Both intrauterine and early infancy are periods of rapid growth and development during which insufficient supply of energy and nutrients might result to metabolic or body composition alterations. Its relative impact on the different periods has not yet been elucidated(Reference Gluckman, Hanson, Cooper and Thornburg1) but it appears to be modulating early life outcomes and later risk of chronic disease(Reference Gluckman, Hanson, Cooper and Thornburg1, Reference McMillen and Robinson2). Specifically, early life nutrition has been shown to significantly contribute to adiposity development variability(Reference McMillen and Robinson2–Reference Labayen, Ruiz, Vicente-Rodriguez, Turck, Rodriguez and Meirhaeghe4). The fetal-infant programming hypothesis states that increased risk of adiposity later in life is originated from early exposure to detrimental environments including nutritional aspects; however, the mechanisms are still unclear(Reference Gluckman, Hanson, Cooper and Thornburg1, Reference McMillen and Robinson2). Therefore, effective preventive measures of the obesity epidemic require knowledge of the dietary risk factors and their consequences which act during critical life periods(Reference Moreno and Rodriguez5).
The effects of essential long chain polyunsaturated fatty acids (LCPUFA) supplementation during the perinatal period on neurobehavioral development or visual acuity, infant growth as well as safety monitoring outcomes has been addressed by a number of clinical trials mainly in preterm infant populations(Reference Gaillard, Negrel, Lagarde and Ailhaud6, Reference Simmer, Patole and Rao7). The majority of the studies tested the effect of specific LCPUFA concentrations added to an infant formula on postnatal growth outcomes including body weight and length(Reference Schulzke, Patole and Simmer8, Reference Makrides, Gibson, Udell and Ried9). Some of the limited reporting effects on body composition and long-term body fat programming, both from animal and human studies, indicate that early availability of LCPUFA might influence development of adipose tissue during fetal life and infancy(Reference Vollhardt10). Eicosanoids derived from arachidonic acid (AA), n-6 LCPUFA, appear to have an adipogenic effect, by providing a molecular link between fatty acid uptake and preadipocytes differentiation during early hyperplasic growth stages of adipose tissue. In contrast, those derived from eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA), n-3 LCPUFA, have an antiadipogenic effect(Reference Amri, Ailhaud and Grimaldi11, Reference Makrides, Collins and Gibson12). Consequently, fatty acid levels and the ratio between n-6 LCPUFA and n-3 LCPUFA in the maternal diet, during pregnancy and lactation, may play an important role in early adipose tissue development(Reference Ailhaud and Guesnet13).
The present systematic review presents studies addressing the relationship between n-3 LCPUFA intakes, during pregnancy and postnatally, on early and long-term body composition variability.
Methods
The research question to be answered by the systematic review was if early life n-3 LCPUFA intake (prenatal and early postnatal periods) has any influence on childhood body composition. The flow chart of the process is illustrated in Fig. 1. The search process was not limited to any language, timeframe or country of publication and was performed in three electronic databases (MEDLINE, EMBASE and LILACS). The general search strategy included terms related to the population under question (infants and children), predictor (n-3 LCPUFA intake) and dependent variables (obesity and body composition). The shared terms used in MEDLINE and EMBASE search included (pregnant women [MeSH] OR breastfeeding [MeSH] OR age group [MeSH]) AND (Fatty Acids, Unsaturated [MeSH]) AND (body weight [MesH] OR metabolic syndrome X [MesH]). In LILACS, terms used were slightly different: (Risk groups Nutrition [MeSH] OR feeding behaviour [MeSH] OR age groups [MeSH]) AND (Fatty Acids, Unsaturated [MeSH]) AND (body weight [MesH] OR Nutritional and Metabolic Diseases [MesH]).
Fig. 1 Stages of the systematic review process
The initial search yielded 2605 references after exclusion of duplicates. Additional publications were identified from references listed in the identified original papers reviewed. This secondary search added 47 potential relevant papers (total of 2653). Selected studies were then classified into two different groups: 1) maternal n-3 LCPUFA intake during pregnancy and lactation on infants and young children's body composition; and 2) effects of infant's n-3 LCPUFA intake (from birth or early postnatal period) on later body composition.
The results of the searches were stored in an Endnote XII library. Firstly, references were screened on the basis of title and abstract. Those clearly not meeting the review's criteria were excluded. Selected references in the previous step were all screened based on full text. Reasons for exclusion were registered in the Endnote library. Criteria for inclusion/exclusion are stated in Table 1. References were excluded on the basis of irrelevant health outcome (no adiposity or body composition measurements) and targeted population or dietary exposure (only ALA as n-3 LCPUFA or only n-6 essential fatty acids as LCPUFA supplementations). When in doubt, the team reviewed the papers to ensure alignment and quality control. Only papers fulfilling the inclusion criteria were considered in the present review.
Table 1 Inclusion and exclusion criteria.
