The wealth of experimental evidence dating back to the 1970s [2] is supported by the Avon Longitudinal Study of Parents and Children (ALSPAC) study of over 14,000 pregnancies which found that the mothers’ dietary intake of long chain omega 3 fatty acids from fish and sea foods, were predictive of the verbal IQ and behavioural scores of the children at 8 years of age [3]. This evidence has led to randomised clinical trials, the largest of which was the DoMINO study [4] on mothers at risk of gestational diabetes but their results regarding cognitive development 2

enhancement have been disappointing. With respect to preterm labour, many trials on the supplementation of fatty acids have focussed on DHA, an omega-3 fatty acid (ω-3 FA) [5]. While some studies showed that omega-3 supplementation during pregnancy resulted in an increased length of gestation and increased birth weight, a recent metaanalysis refuted the above [6]. However, such studies are based on triglyceride supplements and often late during the pregnancy. A large part of neurogenesis takes place very early with neurons in place ready to migrate to form the cortex at the time a woman might report for her first antenatal visit. In addition, cells use phosphoglycerides whereas triglycerides are stored in adipose fat.

The change of hormone profile in

pregnancy moreover, leads to the deposition of triglycerides in fat stores, which might frustrate the use of triglyceride supplements. The failure of such studies does not dispute the demand for long chain super-unsaturated fatty acids for brain growth. The importance of lipid or fat nutrition during pregnancy is that the priority in human early development is the brain and the brain is a fat rich organ with over 60% of its structural material a highly specialised fat. The requirement for long chain super unsaturated fatty acids specifically includes arachidonic acid (AA), adrenic acid and DHA for structure, growth and function [7]. Hence nutrition of the brain is fundamentally different to that of the protein needs for the body on which present day food and agricultural policies are based. This fact combined with the rise in mental ill health provides the rationale for seeking a better understanding of lipid nutrition in pregnancy. The critical period for organogenesis and brain development is early after conception when neurolocation takes place and the cardiovascular system emerges from the embryo to initiate organogenesis. Neurogenesis and cell migration follows. In the last trimester the brain growth spurt exerts a quantitative demand for structural lipid and highly specific fatty acids. In 1976, we reported on the principle of bio-magnification of the super-unsaturated fatty acids, which meets this demand by specifically, and selectively increasing the proportion of AA in the fetal circulation and DHA in favour of the fetal brain. This enhancement is consistent with the need for both fetal vascular and neural development [8]. Experimental lack of DHA has been shown to restrict neurogenesis [9]. With the food system increasingly based on protein and energy and

3

with low fat diets now in vogue and mental ill health rising, it seems appropriate to reexamine the fatty acids status during pregnancy. For this re-examination , this study assessed the membrane status for essential fatty acids at recruitment and delivery in women who were part of a randomised trial of a ’fish oil’ supplement rich in DHA and AA. The red cell lipid levels were taken as a marker on the basis that “the tissue is the issue” a comment from Bill Lands to emphasise thar diet may have an effect on many biological facets but ultimately the impact has to be felt in the tissues as a combination of both diet and genetic variance. In the case of lipids the impact will be reflected in the quantitative and qualitative nature of the cell m membranes. . In this paper we have used the red cell membrane as a representative tissue of plasma membranes as in endothelial tissue. The evidence presented here we believe can help explain the low efficacy of the clinical trials with eicosapentaenoic acid (EPA) and DHA supplements and point to the implications for public health.

2. Materials And Methods 2.1. Study Population

The study population were women booked to deliver at the Chelsea and Westminster Hospital, London, UK. They were part of a double blind randomised controlled trial studying the effect of essential fatty acid supplementation in high risk pregnancies compared to normal healthy controls. The data presented here will focus on the recruitment data in relation to spontaneous preterm birth and birthweight. Many women today take supplements during pregnancy. This use and wide variations in diet will affect tissue lipids to a varying degree depending on genetics and several other nutritional, life style and environmental factors. Hence the analysis of the whole population as opposed to the control groups has relevance to practice. The results pertaining to the effect of supplements will be discussed in a subsequent paper.

Exclusions were: Women with known allergy to fish and fish oil, Non-English speaker who decline the use of an interpreter and those who are unable or unwilling to attend 4

follow

up

appointments.

Women

with

chronic

disease

such

as

human

immunodeficiency virus (HIV), cirrhosis or other chronic liver disease, Hepatitis B and C carriage Women previously on regular pre-conceptual fish oil supplement or for different reasons were not able to give competent, written consent.

The total number of subjects recruited was 300, divided as follows: Normal healthy controls (NHC) (n= 50), women identified at risk of having a low birthweight baby either spontaneously (n = 100) or because of developing pre-eclampsia (n = 100), and women at risk of gestational diabetes (n= 50). Table 1 provides details of the subjects in this study. Supplementation was 2 capsules daily of either “DHA-enriched formula” or “placebo (high oleic acid sunflower seed oil)”. Each active supplement capsule contained 300mg of DHA, 42mg of eicosapentaenoic acid (EPA) and 8.4 mg of AA, and placebo 721 mg of oleic acid (Vifor Pharma Switzerland). In a previous study of 513 pregnancies, we found that maternal intake for 14 components of diet fell progressively in a dose responsive manner as birthweight fell, but only for the mothers of smaller babies below 3270g [12]. We used 3,200g as a pragmatic dividing birthweight to test for those at a nutritional risk for low birthweight. In those above 3,200g we would have expected differences in maternal nutrition to be powerless to influence birthweight. Blood specimens: 5mls of blood was taken from pregnant women at first booking, at delivery and from cord blood. Blood cells separated from the serum were collected and stored at -70C until required for analysis. The study was conducted according to guidelines laid down in the Declaration of Helsinki and all study procedures involving participants were reviewed and approved by the East London Research Ethics Committee. The study was indemnified by the Chelsea & Westminster NHS Healthcare Foundation Trust.

2.2. Fatty Acid Analysis

Total plasma and blood cell lipids were extracted by the method of Folch and Stanley [11] by homogenising the samples in chloroform and methanol (2:1 v/v) 5

containing 0·01% butylated hydroxytoluene (BHT) as an antioxidant under N 2. Fatty acid methyl esters (FAME) were prepared by heating the lipid fractions with 5ml of 15% acetyl chloride in methanol in a sealed vial at 70ºC for 3 hours under N2. FAME were separated by a gas liquid chromatograph fitted with a capillary column (25m X 0·32mm ID, 0·25µ film, BP20). Hydrogen was used as a carrier gas, and the injector, oven and detector temperatures were 235, 210 and 260ºC. The FAME were identified by comparison of retention times with authentic standards and calculation of equivalent chain length values confirmed by mass-spectrometry. Peak areas were quantified by a computer chromatography data system (EZChrom Chromatography Data System, Scientific Software, Inc., San Ramon, CA).

