European Journal of Clinical Nutrition (1997) 51, 578±584 ß 1997 Stockton Press. All rights reserved 0954±3007/97 $12.00 Effect of increasing breast milk docosahexaenoic acid on plasma and erythrocyte phospholipid fatty acids and neural indices of exclusively breast fed infants

RA Gibson1, MA Neumann1 and M Makrides2

1Department of Paediatrics & Child Health, Flinders Medical Centre, Bedford Park, Adelaide, SA 5042, Australia; and 2Department of Paediatrics & Child Health, Flinders University of South Australia, Bedford Park, Adelaide, SA 5042, Australia

Objectives: To determine the effect of increasing docosahexaenoic acid (DHA, 22 : 6 n-3) in breast milk on infant fatty acid pro®les. A secondary aim was to examine aspects of neural development. Design and Setting Double blind, placebo controlled study of infants recruited from postnatal wards at Flinders Medical Centre. Subjects: Fifty-two healthy term infants who were breast fed for at least 12 weeks and were from middle class families. Intervention: Breast milk with DHA concentrations that ranged from 0.1±1.7% of total fatty acids. This was achieved by supplementation of the maternal diet for the ®rst 12 weeks post partum. Results: Breast milk DHA was related to infant plasma (r ˆ 0.89, P < 0.001) and erythrocyte (r ˆ 0.88, P < 0.001) phospholipids in a saturable curvilinear manner so that breast milk DHA above 0.8% of total fatty acids resulted in little further increase in infant plasma or erythrocyte DHA levels. The rise in plasma and erythrocyte DHA was approximated by a fall in total n-6 polyunsaturated fatty acids. We could detect no relationship between visual evoked potential acuity (measured at 12 and 16 weeks) of infants by either the dietary grouping or the DHA status of individuals. A stepwise multiple regression showed that infant erythrocyte DHA at 12 weeks and home stimulation were the only independent factors associated with Bayley's MDI at 1 y (adjusted model r2 ˆ 0.18, P < 0.005); while at 2 y gender and social score of the spouse were the only signi®cant predictors of Bayley's MDI (adjusted model r2 ˆ 0.22, P < 0.005). Conclusions: Increasing breast milk DHA levels caused a dose dependent saturable increase in infant plasma and erythrocyte phospholipid DHA. There were no long-term effects of infant DHA status on indices of neurodevelopment. Sponsorship: Financial support was provided by Martek Biosciences, MD, USA and the National Health and Medical Research Council, Australia. Descriptors: Breast milk, docosahexaenoic acid (DHA), neural development.