Assessment of risk of bias in included studies
The quality and risk of bias were assessed as indicators of validity for identified observational and intervention studies. The studies included in this review were checked for a minimum quality score system developed by EURopean micronutrient RECommendation Aligned (EURRECA -network of excellence-) which was adapted from The Cochrane Handbook(Reference Higgins and Green14). Criteria for intervention studies were based on method of sequence generation (adequate randomization procedure) and allocation concealment, blinding, potential funding bias, number of participants at start, number of dropouts and suggested reasons, dose check, dietary intake data reported, and similarities between the most and least exposed groups at baseline. For longitudinal studies criteria were based on number of dropouts and reasons, potential funding bias, lack of other potential threats to validity, inclusion of confounders, and assessment of adequacy of exposure. Concepts were evaluated as of high, low or of uncertain risk of bias. Overall risk of bias was judged as high if more than one of the following concepts were uncertain or inadequately addressed: confounders, exposure assessment, potential funding bias for observational studies and sequence generation, allocation concealment, blinding or potential funding bias for intervention studies. Observational and intervention studies were judged of moderate risk of bias, if reviewed studies had one of the above stated criteria judged as high risk. If there was no risk of bias or the risk of bias was present only in other criteria different from those mentioned above, the overall risk of bias was judged as low.
Results
A total of 13 publications(Reference Asserhoj, Nehammer, Matthiessen, Michaelsen and Lauritzen15–Reference Hauner, Much, Vollhardt, Brunner, Schmid and Sedlmeier27) were eventually selected for inclusion in this systematic review. All studies are summarised in Tables 2 and 3 together with extracted information on country where the study was performed, number of participants and age at enrolment, intervention and follow-up duration, intervention details and diet/formula LCPUFA composition, body composition-related outcomes and conclusions. Studies were split into two categories; those reporting data on maternal intake of n-3 LCPUFA during pregnancy or lactation (Table 2)(Reference Asserhoj, Nehammer, Matthiessen, Michaelsen and Lauritzen15, Reference Donahue, Rifas-Shiman, Gold, Jouni, Gillman and Oken18, Reference Helland, Smith, Blomen, Saarem, Saugstad and Drevon20, Reference Lauritzen, Hoppe, Straarup and Michaelsen23, Reference Lucia Bergmann, Bergmann, Haschke-Becher, Richter, Dudenhausen and Barclay24, Reference Hauner, Much, Vollhardt, Brunner, Schmid and Sedlmeier27) and those assessing infant n-3 LCPUFA intake (beginning during neonatal period) and potential effects on later body composition (Table 3)(Reference Birch, Hoffman, Uauy, Birch and Prestidge16, Reference de Jong, Boehm, Kikkert and Hadders-Algra17, Reference Groh-Wargo, Jacobs, Auestad, O'Connor, Moore and Lerner19, Reference Innis, Adamkin, Hall, Kalhan, Lair and Lim21, Reference Kennedy, Ross, Isaacs, Weaver, Singhal and Lucas22, Reference Ryan, Montalto, Groh-Wargo, Mimouni, Sentipal-Walerius and Doyle25, Reference Scholtens, Wijga, Smit, Brunekreef, de Jongste and Gerritsen26).
Table 2 Included studies on maternal n-3 polyunsaturated fatty acid intake during pregnancy and lactation and possible effects on infant body composition.
ALA, alpha-linolenic acid; DHA, docosahexaenoic acid; EPA, eicosapentaenoic acid; AA, araquidonic acid; RCT, randomized controlled trial; LCPUFA, polyunsatured fatty acid; BMI, body mass index.
Table 3 Included studies on infant's n-3 polyunsaturated fatty acid intake and possible effects on body composition
NA, Not applicable; RCT, randomized controlled trial; LA, linoleic acid; ALA, alpha-linolenic acid; LCPUFA, polyunsatured fatty acids; DHA, docosahexaenoic acid; AA, araquidonic acid; PMA, postmenstrual age; EPA, eicosapentaenoic acid; FFM, fat free mass, TBF, total body fat; TOBEC, total body electrical conductivity; DEXA, X-ray absorptiometry; GLA, gamma-linolenic acid; BMI, body mass index
Effects of maternal n-3 LCPUFA intake during pregnancy and lactation on body composition during infancy and childhood
A total of six studies assessed maternal intakes of n-3 LCPUFA, one being an observational and five intervention studies (Table 2).
Only one study(Reference Donahue, Rifas-Shiman, Gold, Jouni, Gillman and Oken18) addressed longitudinally the relationship between prenatal n-3 fatty acid intake and long term adiposity at 3 years of age. This recently US published study was carried out in a cohort of pregnant women where n-3 LCPUFA maternal intake was measured at 29 weeks of pregnancy (mean) and a month before delivery, using a previously validated food frequency questionnaire. LCPUFA intake was reported as total n-3 fatty acids, alpha-linolenic acid (ALA) and DHA + EPA. Blood samples were obtained at mid-pregnancy and after delivery, from the umbilical cord, for quantification of fatty acids from erythrocyte membranes. Children's body composition parameters (height, weight, skinfold thickness) and dietary intakes were measured at the age of 3. The authors concluded that higher maternal prenatal n-3 intake was associated with lower adiposity in early childhood. They observed that a higher DHA + EPA intake during mid-pregnancy was associated with lower subscapular and triceps skinfold thickness, and with reduced odd of obesity at 3 years (OR = 0·68, 95 % CI = 0·50-0·92). Moreover, higher DHA + EPA concentrations in umbilical cord plasma were similarly associated with lower adiposity (skinfold thickness) and obesity (OR = 0·09, 95 % CI = 0·02-0·52).