2.3. Data analyses Receiver operating characteristic (ROC): When using normalized units, the area under the curve (often referred to as simply the AUC, or AUROC) is equal to the probability that a classifier will rank a randomly chosen positive instance higher than a randomly chosen negative one. In a ROC curve the true positive rate (Sensitivity) is plotted in function of the false positive rate (10-Specificity) for different cut-off points of a parameter. Each point on the ROC curve represents a sensitivity/specificity pair corresponding to a particular decision threshold. The AUC is a measure of how well a parameter can distinguish between two diagnostic groups (diseased/normal) and are consequently useful measure for predicating outcomes. Data are expressed as mean  standard deviations. The strength of association between AA and DHA was calculated as the Pearson Correlation Coefficient, r using a statistical package, SPSS for Windows (Release no 22) to analyse the data. The key ROC curves with area under the curve greater than 0·70 are reported below. The principle adopted is the “Tissue is the Issue”. Whilst dietary analyses are useful the ultimate lipid arbiter of cell membrane function is the composition achieved. Membrane lipid composition and function will depend on genetic differences and a wide range of lipid and non-lipid influences and even the extent and type of exercise, 6

emotional trauma and sleep. The lipid composition of the red cell is essentially that of a plasma membrane and similar to that of the vascular endothelium. It contains significant amounts of AA and adrenic acid (the omega 6 docosatetraenoic acid) and DHA. All three are strongly represented in neural membrane lipids. The half-life of the red cell being 120 days, the compositional data provides an integrated image of the dietlifestyle-genetic-metabolic sense that ends up as membrane composition.

3. Results Table 1: describes maternal demographic characteristics of the population of women who entered the trial. In the tables, the proportion of fatty acids in the red cell membrane is presented as percentages of the individual fatty acids in the blood cells from the mother and fetus (cord blood) at delivery. The time of sampling is designated as follows: 

Recruitment (R)



At delivery (D)



And the cord blood of the neonate at birth (C).

Table 2: The change in lipid profiles at recruitment to delivery and the cord blood findings are given. The data shows significant bio-magnification of saturated fats (16.0 and 18.0) from mother to fetus and the converse with the mono-unsaturated fats.

Figures 1 & 2: Show the effects mentioned in Table 2 graphically

Table 3: Data is reported on the relationship of Lipid Profiles to Infant birth weight, gestational age and head circumference by sub – groups. Data are presented only when there is more than one statistically significant figure in one row to avoid type I errors. This table shows how the Pearson Correlation Coefficients in most cases are strongest for birthweights below 3,200g. For example, for birthweight, recruitment AA of R = 0·134 is insignificant but at birthweights below 3,200g is 0·388 p = 0·000· It is also significant for gestational age at delivery. Moreover the ratio between AA and oleic acid was significant for all three categories. DHA was significant for all 3 categories and 7

again the DHA/oleic acid ration was significant across all three. That is the lower the AA and DHA in the RBC lipids the higher the oleic acid consistent with the strong independent ROC for oleic acid predicting preterm delivery at 34 and 39 weeks and also for prediction of low birthweight. Further confirmation for this status comes from summating AA and DHA and dividing by oleic acid. That and (AA+DHA)/MUFA gave the strongest correlation coefficients in table 3. For example, the sum of all ω6 and all ω3 divided by total PUFA (1ω6+1ω3) / MUFA; had a correlation coefficient at 0·551 p<0·000 with those eventually born below 3,200g. The data are clearly telling us that the influence of

essential fatty acids on pregnancy outcome is not one fatty acid or another: it is a global profile for optimum membrane function that is at stake. Doubtless the significance of this optimum profile is the need for rapid cell membrane division and growth involved extremes of membrane curvature and the optimum performance of the membrane bounds proteins, the transporters, signalers, ligands for nuclear receptors, ion channels and the rest.

Figure 3: The ROC curve shows a significant relationship between oleic acid levels at recruitment and subsequent spontaneous preterm birth at 34 weeks or less. The new finding was the predictive value for maternal red cell membrane status for oleic acid at recruitment with eventual birthweight and gestational age. For the receiver operating characteristic (ROC), the predictive area under the curve in the whole population prior at randomisation (n=296) was 0·926 for preterm delivery at 34 weeks. This was a powerful predictor of spontaneous preterm birth at 34 weeks, which operated regardless of whether or not the mother took a fatty acid supplement, developed gestational diabetes or became hypertensive.

Figures 1 and 4, Table 5: Data show that Oleic acid and MUFA predict prematurity and low birthweight at recruitment

Table 5: This shows data at recruitment, prior to supplementation, for the ROC curve for all study mothers in relation to gestational age. We have used this subset to relate it to infant outcome and to avoid any confounding effects of the supplement by 8

the time of birth. It also confirms the predictive data for oleic acid at recruitment predicting preterm delivery at 34 weeks gestation using the data at recruitment for the whole population (Figure 1). The results support the initial analysis above, showing a strong association between oleic acid in the red cells taken at recruitment (12 weeks) and preterm birth at 34 and 30 weeks. Analysing the data for the separate groups of all randomised women at recruitment, those who had a supplement and those who did not revealed a similar predictive value in all three groups viz all randomised ROC= 0·802 n=280, control 0·869 n=136 and supplement 0·756 n=144. We have used this subset data to relate it to risk of preterm birth and to avoid any confounding effects of the supplement by the time of birth. It also confirms the predictive data for oleic acid at recruitment predicting preterm delivery at 34 weeks gestation using the data at recruitment for the whole population (Figure 1). The effect of the supplement on fatty acid composition will be reported in separately in a subsequent paper.

Figure 4: It can be seen that oleic acid and all mono-unsaturated fatty acids correlated negatively at recruitment with birthweight and gestational age in all control mothers. The prediction also included head circumference for births <3,200g. We have excluded those with diabetes because some born from diabetic mothers can be exceptionally large.

3.1. Relationships at recruitment and birth in healthy controls We have presented data set in Table 3 for significant correlations at recruitment, the mother at birth and cord blood which provides confirmation of the RPC data. The data is for all birthweights and those <3,200g and gestational age for all healthy mothers which included the normal healthy control group and those considered to be at risk of low birthweight at recruitment (see above) but otherwise healthy. It is especially interesting to see the strength of the correlations of various fatty acids in the red cell membrane at recruitment and relatively little in the maternal RBC at birth. All MUFAs at recruitment had negative r values for all 3 outcome measures of birthweight, birthweight below 3,200g and gestational age. 9

For most, the correlation coefficients were strongest for those born below 3,200g. Above 3,270g we previously found no relationship with birthweight and maternal macro and micronutrient intakes during pregnancy [12]. Below 3,270g there was a significant and dose response relationship. For this paper we chose the arbitrary cut off point as 3,200g. This data again emphasises the point that nutritional impacts under present circumstances are best studied in those at highest risk. Averaging the data for a whole population dilutes the test as it includes the large proportion of those in whom any real relationship is marginal or non-existent. The sum of both AA and DHA correlated with birthweight (r=0·286, p< 0·000, at birthweight below 3,200g r= 0·467 p<0·000 and for gestational age r=0·383, p<0·000) (Table 3). In figure 3 we have illustrated the correlations at recruitment for oleic acid and all monounsaturated fatty acids which correlated negatively with birthweight and gestational age in all control mothers. By comparison with the data at recruitment there was little data of significance at delivery either in the mother. In the fetus at birth all the omega 6 r values were negative but not statistically significant. The DHA and the omega 3 index were positive for those born below 3,200g (Table 3 for DHA r=0·381, p=0·018). AA levels were insignificant for all outcomes in the cord blood. The negative result in the mother and null result in the cord red cells could be influenced by the powerful bio-magnification AA from mother to fetus (Table 2).