Introduction PUFA (Makrides et al, 1996). This paper reports on the primary aim of the study which was to examine the effect Breast milk contains a variety of long chain polyunsatu- of varying the DHA concentration in breast milk on the rated fatty acids (LCPUFA) including docosahexaenoic LCPUFA status of exclusively breast fed term infants. acid (DHA, 22 : 6 n-3). Infants can be exposed to a wide Several studies have reported that children who were range of DHA levels in breast milk that are largely breast fed as infants had superior development compared dependent on the dietary intake of the mother (Harris et with those who were formula fed (Rogan & Gladen, 1993; al, 1984; Henderson et al, 1992; Makrides et al, 1996). For Lucas et al, 1992; Morrow-Tlucak et al, 1988). This has example, breast fed Korean or Malaysian infants would been related to the presence of the DHA in breast milk, a receive up to ®ve times the amount of DHA of their fatty acid which is not included in most formulas. The Western cousins (Kneebone et al, 1985; Choue et al, association between dietary DHA and improved retinal and 1995). Supplementing the diet of Western women with neural outcomes has been veri®ed in both observational ®sh oil rich in n-3 LCPUFA results in elevated levels of the studies and randomised trials involving both term and n-3 LCPUFA eicosapentaenoic acid (EPA, 20 : 5 n-3) and preterm infants (Birch et al, 1992a; Birch et al, 1992b; DHA (Henderson et al, 1992), but no systematic study on Makrides et al, 1995; Carlson et al, 1993; Agostoni et al, the effect of a single individual n-3 PUFA have been 1995). conducted. The availability of a unicellular oil containing The dose of DHA required for optimal neural response DHA as its only PUFA gave us the opportunity of examin- has not been addressed. Such reports have prompted some ing the effect of increasing DHA in the maternal diet on authors (Nettleton, 1993; Holman et al, 1991) and Nutrition breast milk fatty acid composition (Makrides et al, 1996); Committees (British Nutrition Foundation Task Force, increased dietary DHA resulted in a linear increase in 1992) to recommend that nursing mothers increase their breast milk DHA but did not affect other breast milk intake of DHA in order to furnish DHA for their infants. However, such recommendations assume that the DHA Correspondence: Dr RA Gibson. requirement for optimal infant development is known and Received 7 February 1997; revised 18 April 1997; accepted 1 May 1997 that the level of DHA in the breast milk of Western Increasing breast milk docosahexaenoic acid RA Gibson et al 579 omnivorous women is limiting for neural function. Thus, tested at each age (10, 14, 20, 28, 35, 42 and 55 min of our secondary aim was to investigate whether two neural arc at 12 weeks and 7, 10, 14, 20, 28, 42 and 55 min arc at indices of infants (visual evoked potential (VEP) acuities 16 weeks). The P100 latency of the averaged VEP at each and Bayley's Scales of Development) were dependent on checkerboard pattern was recorded. The peak to peak the DHA concentration of breast milk. amplitude of the VEP (N1-P100) response was measured and plotted against log of the angle subtended by each check size. The linear portion of the plot was extrapolated Methods to 0 mV to give the theoretical value that would just elicit a Subjects response (log of the minimum angle of resolution, Mothers of term infants (> 37 weeks gestation) who logMAR). Hence, lower logMAR values represent better intended to breast feed for at least 12 weeks were recruited visual acuity as these infants are capable of eliciting to participate in this study. These mothers were randomised responses to smaller checkerboard patterns. Points were to receive one of ®ve doses (0, 0.2, 0.4, 0.9 or 1.3 g DHA/ excluded from the regression if they were not on the linear d) of a DHA-rich algal oil (DHASCO, Martek Biosciences, portion of the stimulus-response function or represented MD, USA), between day 5 and week 12 post partum amplitudes of  2 mV. VEP acuity extrapolations were (Makrides et al, 1996). The oil contained 43% DHA, 1% accepted as valid only if there were at least three points n-6 PUFA, 38% saturates and 18% monounsaturates. and the regression line was signi®cant (P < 0.05). The Infants who were exclusively breast fed for 12 weeks process of conducting the VEP test was similar to our were assessed. All infants were healthy, appropriate earlier published work (Makrides et al, 1995) although the weight for gestation and had apgar scores greater than 7 hardware and software was updated to allow a greater range at 5 minutes post birth. Written consent was obtained from of checkerboard patterns to be presented to the infant, and all mothers according to the protocol approved by the thus more points were available for a more reliable extra- Committee on Clinical Investigations (Ethics) at Flinders polation. Medical Centre. Each infant was also assessed at 1 and 2 y of age using the Bayley's Scales of Infant Development. Mental (MDI) Research plan and Psychomotor (PDI) Developmental Indices were stan- Length of gestation, birth anthropometrics, parental social dardised for age based on established reference norms class (Daniel, 1983) and education levels (Makrides et al, (Bayley, 1969). All mothers also completed a short ques- 1995) were recorded at enrolment. Infant growth was tionnaire assessing the amount of stimulation a child assessed by weight, length and head circumference at 12 receives in the home (Coons et al, 1981). weeks, 1 and 2 y of age. Growth measurements were used as an index of nutritional status and to con®rm that all Statistics infants were appropriate weight, length and head circum- Differences in plasma and erythrocyte fatty acids between ference for age. All 52 infants, whose mothers successfully diet groups were examined by one-way analysis of variance completed the supplementation phase of the study (12 (ANOVA). Relationships between breast milk DHA and weeks) were assessed for visual function at 12 and 16 infant plasma and erythrocyte phospholipid fatty acids were weeks, and for global development at 1 and 2 y of age. plotted and curve ®tting models were used to establish Infants' dietary intake of DHA was assessed by measuring regressions. One way ANOVA was also used to test for breast milk DHA levels at the end (week 12) of supple- differences between dietary groups for VEP acuity, Bay- mentation. ley's MDI and PDI. Paired t-test was used to determine the Regular telephone contact was maintained with all effect of age on VEP acuity between 12 and 16 weeks. mothers to support and encourage breast feeding. Length Associations between DHA in breast milk, DHA status of of breast feeding was de®ned as weeks of exclusive breast the infant and neural indices were tested using Pearson feeding without formula or cows milk supplements. correlation coef®cients. One MDI data point appeared to lie Blood samples from infants (200 mL) were taken by heel outside the range of all others. Investigations failed to ®nd a prick at 12 weeks of age for assessment of plasma and legitimate reason to exclude the value from the data set but erythrocyte phospholipid fatty acids. Plasma and erythro- outlier analysis indicated that it was a major weighting cytes were separated by centrifugation and the erythrocytes factor to the correlation calculation (Mahalanoblis distance were washed three times with isotonic saline. Plasma and of greater than 6). This point was therefore excluded from erythrocyte lipids were extracted with chloroform : metha- the statistical analysis. A stepwise multiple regression was nol (Bligh & Dyer, 1959) and chloroform : propanol (Broe- used to predict independent factors associated with Bay- khuyse, 1974), respectively. The phospholipid fractions of ley's MDI. Statistical analyses were completed using SPSS both plasma and erythrocyte lipid extracts were separated for Windows 6.0 (SPSS Inc, Chicago) and Sigma Plot for by thin-layer chromatography, methylated and quanti®ed Windows 2.0 (Jandel Corp, San Rafael) was used for curve by capillary gas chromatography (Makrides et al, 1996). ®tting. VEPs were recorded under transient conditions using the Enfant 4010 system (Neuroscienti®c Corp. Farmingdale, Results NY, USA) when infants were 12 and 16 weeks of age. Infants were seated, with a parent, 1 m away from a 1900 All 52 infants who were exclusively breast fed for 12 weeks monitor presenting high contrast black and white checker- were assessed at 12 and 16 weeks and 1 y of age. Two board pattern reversal (2 Hz) stimuli. The active electrode infants were lost to follow-up at 2 y because the families was placed 3 cm above the inion, the reference electrode at had relocated interstate. Infants who completed the study 30% of the nasion-to-inion distance and the inactive elec- had mature mothers (mean age 30 Æ 4 y), were from middle trode was on the forehead. Two recordings were performed class families and on average had parents who had com- at each checkerboard pattern and these were subsequently pleted year 12 education (Table 1). The parents of these averaged. Seven different checkerboard patterns were infants also provided the infants with above average Increasing breast milk docosahexaenoic acid RA Gibson et al 580 Table 1 Description of study subjects according to DHA dose received in breast milk