Five trials relating infant body composition with pre- and postnatal maternal intake of n-3 LCPUFA were included(Reference Asserhoj, Nehammer, Matthiessen, Michaelsen and Lauritzen15, Reference Helland, Smith, Blomen, Saarem, Saugstad and Drevon20, Reference Lauritzen, Hoppe, Straarup and Michaelsen23, Reference Lucia Bergmann, Bergmann, Haschke-Becher, Richter, Dudenhausen and Barclay24, Reference Hauner, Much, Vollhardt, Brunner, Schmid and Sedlmeier27) of which all were double-blinded(Reference Asserhoj, Nehammer, Matthiessen, Michaelsen and Lauritzen15, Reference Helland, Smith, Blomen, Saarem, Saugstad and Drevon20, Reference Lauritzen, Hoppe, Straarup and Michaelsen23, Reference Lucia Bergmann, Bergmann, Haschke-Becher, Richter, Dudenhausen and Barclay24) except one(Reference Hauner, Much, Vollhardt, Brunner, Schmid and Sedlmeier27) where no information was given. All presented results come from European countries (Denmark, Norway and Germany), conducted between 2000 and 2009 with the number of participants varying from 175 to 198 mother-term infant pairs. Body mass index (BMI) was reported in all five trials, skinfold thickness in four(Reference Asserhoj, Nehammer, Matthiessen, Michaelsen and Lauritzen15, Reference Lauritzen, Hoppe, Straarup and Michaelsen23, Reference Hauner, Much, Vollhardt, Brunner, Schmid and Sedlmeier27) and waist circumference in two(Reference Asserhoj, Nehammer, Matthiessen, Michaelsen and Lauritzen15, Reference Lauritzen, Hoppe, Straarup and Michaelsen23). Two studies(Reference Lucia Bergmann, Bergmann, Haschke-Becher, Richter, Dudenhausen and Barclay24, Reference Hauner, Much, Vollhardt, Brunner, Schmid and Sedlmeier27) measured head circumference and one(Reference Hauner, Much, Vollhardt, Brunner, Schmid and Sedlmeier27) also calculated weight-for-length and ponderal weight. In order to make possible the comparison of results among intervention studies, sources of n-3 LCPUFA, intervention periods as well as point of time in which the outcomes were measured were considered. In this sense, studies showed some differences (Table 2). Hauner et al. (Reference Hauner, Much, Vollhardt, Brunner, Schmid and Sedlmeier27) started supplementing pregnant women at 15 weeks of pregnancy till 4 months of lactation and infant's measurements were taken at birth, 6 weeks, 4 months and 12 months postpartum. In Lauritzen et al. (Reference Lauritzen, Hoppe, Straarup and Michaelsen23), the intervention was carried out during the first 16 weeks of lactation and outcome assessment took place at 9 months and at 2·5 years respectively. The trial of Asserhøj et al. (Reference Asserhoj, Nehammer, Matthiessen, Michaelsen and Lauritzen15), a follow-up of the Lauritzen et al. (Reference Lauritzen, Hoppe, Straarup and Michaelsen23) study, described results at 7 years of age. Helland et al. (Reference Helland, Smith, Blomen, Saarem, Saugstad and Drevon20), measured BMI as a secondary outcome at 7 years of age, in an intervention trial performed between 18 weeks of pregnancy and 3 months following delivery with either cod liver oil or corn oil differed in the LCPUFA type. In the study by Lucia Bergman et al. (Reference Lucia Bergmann, Bergmann, Haschke-Becher, Richter, Dudenhausen and Barclay24) pregnant women were supplemented with fish oil DHA (200 mg) from the 21st week of pregnancy until the 3rd month of lactation. Infants were measured at birth and thereafter at 1, 3 and 21 months.