The data for head circumference are presented in Table 4. Again, the strongest coefficients were mostly observed for births <3200g except on the low birthweight group. In the normal healthy control group there appeared to be highly statistically significant correlations in the small number of healthy controls and a strong relationship with Mead Acid C20:3,n-9 (0·642 p=0·004 n=18) in the cord blood was shown. Some of the very high correlations in maternal and cord blood of the NHC group (r>0·8) were obtained in small numbers of only six cases where we had the coincident data and should be viewed with caution. These could be statistical artefacts and will be examined further. 10

Biomagnification

3.1.1. Saturated fatty acids: (Figure 2, Table 2)

As far as we know the bio-magnification of the saturated fatty acids has not been reported before. All saturated fatty acids showed bio-magnification from the mother to the fetal circulation measured at birth (16:0, 18:0, 20:0, 22:0 and 24:0) (Figure 2). Interestingly the greatest concentration for the fetus detected was for stearic acid (11.7±1.29 vs 14.9±1.82) a 27% enhancement, and lignoceric acid (2.43±1.87 vs0· 3.17±0·68). The peak sizes for the 20 and 22 chain length fatty acids are small but the consistency provides some confidence that this is a true result.

3.1.2. Monounsaturated fatty acids (Figure 4 and Table 2)

By contrast, the biomagnification was reversed for the monounsaturated fatty acids with higher proportions in maternal compared to fetal red cells. The oleic acid proportion in maternal blood was higher at delivery compared to recruitment (15.2±2.03 n=139 compared to 16.3±1.87 n=82) but unlike the recruitment data it was not related to birthweight or gestational age at delivery. Cord blood oleic acid was significantly lower than the maternal proportion at delivery (11.7±1.46 n= 58 cfd 16.3±1.98) Even nervonic acid (24:1,ω9) exhibited a small decline across the placenta (3.07 ± 0·64 vs

2.62±0·55). That is the placenta is conserving MUFA for the mother and

enhancing saturated fatty acids for the fetus.

3.1.3. Polyunsaturated fatty acids (Table 2) 11

The fatty acid proportion of precursors for AA and DHA were reduced in cord blood by the placenta as found previously [8]. The linoleic acid proportion in the cord blood was lower at 4.42%±1.62 n= 58 compared to 12.6%±1.58 n= 82 (p<0·01) in the mothers: a threefold difference. At the same time the AA portion was incremented from 10·8%± 2.15 to 15.1% ± 3.38 (p<0·01) in the red blood cells (RBC). The biomagnification for the total long chain omega 6 fatty acids was from 14.7%±2.76 in the mother to 21.0%±4.54 p<0·01. This larger effect is mainly accounted for by the biomagnification of adrenic acid (C22:4,ω6) with 1.61%±0·53 in the maternal circulation and 2.49%±0·65 in the fetal blood at delivery. The situation is especially interesting with the 3 family because there is relatively little precursor in the maternal circulation and this is reduced even further in the fetus e.g. α-linolenic acid at 0·26%±0·14 reduced to 0·09% ±0·16 and eicosapentaenoic acid (EPA) was also lower in the fetus; i.e. 0·49 ± 0·28 to 0·27 ± 0·15. Docosapentaenoic acid (DPA - C22:5,ω3) was similarly low in proportion at term in the mother than at recruitment (1.68%±0·32 and at delivery 1.45%±0·44 which was reduced even further in transit across the placenta at 0·60%±0·19, p<0·01). By comparison with AA, the biomagnification of DHA captured in the red cells was relatively small (4.17% ± 1.18 compared to 4.89% ± 1.41) which was not statistically significant. None the less the proportion of DHA is known to be biomagnified from cord blood to fetal liver and fetal brain [13]. The lack of precursor emphases the significance of the bio-magnification process, selectively incorporating DHA into the membrane phosphoglycerides preformed rather than the operation of desaturation (FADS) enzymes.

3.6 Ratios

Some of the relationships scored highly when tested not as individual fatty acids but in relation to another implying that the profile was more significant than the individual quantity. Arachidonic acid individually was significant at recruitment for those born below 3,200g and for gestation. However AA/oleic acid was significant in all 3 12

parameters with greater R values. DHA/oleic acid similarly was stronger than for DHA alone. Indeed the ratio of the sum of AA+DHA divided by MUFA and Oleic acid provided an even greater R value. These results are consistent with the negative value for oleic acid and MUFA on their own. The omega 3 Index at recruitment was significant for all 3 parameters potentially linking risk of brain and cardiovascular disorders (13).

There were surprisingly few correlations at delivery in the mother and fetus. Notably AA and all long chain ω6 were negative in contrast to recruitment which could be a consequence of the high rate of bio-magnification for AA seen in Table 2. In cord RBC at delivery, again, the ratios provided stronger information than the individual fatty acids. All ω3, all long chain ω3, and DHA were only found to be significant in cord blood for those born below 3200g. On the other hand the ratios of AA/DHA, all ω6/ω3 and the ratio for all long chain ω6 and ω3 were negative for all three parameters with the AA/DHA ratio being the strongest at -0·551 p<0,000· n=38 for those born <3.200g. That is the lower the DHA in relation to AA the greater the risk for birthweight <3200g. The negative relationship also existed for gestation (AA/DHA r= -0·4 p<0·000 n= 182).

.

3.2. Membrane polyunsaturation

The loss of double bonds from the return of MUFA and linoleic acid to the mother compared to the gain by the fetus is nearly balanced. A calculation of the number of double bonds in all polyunsaturated fatty acids returned to the mother yields 27 and the number gained by the fetus is 29. The closeness of these two numbers implies the biomagnification process is not to increase the number of double bonds or degree of polyunsaturation but more likely a very specific selection of the fatty acids for membrane development in the fetus. Evidence for this high selectivity is in the strong AA to DHA ratio in the fetus which had an r value of 0·551 p<0·000 n=38 for birthweights below 3,200g.

13

4. Discussion The three unexpected findings collectively have significance for nutritional science and practice with its relevance to pregnancy, birthweight, gestational age and our understanding of fatty acid biology.