Results on the effect of maternal n-3 supplementation on infant's body composition were inconsistent and did no enable the authors' to underpin any relevant conclusion (Table 2). Additionally, the fact that outcomes were assessed at different age points increased difficulties in result's comparison. Lauritzen et al. (Reference Lauritzen, Hoppe, Straarup and Michaelsen23) observed significantly higher BMI, waist circumference and adipose tissue in the supplemented group at 30 months compared to the control group. These differences, however, disappeared at 7 years in the Asserhøj study(Reference Asserhoj, Nehammer, Matthiessen, Michaelsen and Lauritzen15). Helland et al. (Reference Helland, Smith, Blomen, Saarem, Saugstad and Drevon20), reported no significant effect on BMI at 7 years, in the supplemented group however, concentrations of ALA in breast milk 3 months after birth were positively correlated with BMI at 7 years. On the contrary, Donahue et al. (Reference Donahue, Rifas-Shiman, Gold, Jouni, Gillman and Oken18) concluded that higher maternal prenatal n-3 intake was associated with lower adiposity in early childhood and Lucia Bergman et al. (Reference Lucia Bergmann, Bergmann, Haschke-Becher, Richter, Dudenhausen and Barclay24) showed that DHA supplements during pregnancy and lactation may reduce BMI in late infancy. More specifically, lower weight and BMI at 21 months were observed in infants whose mothers were supplemented with DHA, whereas no effects were found for height and head circumference. Hauner et al. (Reference Hauner, Much, Vollhardt, Brunner, Schmid and Sedlmeier27) however, reported no effect on infants fat mass and growth at ≤ 1 year of life between the intervention and the control groups.
Effects of infant's n-3 LCPUFA intake on body composition
A total of seven studies are presented in Table 3(Reference Birch, Hoffman, Uauy, Birch and Prestidge16, Reference de Jong, Boehm, Kikkert and Hadders-Algra17, Reference Groh-Wargo, Jacobs, Auestad, O'Connor, Moore and Lerner19, Reference Innis, Adamkin, Hall, Kalhan, Lair and Lim21, Reference Kennedy, Ross, Isaacs, Weaver, Singhal and Lucas22, Reference Ryan, Montalto, Groh-Wargo, Mimouni, Sentipal-Walerius and Doyle25, Reference Scholtens, Wijga, Smit, Brunekreef, de Jongste and Gerritsen26). One was an observational cohort study(Reference Scholtens, Wijga, Smit, Brunekreef, de Jongste and Gerritsen26) and six RCTs(Reference Birch, Hoffman, Uauy, Birch and Prestidge16, Reference de Jong, Boehm, Kikkert and Hadders-Algra17, Reference Groh-Wargo, Jacobs, Auestad, O'Connor, Moore and Lerner19, Reference Innis, Adamkin, Hall, Kalhan, Lair and Lim21, Reference Kennedy, Ross, Isaacs, Weaver, Singhal and Lucas22, Reference Ryan, Montalto, Groh-Wargo, Mimouni, Sentipal-Walerius and Doyle25) four of which performed in North America (mainly in the US)(Reference Birch, Hoffman, Uauy, Birch and Prestidge16, Reference Groh-Wargo, Jacobs, Auestad, O'Connor, Moore and Lerner19, Reference Innis, Adamkin, Hall, Kalhan, Lair and Lim21, Reference Ryan, Montalto, Groh-Wargo, Mimouni, Sentipal-Walerius and Doyle25) and three in Europe: two in The Netherlands(Reference de Jong, Boehm, Kikkert and Hadders-Algra17, Reference Scholtens, Wijga, Smit, Brunekreef, de Jongste and Gerritsen26), and one in the UK(Reference Kennedy, Ross, Isaacs, Weaver, Singhal and Lucas22). One was single-blind(Reference Ryan, Montalto, Groh-Wargo, Mimouni, Sentipal-Walerius and Doyle25) and five were double-blind(Reference Birch, Hoffman, Uauy, Birch and Prestidge16, Reference de Jong, Boehm, Kikkert and Hadders-Algra17, Reference Groh-Wargo, Jacobs, Auestad, O'Connor, Moore and Lerner19, Reference Innis, Adamkin, Hall, Kalhan, Lair and Lim21, Reference Kennedy, Ross, Isaacs, Weaver, Singhal and Lucas22).
Preterm infants. Four out of seven studies included preterm newborns and varied in sample size (60 to 194 infants)(Reference Groh-Wargo, Jacobs, Auestad, O'Connor, Moore and Lerner19, Reference Innis, Adamkin, Hall, Kalhan, Lair and Lim21, Reference Kennedy, Ross, Isaacs, Weaver, Singhal and Lucas22, Reference Ryan, Montalto, Groh-Wargo, Mimouni, Sentipal-Walerius and Doyle25), intervention periods (4 weeks to 12 months) as well as in formula composition with n-3 fatty acid. All used DHA as the main n-3 differential fatty acid with contents ranging from 0·2 % to 0·5 % of total fatty acid weight. The composition of the control formulas differed to that of supplemented formulas in different terms: in two studies LA and ALA but no DHA and AA were added(Reference Groh-Wargo, Jacobs, Auestad, O'Connor, Moore and Lerner19, Reference Innis, Adamkin, Hall, Kalhan, Lair and Lim21), in one study the control formula did not contain any added n-3 LCPUFA(Reference Ryan, Montalto, Groh-Wargo, Mimouni, Sentipal-Walerius and Doyle25), and in another the control formula was supplemented with smaller amounts of LA and ALA compared to the supplemented formula(Reference Kennedy, Ross, Isaacs, Weaver, Singhal and Lucas22). Three trials(Reference Groh-Wargo, Jacobs, Auestad, O'Connor, Moore and Lerner19, Reference Kennedy, Ross, Isaacs, Weaver, Singhal and Lucas22, Reference Ryan, Montalto, Groh-Wargo, Mimouni, Sentipal-Walerius and Doyle25) accurately measured body compartments as body fat (BF) and fat-free mass (FFM) although methods of assessment and indicators of body composition varied i.e., skinfolds were assessed in two studies(Reference Kennedy, Ross, Isaacs, Weaver, Singhal and Lucas22, Reference Ryan, Montalto, Groh-Wargo, Mimouni, Sentipal-Walerius and Doyle25), one measured total body conductivity(Reference Ryan, Montalto, Groh-Wargo, Mimouni, Sentipal-Walerius and Doyle25), dual X-ray absorptiometry (DEXA) was used in another one(Reference Groh-Wargo, Jacobs, Auestad, O'Connor, Moore and Lerner19) and both bioelectrical impedance and deuterium dilution in another study(Reference Kennedy, Ross, Isaacs, Weaver, Singhal and Lucas22).