4.1. Oleic acid and MUFA at recruitment predicts adverse pregnancy outcome

The higher the oleic or MUFA in the red cell at recruitment (12 weeks) the more likely the babies were more often preterm and therefore smaller. The analysis of the whole population (n = 296) reported a receiver operating characteristic predicting birth at 34 weeks gestation at 0·926 (p<0·000). This relationship held at 30 weeks and at term. Further analysis of apparently healthy mothers who did not take a supplement was also significant. Although a smaller number, the ROC at 30 weeks was at 0·852 (p=0·017). Birthweight similarly correlated. This is not to say that conditions or interventions during the pregnancy are irrelevant. However, the red cell data implies that a major determinant of a mature pregnancy and safe birthweight is the condition of the mother leading up to pregnancy; possibly over some months before conception. This result is consistent with the notion that nature prepares in advance for important physiological events. Another example would be the deposition of fat during the pregnancy resulting in a fat store, which guarantees one third of the cost of lactation over the first 100 days [14]. The metabolomic significance of the oleic and MUFA relationships is most likely as a marker for an inadequate provision of dietary essential fatty acids in general. Essential fatty acid deficiency is met by an attempt to hold the degree of unsaturation (double bonds) in membrane lipids stable [14] through replacement by oleic acid and its long chain more unsaturated derivative the eicosatrienoate, the Mead acid1 .

1

Viz: 18:1ω9 > 18:2ω9 > 20:2ω9 > 20:3ω9 (Mead Acid)

14

The identification of Mead Acid as a marker for essential fatty acid deficiency was made in rats fed a diet free of essential fatty acids. Such extreme conditions are unlikely in human practice. With the desaturation process limited in humans [10] the ability to convert oleic acid renders the Mead Acid barely detectable. The first FAO-WHO joint consultation on the Role of Dietary Fats and Oils in Human Nutrition [14] makes the point that essential fatty acid deficiency is accompanied by an increase in palmitoleic and oleic acids to compensate for loss of unsaturation in the cell membranes. Pinter et al. [15] presented supporting evidence from embryopathy: “the elevation in embryonic oleic acid level suggests that the teratogenic mechanism could be related to a deficiency in essential fatty acids. The pattern of essential fatty acid deficiency and embryopathy was preventable with arachidonic acid supplementation in this experimental model”. Protein quality is dependent on an optimum balance of essential amino acids. Our oleic acid data are suggesting a similar situation with essential fatty acids. As with essential amino acids, there will be an optimum profile for essential fatty acids. The profile is yet to be described. However, the red cell has a half-life of 120 days hence the data represent the period around the time of conception and possibly 3 to 4 months prior.

4.2. Biomagnification of the saturated fats

As far as we know, the enhancement of saturated fatty acids by the placenta for the fetus, as not been previously reported. At the same time, the proportions of oleic acid and other mono-unsaturated fatty acids were reduced in the fetus compared to the

mother.

Prenatally,

saturated

fatty

acids

were

being

swapped

for

monounsaturated. The most likely explanation of this data is that the saturated fatty acids are being transported to be esterified alongside the long chain super unsaturated fatty acids for cell membrane phosphoglycerides. Cell division and fetal growth requires membrane lipids. That requirement is particularly significant quantitatively for the fetal brain growth thrust of the last trimester. 15

Membrane phosphoglycerides generally have a polyenoic fatty acid in the SN-2 and a saturated fat in the SN-1 positions. Both AA and DHA appear to favour association with saturated fatty acid in the SN-1 position, especially stearic acid (18:0) [16]. Indeed, Hindenes et al. [17] have shown that activation of protein kinase C is achieved by S-stearoyl-2-arachidonoylglycerol (SAG) a derivative of the corresponding membrane phosphoglycerides. The negative concentration gradient for oleic acid and the MUFA is equally intriguing. Post-natally oleic acid is quantitatively the most significant fatty acid in mothers’ milk illustrating the striking shift from the prenatal requirement for superunsaturated fatty acids accompanied by saturated fatty acids for cell division to the post-natal MUFA required for energy and for myelination. This physiological switch may have relevance to feeding preterm infants. Preterm infant formulae based on milk composition result in a precipitous collapse of the profile of super-unsaturated fatty acid [13, 18]. It could be expected that the developmental physiology of a baby born preterm is anticipating the placental input, the biological strategy for which is a quite different from post-natal nutrition. It is however, common practice to design the lipids of preterm feeds based on term human milk composition which may not be ideal.

4.3. Arachidonic acid and DHA A role for DHA has been reported for neurogenesis [9] , dendrite formation [19], selective synaptic incorporation [20] and neural gene activation [21]. Indeed the long chain poly-unsaturated fatty acids (LC-PUFA) of both families are involved as ligands for nuclear receptors [22, 23]. Hence there is a mechanism for the involvement of LC-PUFA in cell structure and function including gene expression hence growth and development. The ALSPAC study of over 14,000 pregnancies found that verbal IQ and behavioural and social attitudes in 8-year-old children were enhanced the greater the maternal consumption of DHA foods (fish and sea foods) during the pregnancy [3]. This body of evidence is consistent with the positive relationships with DHA in cord blood. The data reported here including that on head circumference, supports the ALSPAC outcome and adds the relationship between maternal diet even before pregnancy to birth weight, 16

gestational age and head circumference as markers for health of the new born. One can assume that diet during the pregnancy as reported in ALSPAC, is likely to reflect longterm nutritional trends in contrast to the expectations of supplements. Our data does not say that interventions or events during pregnancy would be irrelevant. However, interventions have had limited effects. Generally fish oils are used in the hope of advancing intelligence and other aspects of visual and brain function. However, fish oils and allied supplements are generally long chain ω3 triglycerides whereas cell membranes use ω6 and ω3 phosphoglycerides. During pregnancy, the hormonal shift results in triglyceride deposition to accumulate fat stores. This mechanism may accumulate as much as 4.5 kg of adipose fat. It is thought to be in advance of lactation when the energy demand is much greater than for the pregnancy guaranteeing one third of the energy needed for the first 10 days [14]. Indeed the strong predictive value of oleic acid and other fatty acid measurements at recruitment emphasises that point. Nature prepares in advance. In view of the long half-life of the red cell, it is likely that the condition of the mother in the several months prior to conception is the important determinant of outcome and is a more forceful decider than what happens during the pregnancy, where in most cases the diet of the mother will have a similar content to the habitual diet.

5. Conclusions

5.1 The conclusion from this study is that the condition of the mother in the months around conception is probably a principle determinant of gestational length regardless of whether or not the mother changed diet or took a supplement during the pregnancy (see also [24].)

5.2 The biomagnification of saturated and reverse for mono-unsaturated fatty acids seen at delivery opens a new chapter in our understanding of lipidomics and the physiology of early development. The evidence of this data similar to that of AA and DHA is that the reproductive process is deliberately prioritising different fatty acids for different periods and different functions: e.g. AA for cardio-vascular (13) the immune 17

system (28) and placental development (29), and DHA ultimately for brain growth. However, both AA and DHA need a saturated fatty acid to accompany them in the SN1 position of the growing, membrane phosphoglycerides. The data is telling us that saturated fatty acids have as yet unheralded and important role in cell membrane growth and integrity.