Results among trials varied in use of supplemented LCPUFA formulas, age and gender groups. Innis et al. (Reference Innis, Adamkin, Hall, Kalhan, Lair and Lim21) reported higher weight, length and weight-to-length ratio in infants fed with DHA and AA from algal/fungal oils (0·33 % DHA and 0·60 % AA) supplemented formula for 4 weeks to the controls (formula 21-22 % LA, 3-3·1 % ALA) and those given single-cell algal oil DHA supplemented formula at 40-57 weeks post-menstrual age (PMA). Infants supplemented with DHA and AA formula during 12 months had significantly more FFM and less BF than controls (16-19 % LA and 2·5 % ALA) at 1 year of age; however, no differences were reported for weight, height, and head circumference(Reference Groh-Wargo, Jacobs, Auestad, O'Connor, Moore and Lerner19). In preterm infants fed with supplemented formula containing DHA and EPA until 9(Reference Kennedy, Ross, Isaacs, Weaver, Singhal and Lucas22) and 59 weeks postpartum of PMA(Reference Ryan, Montalto, Groh-Wargo, Mimouni, Sentipal-Walerius and Doyle25), lower FFM and BF in males(Reference Ryan, Montalto, Groh-Wargo, Mimouni, Sentipal-Walerius and Doyle25) and increased weight and adiposity at 9-11 years among females(Reference Lauritzen, Hoppe, Straarup and Michaelsen23) was observed.
Term infants. Three out of seven studies were RCTs with a sample size of 79(Reference Birch, Hoffman, Uauy, Birch and Prestidge16) and 341(Reference de Jong, Boehm, Kikkert and Hadders-Algra17) infants respectively. One study was a cohort including 244 mother-infant pairs(Reference Scholtens, Wijga, Smit, Brunekreef, de Jongste and Gerritsen26). The intervention periods with LCPUFA supplemented formulas varied from 2 months(Reference de Jong, Boehm, Kikkert and Hadders-Algra17) to 4 months(Reference Birch, Hoffman, Uauy, Birch and Prestidge16). DHA was used as the main n-3 differential fatty acid for supplementation(Reference Birch, Hoffman, Uauy, Birch and Prestidge16, Reference de Jong, Boehm, Kikkert and Hadders-Algra17). No long-term changes in body composition were observed at 1(Reference Birch, Hoffman, Uauy, Birch and Prestidge16) or 9 years of age(Reference de Jong, Boehm, Kikkert and Hadders-Algra17). Reported results in term infants are similarly inconsistent those of preterm infants. No clear relationship between any of the used LCPUFA supplemented formulas and later body composition variability was found. The findings of the cohort study which measured the composition of maternal milk in fatty acids suggested that the n-3 and n-6 LCPUFA content did not have an effect on weight gain or BMI during the first year of life in this group of breast-fed infants(Reference Scholtens, Wijga, Smit, Brunekreef, de Jongste and Gerritsen26).
Quality of included studies
Table 4 summarizes the method used to assess the quality of the included studies. Different levels of risk of bias among the studies were observed. For intervention studies involving infants' mothers, four studies had a high risk of bias(Reference Asserhoj, Nehammer, Matthiessen, Michaelsen and Lauritzen15, Reference Helland, Smith, Blomen, Saarem, Saugstad and Drevon20, Reference Lauritzen, Hoppe, Straarup and Michaelsen23, Reference Hauner, Much, Vollhardt, Brunner, Schmid and Sedlmeier27) and one had moderate risk(Reference Lucia Bergmann, Bergmann, Haschke-Becher, Richter, Dudenhausen and Barclay24). Repeated reasons for having risk of bias included an inadequate or unclear blinding procedure, inadequate explanation on dropouts, and an inadequate funder and other potential threats of validity. The only observational study included(Reference Donahue, Rifas-Shiman, Gold, Jouni, Gillman and Oken18) was defined to be of low risk of bias due to insufficient description of dropouts classified as “unclear”. Regarding the studies focused exclusively on children, the observational study(Reference Scholtens, Wijga, Smit, Brunekreef, de Jongste and Gerritsen26) had a moderate risk of bias because of several threats to validity. Out of the seven RCTs(Reference Birch, Hoffman, Uauy, Birch and Prestidge16, Reference de Jong, Boehm, Kikkert and Hadders-Algra17, Reference Groh-Wargo, Jacobs, Auestad, O'Connor, Moore and Lerner19, Reference Innis, Adamkin, Hall, Kalhan, Lair and Lim21, Reference Kennedy, Ross, Isaacs, Weaver, Singhal and Lucas22, Reference Ryan, Montalto, Groh-Wargo, Mimouni, Sentipal-Walerius and Doyle25), two(Reference de Jong, Boehm, Kikkert and Hadders-Algra17, Reference Ryan, Montalto, Groh-Wargo, Mimouni, Sentipal-Walerius and Doyle25) were classified at high risk of bias and four(Reference Birch, Hoffman, Uauy, Birch and Prestidge16, Reference Groh-Wargo, Jacobs, Auestad, O'Connor, Moore and Lerner19, Reference Innis, Adamkin, Hall, Kalhan, Lair and Lim21, Reference Kennedy, Ross, Isaacs, Weaver, Singhal and Lucas22)at moderate risk of bias. Insufficient description of funding and other potential threats to validity were some of the reasons.
Table 4 Assessment of methodological quality of included RCTs and longitudinal studies.
NA, not applicable
Donahue: Insufficient information on drop-outs.
Lauritzen: Insufficient information on the sequence generation and allocation concealment. Group sizes were based on power calculation for infant visual acuity instead of infant growth which is our main outcome.
Lucia Bergmann: Insufficient information on drop-outs. No explanation on differences between those who completed the study and drop-outs.
Helland: Insufficient information on sequence generation, allocation concealment and blinding procedure. Maternal age and education significantly differed among those mothers included in the study and those who were excluded. The study was co-financed by an enterprise which also provided the supplements. Power calculation of group sizes was based on an infant intelligence questionnaire instead of using infant growth which is our main outcome. Moreover, main outcomes were eventually related to umbilical cord or breast milk concentrations and not to n-3 LCPUFAs intake or supplementation.
Asserhøj: Insufficient information on sequence generation, allocation concealment and blinding procedure. They were not aware of the group allocation at 7y. Maternal age and education significantly differed among those mothers included in the study and those who were excluded. The study was co-financed by an enterprise which also provided the supplements. Power calculation of group sizes was based on an infant intelligence questionnaire instead of using infant growth which is our main outcome.
Hauner: The blinding procedure did not exclude potential bias. Moreover, the study was co-financed by private enterprises.
Birch: The sample size was calculated based on outcome (visual evoked potential) other to the ones examined in this review. Moreover, the supplemented formulas were provided by a private enterprise.
Ryan: No description of blinding method or funding source.
Innis: No description on dropouts. Moreover, supplemented formulas were provided by a private enterprise.
Groh-Wargo: The funder cannot be considered as adequate because the study was partially supported by private enterprise. The fact that subjects were allowed to take supplements but they did not register that information risks study validity.
Scholtens: Mother allocated in the intervention group were highly educated and smoked less often during pregnancy compared to those of the general study population. Moreover, children weight was self-reported.
Kennedy: Gestational age and birth-weight SD scores of dropouts were significantly different from the final included sample. Not controlled intakes of other n-3 LCPUFA. The study was funded by a private enterprise.
DeJong: Sequence generation and allocation concealment were not insufficiently described. The sample size was calculated based on health outcome other to the ones examined in this review. Supplementation period was too short.
Discussion
The aim of this systematic review was to identify and summarize evidence on studies assessing the relationship between n-3 LCPUFA intake on infancy and early childhood body composition across consecutive life stage. A total of 13 studies(Reference Asserhoj, Nehammer, Matthiessen, Michaelsen and Lauritzen15–Reference Hauner, Much, Vollhardt, Brunner, Schmid and Sedlmeier27) met the inclusion criteria and were included in this review. A very sensitive search was performed, considering limited availability of literature addressing the topic. References from excluded papers were also reviewed to avoid the omission of any relevant report. Appraised studies were categorised into two groups; those of mothers and infants and to those of infants. It should be taken into consideration that studies carried out in newborns are less solid due to the ethical implications.
The most important strength of the present study is that the review has been performed systematically. On the other hand, the limited number of articles relating intake of n-3 LCPUFA supplemented formulas and short and long-term body composition variability in infants represents the main study weakness. It is difficult to obtain conclusive and comparable results from the scarce existing information. In addition, there is a major discrepancy between the results obtained across the available studies.