5.3 Additionally, the oleic acid - MUFA signal when predicting preterm delivery and low birthweight is a surrogate for a global inadequacy to meet the optimum conditions for membrane development and hence cell division, proliferation, growth and membrane integrity. Just as protein quality is dependent on the balance of essential amino acids the implication of our data is of a profile of fatty acids being specifically selected for differential requirements for the product of conception. The data is presenting a new biological paradigm on the significance of the lipid profile playing such a complex role in reproduction. Whilst accepting that this new chapter requires much work to fully understand the processes involved it is interesting that Susan Carlson and colleagues reported data on preterm infants in which the biomagnification of saturated and reverse of oleic acid can be clearly seen [25].

6. Implications: The brain is largely made of special fats, which individually occur, in different aspects of the food web. Preterm delivery and low birthweight are known to carry a risk for neurodevelopmental disorder. Very preterm and low birthweight infants are at a greatly increased risk to severe neurodevelopmental disorders such as cerebral palsy. There is good evidence to support the Barker Hypothesis [26] which claims poor maternal/fetal nutrition is responsible for diabetes, heart disease and stroke in later life. Such a condition, although perhaps of a nature more focussed on the brain specific nutrients, is unlikely to leave the developing brain unscathed. The evidence from the ALSPAC study and that reported here questions the changing diet over the last few decades, which has been based more on protein rather than brain specific nutrients. The rise in mental ill-health [27], especially amongst the young is a matter of deep concern. 18

Acknowledgements The authors thank the Mother and Child Foundation, Letten Foundation, the Waterloo Foundation and Vifor Pharma for financial support and encouragement. We are especially grateful to the staff at the NHS Obstetrics and Gynaecology Unit at the Chelsea and Westminster Hospital as well as the mothers for their cooperation in this study. We also wish to thank Paul Seed for undertaking some of the statistical analysis and AnnieBelle Sassine for proof reading and adjusting the references.

7. References [1] FAO, Fats and fatty acids in human nutrition. Report of an expert consultation, FAO Food Nutr Pap, 91 (2010) 1-166.

[2] P.E. Wainwright, Essential fatty acids and behavior, in: Handbook of Essential Fatty Acid Biology, Springer, 1997, pp. 299-341.

[3] J.R. Hibbeln, J.M. Davis, C. Steer, P. Emmett, I. Rogers, C. Williams, J. Golding, Maternal seafood consumption in pregnancy and neurodevelopmental outcomes in childhood (ALSPAC study): an observational cohort study, Lancet, 369 (2007) 578-585.

[4] L.G. Smithers, R.A. Gibson, M. Makrides, Maternal supplementation with docosahexaenoic acid during pregnancy does not affect early visual development in the infant: a randomized controlled trial, Am J Clin Nutr, 93 (2011) 1293-1299.

[5] S.F. Olsen, Is supplementation with marine omega-3 fatty acids during pregnancy a useful tool in the prevention of preterm birth?, Clinical obstetrics and gynecology, 47 (2004) 768-774; discussion 881-762.

[6] G. Saccone, V. Berghella, Omega-3 long chain polyunsaturated fatty acids to prevent preterm birth: a systematic review and meta-analysis, Obstetrics and gynecology, 125 (2015) 663-672.

19

[7] M. Crawford, A. Sinclair, Nutritional influences in the evolution of mammalian brain, in:

Ciba

Foundation Symposium 3-Lipids, Malnutrition & the Developing Brain, Wiley Online Library, 1972, pp. 267-292.

[8] M.A. Crawford, A.G. Hassam, G. Williams, Essential fatty acids and fetal brain growth, Lancet, 1 (1976) 452-453.

[9] E. Yavin, E. Himovichi, R. Eilam, Delayed cell migration in the developing rat brain following maternal omega 3 alpha linolenic acid dietary deficiency, Neuroscience, 162 (2009) 1011-1022. [10] J.T. Brenna, N. Salem, A.J. Sinclair, S.C. Cunnane, α-Linolenic acid supplementation and conversion to ω3 long-chain polyunsaturated fatty acids in humans, Prostaglandins, leukotrienes and essential fatty acids, 80 (2009) 85-91.

[11] J. Folch, M. Lees, G.H. Sloane Stanley, A simple method for the isolation and purification of total lipides from animal tissues, J Biol Chem, 226 (1957) 497-509. [12] S. Wynn, A. Wynn, W. Doyle, M. Crawford, The association of maternal social class with maternal diet and the dimensions of babies in a population of London women, Nutrition and Health, 9 (1994) 303315.

[13] WS Harris (2009) The omega-3 index: from biomarker to risk marker to risk factor.Curr Atheroscler Rep. 2009 Nov;11(6):411-7.

[14] FAO/WHO, Dietary fats and oils in human nutrition. Report of an Expert Consultation (21-30 September 1977), in, FAO, Rome, 1978.

[15] E. Pinter, E.A. Reece, P.L. Ogburn, S. Turner, J.C. Hobbins, M.J. Mahoney, F. Naftolin, Fatty acid content of yolk sac and embryo in hyperglycemia-induced embryopathy and effect of arachidonic acid supplementation, American journal of obstetrics and gynecology, 159 (1988) 1484-1490·

[16] A. Brand, M.A. Crawford, E. Yavin, Retailoring docosahexaenoic acid-containing phospholipid species during impaired neurogenesis following omega-3 alpha-linolenic acid deprivation, J Neurochem, 114 (2010) 1393-1404.

[17] J.O. Hindenes, W. Nerdal, W. Guo, L. Di, D.M. Small, H. Holmsen, Physical properties of the transmembrane signal molecule, sn-1-stearoyl 2-arachidonoylglycerol. Acyl chain segregation and its biochemical implications, J Biol Chem, 275 (2000) 6857-6867.

20

[18] M. Crawford, I. Golfetto, K. Ghebremeskel, Y. Min, T. Moodley, L. Poston, A. Phylactos, S. Cunnane, W. Schmidt, The potential role for arachidonic and docosahexaenoic acids in protection against some central nervous system injuries in preterm infants, Lipids, 38 (2003) 303-315.

[19] H.-Y. Kim, A.A. Spector, Synaptamide, endocannabinoid-like derivative of docosahexaenoic acid with cannabinoid-independent function, Prostaglandins, Leukotrienes and Essential Fatty Acids (PLEFA), 88 (2013) 121-125.

[20] H. Suzuki, S. Manabe, O. Wada, M. Crawford, Rapid incorporation of docosahexaenoic acid from dietary sources into brain microsomal, synaptosomal and mitochondrial membranes in adult mice, International journal for vitamin and nutrition research. Internationale Zeitschrift fur Vitamin-und Ernahrungsforschung. Journal international de vitaminologie et de nutrition, 67 (1996) 272-278.

[21] K. Kitajka, A.J. Sinclair, R.S. Weisinger, H.S. Weisinger, M. Mathai, A.P. Jayasooriya, J.E. Halver, L.G. Puskás, Effects of dietary omega-3 polyunsaturated fatty acids on brain gene expression, Proceedings of the National Academy of Sciences of the United States of America, 101 (2004) 1093110936.