Effects of maternal n-3 LCPUFA intake during pregnancy and lactation on body composition during infancy and childhood
All intervention studies had an RCT design. Points which should be considered when interpreting the results of the findings include: geographical location, patterns of intake regarding cold water fish, source of n-3 LCPUFA, and comparability of outcome assessment.
Other important points affecting comparisons and which should be considered include the type of LCPUFA supplemented, formula composition, the different sources of n-3 LCPUFA, the duration of the intervention and the timeframe in which the outcome was measured. The results of this review showed wide variability. The duration of intervention, which ranged from 16 weeks(Reference Lauritzen, Hoppe, Straarup and Michaelsen23) to 36 weeks(Reference Helland, Smith, Blomen, Saarem, Saugstad and Drevon20), as well as the point in time when the outcome was assessed may have an influence on the outcomes evaluated. While two studies measured body composition at 7 years(Reference Asserhoj, Nehammer, Matthiessen, Michaelsen and Lauritzen15, Reference Helland, Smith, Blomen, Saarem, Saugstad and Drevon20), in one trial measurements were taken at 1 year(Reference Hauner, Much, Vollhardt, Brunner, Schmid and Sedlmeier27). Three studies(Reference Helland, Smith, Blomen, Saarem, Saugstad and Drevon20, Reference Lucia Bergmann, Bergmann, Haschke-Becher, Richter, Dudenhausen and Barclay24, Reference Hauner, Much, Vollhardt, Brunner, Schmid and Sedlmeier27) comprised both pregnancy and lactation periods; Donahue et al. (Reference Donahue, Rifas-Shiman, Gold, Jouni, Gillman and Oken18) only considered pregnancy and two studies evaluated the intake of n-3 LCPUFA exclusively during lactation(Reference Asserhoj, Nehammer, Matthiessen, Michaelsen and Lauritzen15, Reference Lauritzen, Hoppe, Straarup and Michaelsen23). That variability on methodology could explain the different findings observed across trials. Two studies(Reference Helland, Smith, Blomen, Saarem, Saugstad and Drevon20, Reference Hauner, Much, Vollhardt, Brunner, Schmid and Sedlmeier27) did not observe any effect on body composition in terms of fat mass, growth or BMI, although in Hauner et al. (Reference Hauner, Much, Vollhardt, Brunner, Schmid and Sedlmeier27) outcome measures were obtained much earlier (12 months after birth) than in the other studies. One study(Reference Lucia Bergmann, Bergmann, Haschke-Becher, Richter, Dudenhausen and Barclay24) positively associated n-3 LCPUFA intake with body composition at 2·5y but further measurements at 7y of age(Reference Asserhoj, Nehammer, Matthiessen, Michaelsen and Lauritzen15) did not confirm that association suggesting a time-dependent effect on later body composition. Lucia Bergman et al. (Reference Lucia Bergmann, Bergmann, Haschke-Becher, Richter, Dudenhausen and Barclay24), however, associated maternal supplementation with DHA with a decrease in children weight and BMI, although no effect was observed on length and head circumference. It is clear that included articles are heterogeneous regarding the interventions as well as the type, sources and doses of n-3 LCPUFA used (see Tables 2 and 3). It is important to remark that exposure time to LCPUFA and EPA/DHA or n-3/n-6 fatty acid ratios might contribute to differential outcome effects.
Literature availability addressing this topic longitudinally is scarce and did not enable conclusion drawing. This is reflected by the nearly non-existent identified literature, since only one paper showing results on the effects of n-3 LCPUFA intake in early periods of life on later adiposity met the inclusion criteria and is presented in this review.
Effects of infant n-3 LCPUFA intake on body composition
Limited literature availability similarly to available evidence on maternal intake should be stressed. Only one paper presented results on the effects of n-3 LCPUFA intake at early life on later body composition in term newborns(Reference Scholtens, Wijga, Smit, Brunekreef, de Jongste and Gerritsen26). In this study, neither n-3 nor n-6 LCPUFA breast milk content influences weight gain or BMI.
All intervention studies were performed in developed countries. Differences in applied protocols in terms of LCPUFA formula composition or the source of n-3 LCPUFA, target sample, sample size, supplementation design of intervention / control groups intervention and monitoring period duration, outcomes measures and implications of accuracy of body composition assessment should be noted.
Preterm infants supplemented with DHA and AA formula in two studies(Reference Groh-Wargo, Jacobs, Auestad, O'Connor, Moore and Lerner19, Reference Innis, Adamkin, Hall, Kalhan, Lair and Lim21) had significantly higher weight, length, weight-to-length ratio(Reference Innis, Adamkin, Hall, Kalhan, Lair and Lim21) and FFM but less BF(Reference Groh-Wargo, Jacobs, Auestad, O'Connor, Moore and Lerner19) compared with controls at 1 year of age. Although n-3 LCPUFAs may have antiadipogenic effects inhibiting fat development, AA appears to have an adipogenic effect(Reference Makrides, Gibson, Udell and Ried9, Reference Amri, Ailhaud and Grimaldi11). Recent systematic reviews however, showed that global growth of both term and preterm infants assessed in terms of weight, length and head circumference was unaffected by LCPUFA intake(Reference Schulzke, Patole and Simmer8). Despite the lack of evidence on the relationship between changes on body composition and LCPUFA intake, n-3/n-6 LCPUFA ratio as well as doses applied and fatty acid sources are still taken into consideration by the studies(Reference Kennedy, Ross, Isaacs, Weaver, Singhal and Lucas22, Reference Ryan, Montalto, Groh-Wargo, Mimouni, Sentipal-Walerius and Doyle25). In addition, in preterm infants fed supplemented formulas with fish oil containing DHA and EPA (without AA supplementation) during the first months of life it was observed lower FFM and BF among males(Reference Ryan, Montalto, Groh-Wargo, Mimouni, Sentipal-Walerius and Doyle25), and higher weight and adiposity at 9-11 years among females(Reference Kennedy, Ross, Isaacs, Weaver, Singhal and Lucas22). These findings are in concordance to those observed by Donahue et al. (Reference Donahue, Rifas-Shiman, Gold, Jouni, Gillman and Oken18) where higher prenatal fish intake and exposure to n-3 LCPUFAs were associated with lower adiposity in early childhood. A programming effect could be involved in the modulation of pre-adolescent body composition in subjects with low early adiposity having a different effect in each gender. It should be noted however, that there is a number of studies which failed to show any body composition variations later in life(Reference Birch, Hoffman, Uauy, Birch and Prestidge16, Reference de Jong, Boehm, Kikkert and Hadders-Algra17, Reference Scholtens, Wijga, Smit, Brunekreef, de Jongste and Gerritsen26).
Conclusions and final comments
Two studies(Reference Donahue, Rifas-Shiman, Gold, Jouni, Gillman and Oken18, Reference Lucia Bergmann, Bergmann, Haschke-Becher, Richter, Dudenhausen and Barclay24) showed positive effects of n-3 LCPUFA (only DHA or DHA+EPA) maternal intake during pregnancy and lactation on infant and young children's later body composition i.e., decreasing adiposity or BMI. Breastfed infants whose mothers were supplemented with fish oil during lactation had higher BMI, adipose tissue and waist circumference at 2·5 and at 7 years of age(Reference Asserhoj, Nehammer, Matthiessen, Michaelsen and Lauritzen15, Reference Lauritzen, Hoppe, Straarup and Michaelsen23). Breast milk n-3 LCPUFA content but not umbilical cord levels were associated with BMI at 7 years of age(Reference Helland, Smith, Blomen, Saarem, Saugstad and Drevon20). One study did not found any related effect(Reference Hauner, Much, Vollhardt, Brunner, Schmid and Sedlmeier27). Focusing on the perinatal period, three out of seven studies did not observed any effects on later body composition among infants supplemented with n-3 LCPUFA enriched formulas(Reference Birch, Hoffman, Uauy, Birch and Prestidge16, Reference de Jong, Boehm, Kikkert and Hadders-Algra17, Reference Scholtens, Wijga, Smit, Brunekreef, de Jongste and Gerritsen26). On the other hand, two studies(Reference Innis, Adamkin, Hall, Kalhan, Lair and Lim21, Reference Kennedy, Ross, Isaacs, Weaver, Singhal and Lucas22) reported higher weight and adiposity when consuming increased amounts of n-3 LCPUFA (DHA+AA), whereas two(Reference Groh-Wargo, Jacobs, Auestad, O'Connor, Moore and Lerner19, Reference Ryan, Montalto, Groh-Wargo, Mimouni, Sentipal-Walerius and Doyle25) showed a decrease in weight and fat mass (DHA+EPA and high ALA intake).
In summary, evidence on the potential relationship between maternal n-3 LCPUFA intake and infant growth or later body composition are not conclusive. In addition, contradictory findings among trials on use of varied supplemented n-3 LCPUFA formulas and on short and long-term effects on body composition or body fat were observed. Results derived from the studies included in this systematic review suggest that mechanisms are not understood and data synthesis is inconclusive. Differences in n-3 LCPUFA formula composition due to the heterogeneity in the type, sources and doses of LCPUFA as well as the timeframe of exposure prevent conclusive findings. Therefore, the association between early n-3 LCPUFA exposure during perinatal period and long-term body composition remains unclear. More studies addressing this relationship are needed.
Acknowledgements
This study has been supported by a grant from the Spanish Health Institute Carlos III (RD08/0072: Maternal, Child Health and Development Network) within the framework of the VI National R+D+i Research Programme (2008-2011). SBS is funded by a grant from the Aragon's Regional Government (Diputación General de Aragón, DGA). This entity also partially supports the entire research group. II is financially supported by EURRECA-Network of Excellence-. The authors have no conflict of interest. The author's contributions were as follows: GR, II and SBS designed the study, the literature search and the systematic review; GR, II, SBS and LAM wrote the manuscript and critically discussed and revised the article.