[22] A. Chawla, J.J. Repa, R.M. Evans, D.J. Mangelsdorf, Nuclear receptors and lipid physiology: opening the X-files, Science, 294 (2001) 1866-1870·

[23] A.M. de Urquiza, S. Liu, M. Sjoberg, R.H. Zetterstrom, W. Griffiths, J. Sjovall, T. Perlmann, Docosahexaenoic acid, a ligand for the retinoid X receptor in mouse brain, Science, 290 (2000) 21402144.

[24] P. Dominguez-Salas, S.E. Moore, M.S. Baker, A.W. Bergen, S.E. Cox, R.A. Dyer, A.J. Fulford, Y. Guan, E. Laritsky, M.J. Silver, G.E. Swan, S.H. Zeisel, S.M. Innis, R.A. Waterland, A.M. Prentice, B.J. Hennig, Maternal nutrition at conception modulates DNA methylation of human metastable epialleles, Nat Commun, 5 (2014).

[25] S. Scholtz, E. Kerling, D. Shaddy, S. Li, J. Thodosoff, J. Colombo, S. Carlson, Docosahexaenoic acid (DHA) supplementation in pregnancy differentially modulates arachidonic acid and DHA status across FADS genotypes in pregnancy, Prostaglandins, Leukotrienes and Essential Fatty Acids (PLEFA), 94 (2015) 29-33.

[26] D.J. Barker, The origins of the developmental origins theory, J Intern Med, 261 (2007) 412-417.

21

[27] P. Andlin-Sobocki, B. Jonsson, H.U. Wittchen, J. Olesen, Cost of disorders of the brain in Europe, Eur J Neurol, 12 Suppl 1 (2005) 1-27.

[28] T. Moodley, C. Vella, O. Djahanbakhch, C.J. Branford-White, M.A. Crawford, Arachidonic and docosahexaenoic acid deficits in preterm neonatal mononuclear cell membranes. Implications for the immune response at birth, Nutr Health, 20 (2009) 167-185.

[29] D. Bitsanis, M.A. Crawford, T. Moodley, H. Holmsen, K. Ghebremeskel, O. Djahanbakhch, Arachidonic acid predominates in the membrane phosphoglycerides of the early and term human placenta, J Nutr, 135 (2005) 2566-2571.

Figure Legends: Figure 1. . ROC for oleic acid at recruitment in the whole population in relation to gestational age (n=296). This data includes those with or without supplement during the pregnancy. 50 with gestational diabetes and 50 with hypertension. It is thus predicting preterm delivery at 34 weeks regardless of events during the pregnancy. Figure 2. Consistent biomagnification of saturated fatty acids across the placenta. The likelihood is that the saturated fatty acids are required in the SN1 position of the phosphoglycerides alongside the long chain PUFA as well as in the sphingolipids and ceramides being incorporated into the growth cell membranes. ** p<0.01, .*** p<0.001 Figure 3. Stearic (18:0) versus oleic acid (18:1 ω9) in maternal blood at delivery and fetal cord blood. The data illustrates placental biomagnification for stearic but the opposite for oleic acid the proportion of which is reduced in the fetal circulation (n=189). .*** p<0.001 Figure 4. Illustrates the significant, negative “r” values for oleic acid and all monounsaturated fatty acids in maternal red cells at recruitment with eventual birthweight (BWT) and gestational age (GEST) in control mothers, for the group at risk to low birthweight (LBW), and for head circumference for births <3,200g our arbitrary cut off 22

point below which risk of low birthweight was found to increase with increasing nutritional inadequacy. Above 3,270 g there was no relationship could be found between birthweight and maternal nutrition in a study following some 43 nutrients (12). For birthweight and gestational age, the total MUFA are significantly stronger than oleic acid on its own (p<0.001). For head circumference only the oleic acid – MUFA was significant (p<0.01)

Table 1: Demography of the population studied N 281 281 280

Maternal height (cm) Maternal weight (kg) Body Mass Index (BMI) Gestation age at delivery (weeks) Head circumference (cm) Infant length (cm) Birth weight (BW)(g) BW<3200g BW>3200g LBW <2500g Apgar at 1 Apgar at 5 Gender, male (%)

Mean± 164·81± 73·71± 27·20±

SD 6·45 18·48 6·44

259

37·97± 2·84

231 208 260 134 128 43 256 261 259

34·11± 1·74 50·88± 3·87 3119·54± 723·21 2606·72± 603·61 3621·72± 480·66 1903·16± 583·20 8·43± 1·61 9·48± 1·59 51%

Maximum 180·00 144·00 45·00

Minimum 142·00 40·00 16·00

42·00

22·00

40·50 58·50 4740·00 3200·00 4740·00 2490·00 10·00 10·00

27·50 37·00 470·00 470·00 900·00 470·00 1·00 0·00

Table 2: Fatty acid composition (%) of red blood cells in whole unsupplemented group at different time points Recruit (R)

14:0 16:0 18:0 20:0

n=139

Delivery (D)

n=82

Cord(C)

n=58

Mean

± SD

Mean

± SD

Mean

± SD

0·74 24·3 11·5 0·34

± 0·25 ± 1·21 ± 1·26 ± 0·07

0·73 25·9 11·7 0·34

± 0·23 ± 2·59 ± 1·29 ± 0·07

0·65 28·40 14·9 0·49

± 0·16^ ### ± 3·20^^^ ### ± 1·82^^^ ### ± 0·11^^^

23

22:0 24:0

1·03 2·21

± 0·24 ± 0·54

1·06 2·43

± 0·26 ± 0·65

1·17 3·17

± 0·27^^^ ### ± 0·68^^^

16:17 18:19 18:17 18:1 (all) 20:19 22:19 24:19

0·72 14·06 1·16 15·2 0·26 0·06 2·88

± 0·36 ± 2·03 ± 0·17 ± 2·11 ± 0·05 ± 0·03 ± 0·78

0·82 15·1 1·14 16·3 0·28 0·05 3·07

± 0·42 ± 1·87 ± 0·19 ± 1·98 ± 0·05 ± 0·03 ± 0·64

0·77 10·05 1·71 11·75 0·22 0·04 2·62

± 0·21 ### ± 1·34^^^ ### ± 0·33^^^ ### ± 1·46^^^ ± 0·30 ± 0·04 ### ± 0·55^

18:26 18:36 20:26 20:36 20:46 22:26 22:46 22:56

13·4 0·05 0·04 1·59 12·4 0·15 2·00 0·30

± 2·07 ± 0·03 ± 0·02 ± 0·30 ± 1·69 ± 0·06 ± 0·45 ± 0·12

12·6 0·06 0·04 1·47 10·8 0·12 1·85 0·41

± 1·58 ± 0·03 ± 0·04 ± 0·28 ± 2·15 ± 0·07 ± 0·51 ± 0·16

4·42 0·05 0·06 2·37 15·2 0·06 2·59 0·86

± 1·62^^^ ± 0·02 ± 0·05 ### ± 0·62^^^ ### ± 3·38^^^ ### ± 0·04^^^ ### ± 0·65^^^ ### ± 0·36^^^

18:33 20:33 20:53 22:53 22:63

0·37 0·03 0·68 1.68 4.72

± 0·19 ± 0·02 ± 0·40 ± 0·32 ± 1.10

0·26 0·04 0·49 1.45 4.17

± 0·14 ± 0·02 ± 0·28 ± 0·44 ± 1.18

0·09 0·04 0·27 0·60 4.89

± 0·16^^^ ± 0·02^^^ ### ± 0·15^^^ ### ± 0·19^^^ ## ± 1.41

20:39

0·22

± 0·06

0·21

± 0·07

0·14

± 0·05^^^

42·3 20·50 27·4 14·74 6·42 6·16 2·76 0·22 4·66 2·55 4·66

± 3·40 ± 2·32 ± 3·35 ± 2·76 ± 1·76 ± 1·70 ± 0·74 ± 0·06 ± 1·65 ± 0·71 ± 1·35

49.2 15·4 25·5 21·0 5·89 5·80 3·22 0·32 4·51 3·74 5·16

± 4.99^^^ ### ± 1·62^^^ # ± 4·54^^^ ### ± 4·54^^^ ± 1·51^^^ ± 1·53^^^ ### ± 0·73^^^ ### ± 0·10^^^ ± 0·99 ### ± 0·81^^^ ± 1·44

Saturates 40·3 ± 1·89 Mufa 19·1 ± 2·25 29·9 ± 2·56 6 16·5 ± 2·05 6LC 7·48 ± 1·62 3 7·11 ± 1·59 3LC AA/DHA 2·76 ± 0·75 0·15 ± 0·06 22:56/22:46 4·19 ± 0·99 6/3 2·43 ± 0·60 6LC/3LC Omega3 Index 5·40 ± 1·39 R vs D: *** P<0·005; ** P<0·01; * P<0·05 R vs C: ^^^ P<0·005; ^^ P<0·01; ^ P<0·05 D vs C: ### P<0·005; # # P<0·01; # P<0·05

###

###

###

###

Table 3. Significant Pearson coefficients for control women excluding any with diabetes or PET at 3 times in the study who were not known to have taken any fatty acid supplement Birthweight (g)

BW <3200g

Birth Gestation age

24

AT RECRUITMENT c18:1 ω 9ω c18:1 18:2 ω 6 c20:1 c20:4 ω 6 c20:4 ω 6/18:1 c20:5 ω 3 c24:1 c22:6 ω 3 c22:6 ω 3/18:1 AA+DHA (AA+DHA)/Mono (AA+DHA)/Oleic Sat FAs MonouFAs ω6 ω6LC ω3 ω3LC ω6/Mufa ω6LC/Mufa ω3/Mufa ω3LC/Mufa AA/DHA ω6/ω3 ω6LC/ω3LC Omega3 Index ω6LC+ ω3LC ω6LC+ωLC)/Mufa ω6+ω3 ω6+ω3)/Mufa S/P

n=180 -0·275 -0·289 0·130 -0·215 0·134 0·223 0·191 -0·252 0·335 0·347 0·286 0·340 0·304 -0·100 -0·362 0·175 0·095 0·291 0·300 0·302 0·251 0·336 0·340 -0·249 -0·218 -0·244 0·314 0·277 0·337 0·322 0·353 -0·238

AT DELIVERY

n=102

c20:4n6 ω6LC

-0·199 -0·211

CORD cω22:6n3 ω3

n=82 0·141 0·142

P 0·000 0·000 0·081 0·004 0·073 0·003 0·010 0·001 0·000 0·000 0·000 0·000 0·000 NS 0·000 0·019 0·203 0·000 0·000 0·000 0·001 0·000 0·000 0·001 0·003 0·001 0·000 0·000 0·000 0·000 0·000 0·001

n=93 -0·468 -0·463 0·229 -0·279 0·388 0·483 0·228 -0·173 0·296 0·377 0·467 0·530 0·502 -0·175 -0·528 0·454 0·339 0·282 0·290 0·540 0·498 0·383 0·383 -0·091 -0·128 -0·130 0·295 0·452 0·523 0·543 0·551 -0·457

P 0·000 0·000 0·027 0·007 0·000 0·000 0·028 NS 0·004 0·000 0·000 0·000 0·000 NS 0·000 0·000 0·001 0·006 0·005 0·000 0·000 0·000 0·000 NS NS NS 0·004 0·000 0·000 0·000 0·000 0·000

n=51 0·045 0·033

-0·306 -0·313

0·381 0·358

P 0·000 0·000 0·026 0·007 0·000 0·000 0·001 0·004 0·000 0·000 0·000 0·000 0·000 0·024 0·000 0·000 0·006 0·000 0·000 0·000 0·000 0·000 0·000 0·010 0·003 0·001 0·000 0·000 0·000 0·000 0·000 0·000

n=104 0·029 0·025

n=38 NS NS

n=182 -0·395 -0·395 0·165 -0·198 0·272 0·373 0·237 0·213 0·308 0·369 0·383 0·441 0·425 -0·168 -0·45 0·288 0·204 0·298 0·307 0·411 0·366 0·369 0·371 -0·19 -0·22 -0·23 0·307 0·365 0·431 0·420 0·449 -0·34

-0·179 -0·213

0·069 0·030

n=83 0·018 0·027

0·18:1 0·180

NS NS

25

ω3LC AA/DHA ω6/ω3 ω6LC/ω3LC Omega3 Index

0·156 -0·360 -0·265 -0·338 0·143

NS 0·001 0·016 0·002 NS

0·383 -0·551 -0·451 -0·546 0·389

0·017 0·000 0·005 0·000 0·016

0·191 -0·40 -0·279 -0·37 0·179

NS 0·000 0·011 0·001 NS

26

Table 4: Pearson’s correlation coefficients for head circumference in different groups. (Negative correlations highlighted in bold and positive correlations in italics) 1n=225 2n=130 3n=106

All

1n=112 2n=66 3n=51

P

HC-BW <3200g

P

-0·199 -0·201 0·099 0·147 0·125 -0·200 0·133 0·138

0·003 0·003 NS 0·028 NS 0·003 0·046 0·039

0·309 -0·302 0·210 0·271 0·220 -0·299 0·246 0·253

0·001 0·001 0·026 0·004 0·020 0·001 0·009 0·007

0·126

NS

0·231 0·014

0·234 NS

-0·111 -0·072

.208 .415

-0·251 0·042 -0·193 NS

0·243 0·012 0·235 0·015 0·192 0·048 0·144 NS -0·088 NS -0·020 .843 -0·073 .456 -0·030 NS

0·073 NS 0·091 NS 0·002 NS 0·331 0·018 -0·013 NS -0·001 NS -0·002 NS -0·040 NS

All HC (cm) Recruitment c18:1ω9 c18:1 c20:5ω3 c22:5ω3 22:6ω3 MUFA ω3 ω3LC Omega3 Index Delivery c24:1 AA/DHA Cord blood c18:1ω9 c18:1 c18:2ω6 20:3ω9 c20:4ω6 ω6 ω6LC S/P

1n=67 1n=42 LBW LBW 2n=39 2n=25 Risk Risk 3n=32 3n=18 HC HC-BW P P (cm) <3200g -0·185 -0·187 0·121 0·089 0·254 -0·253 0·193 0·214

.NS NS NS .NS 0·038 0·039 NS NS

-0·141 -0·141 0·073 0·025 0·241 -0·164 0·172 0·181

NS NS NS NS NS NS NS NS

1n=33 1n=13 NHC 2n=21 NHC 2n=6 3n=20 3n=6 HC HC-BW P P (cm) <3200g -0·099 -0·119 0·015 0·128 0·156 -0·082 0·134 0·125

NS NS NS NS NS NS NS NS

-0·465 NS -0·459 NS 0·600 0·030 0·760 0·003 0·687 0·009 -0·372 NS 0·732 0·004 0·719 0·006

0·114 NS

0·672 0·012

-0·299 NS -0·209 NS

-0·474 0·017 0·081 NS -0·451 0·024 -0·107 NS

-0·920 0·009 -0·619 NS

0·363 0·377 0·262 0·408 -0·227 -0·171 -0·222 0·195

0·230 NS 0·168 NS 0·261 NS 0·143 NS 0·067 NS 0·133 NS 0·642 0·004 0·028 NS -0·492 0·027 -0·193NS -0·161NS -0·394 NS -0·189 NS -0·456 NS 0·475 0·034 0·184 NS

-0·720 NS -0·734 NS -0·847 0·033 0·038 NS -0·929 0·007 -0·954 0·003 -0·925 0·008 0·899 0·015

0·211 NS

0·041 0·034 NS 0·020 NS NS NS NS

Table 5a: Receiver Operator Curve area for ‘control’ mothers at recruitment; n= 136 Asymptotic Numbers (yes/no)

Area

Std. Error

a

Asymptotic b

Sig.

95%

Confidence Interval Lower

Upper

Bound

Bound

37 wks oleic acid

25/109

0·677

0·064

0·006

0·551

0·804

37 wks Monoene

25/109

0·669

0·067

0·009

0·538

0·800

34 wks oleic acid

7/129

0·869

0·048

0·001

0·775

0·964

34 wks Monoene

7/129

0·886

0·059

0·001

0·770

1.000

30 wks oleic acid

4/132

0·852

0·072

0·017

0·711

0·994

30 wks Monoene

4/132

0·830

0·093

0·025

0·647

1.000

27

2500g oleic acid

18/117

0·742

0·057

0·001

0·631

0·853

2500g Monoene

18/117

0·731

0·062

0·002

0·610

0·853

Table 5b: Receiver Operator Curve area for all mothers at recruitment; n= 280

37 wks oleic acid 37 wks Monoene

50/224 50/224

0·668 0·682

0·044 0·044

0·000 0·000

Asymptotic 95% Confidence Interval Lower Upper Bound Bound 0·581 0·755 0·596 0·768

34 wks oleic acid

17/263

0·802

0·066

0·000

0·673

0·932

34 wks Monoene 30 wks oleic acid 30 wks Monoene 2500g oleic acid 2500g Monoene

17/263 7/273 7/273 39/240 39/240

0·811 0·760 0·763 0·724 0·738

0·068 0·113 0·111 0·046 0·046

0·000 0·019 0·018 0·000 0·000

0·678 0·538 0·546 0·634 0·649

0·945 0·982 0·980 0·814 0·827

Numbers (yes/no)

Area

Std. a Error

Asymptotic b Sig.

Table 5c: Receiver Operator Curve area for supplemented mothers; n= 144 Numbers (yes/no) 37 wks oleic acid 37 wks Monoene 34 wks oleic acid 34 wks Monoene 30 wks oleic acid 30 wks Monoene 2500g oleic acid 2500g Monoene

25/115 25/115 10/134 10/134 3/141 3/141 21/123 21/123

Area

Std0· Error

0·657 0·696 0·756 0·766 0·645 0·697 0·712 0·746

0·062 0·058 0·104 0·104 0·228 0·214 0·069 0·065

a

Asymptotic b Sig0· 0·014 0·002 0·007 0·005 0·390 0·243 0·002 0·000

Asymptotic 95% Confidence Interval Lower Upper Bound Bound 0·536 0·778 0·582 0·810 0·553 0·959 0·563 0·970 0·199 1·000 0·278 1·000 0·576 0·848 0·619 0·874

Figure 1: Oleic acid ROC at recruitment for the whole population before randomisation n=7/296 ROC at 34 weeks = 0·926 28

Number yes/no 9/287, (p<0·000), term = 0·729 (p<0·000)

Area under ROC curve = 0.7293

1.00

1.00

0.75

0.75

Sensitivity

Sensitivity

Area under ROC curve = 0.9258

0.50

0.25

0.50

0.25

0.00

0.00 0.00

0.25

0.50 1 - Specificity

0.75

1.00

0.00

0.25

0.50 1 - Specificity

0.75

1.00

Figure 2: Biomagnification of saturated fatty acids. (Data from normal healthy controls only n=50)

29

Figure 3: Consistent biomagnification of saturated fatty acids

40

[% 20:0, 22:0 & 24:0] x 10

% 16:0 & 18:0 in RBC membranes

***

35

**

30

3.0

25

2.5 ***

20

2.0 ***

15

1.5

***

10

1.0 0.5

5 0 16:0

18:0

20:0

Recruitment

22:0

Delivery

24:0

Fetus

Figure 3:

SUMMARY - % 18:0 vs 18:1ω9 in RBC membranes

18 16 14 12 10 8 6 4 2 0

***

***

Delivery Fetus

18:0

18:1

Data from un-supplemented, normal healthy controls only n=50) 30

Figure 4: Negative Pearson correlations at recruitment for birthweight (BWT) gestational age (GEST) and head circumference (HEAD CIRC) in

Figure 5 Oleic acid and all mono-unsaturated correlations with -ve “r” values at recruitment with

only un-supplemented with red and monobirthweight and gestational age inmothers all control mothers, for thecell groupoleic at risk acid to low birthweight, and unsaturates. P<0.01 in <3.200g all tests for head circumference for births

BWT N= 112

GEST N= 126

HEAD CIRC N=67

0 -0.1 -0.2

ALL OLEIC -0.3 -0.4

ALL MUFA LBW RISK OLEIC LBW RISK MUFA

-0.5 -0.6 -0.7 With the exception of head circumference in the <3,200g group the sum of all monounsaturates was greater than oleic acid independently. [Data from mothers in control groups excluding diabetes and PET.

31