Parameter 0.21 Æ 0.07 a 0.35 Æ 0.04 0.46 Æ 0.16 0.86 Æ 0.24 1.13 Æ 0.33 (n ˆ 12) (n ˆ 10) (n ˆ 12) (n ˆ 10) (n ˆ 8)

Maternal age (y Æ s.d.) 30 Æ 4 32 Æ 3 29 Æ 3 30 Æ 6 29 Æ 2 Maternal BMI (kg/m2 Æ s.d.) 23 Æ 3 23 Æ 2 25 Æ 6 24 Æ 5 23 Æ 3 Gestation (wks Æ s.d.) 39 Æ 1 40 Æ 1 40 Æ 2 40 Æ 1 39 Æ 1 M : F ratio 5 : 7 3 : 7 7 : 5 3 : 7 7 : 1 Birth weight (kg Æ s.d.) 3.4 Æ 0.3 3.6 Æ 0.4 3.7 Æ 0.6 3.4 Æ 0.5 3.7 Æ 0.6 Birth length (cm Æ s.d.) 50 Æ 2 51 Æ 2 52 Æ 3 51 Æ 2 53 Æ 2 Birth head circumf. (cm Æ s.d.) 35 Æ 1 35 Æ 1 35 Æ 1 35 Æ 1 36 Æ 2 Apgar score at 5 min ( Æ s.d.) 9.3 Æ 0.6 9.2 Æ 0.4 9.2 Æ 0.8 9.1 Æ 0.6 9.1 Æ 0.4 Siblings (n Æ s.d.) 0.4 Æ 0.8 0.6 Æ 0.7 0.8 Æ 1.3 1.2 Æ 1.0 1.1 Æ 1.1 Maternal social scoreb ( Æ s.d.) 4.16 Æ 1.04 4.24 Æ 1.48 4.98 Æ 0.83 5.13 Æ 0.95 4.84 Æ 0.88 Spouse social score ( Æ s.d.) 4.37 Æ 0.92 4.12 Æ 1.47 4.75 Æ 1.18 4.76 Æ 0.83 4.25 Æ 1.00 Maternal educationc (score Æ s.d.) 3.33 Æ 1.15 3.50 Æ 1.43 2.75 Æ 0.97 2.50 Æ 0.71 3.63 Æ 0.52 Partner education (score Æ s.d.) 3.42 Æ 1.08 4.20 Æ 1.55 3.18 Æ 1.33 3.33 Æ 1.12 3.25 Æ 0.89 Mother smokes yes : no 1 : 11 2 : 8 3 : 9 3 : 7 0 : 8 Partner smokes yes : no 2 : 10 3 : 7 3 : 9 1 : 9 3 : 5 Home stimulation (score Æ s.d.) 35 Æ 3 36 Æ 1 35 Æ 4 33 Æ 5 36 Æ 2 Length breast fed (weeks Æ s.d.) 40 Æ 15 40 Æ 16 37 Æ 15 33 Æ 16 45 Æ 9 n breast fed for 52 wks 5 6 3 2 4

a % total fatty acids in human milk as DHA Æ s.d., from the groups of women consuming supplements with 0, 0.2, 0.4, 0.9 and 1.3 g DHA per day. b The highest rank (1) was assigned to professionals and academic occupations and the lowest rank (6) was assigned to unskilled occupations. c Education was ranked using a seven point scale where 0 was used for no formal education, 1 for primary school level, 2 for mid secondary school level, 3 for the completion of secondary school, 4 for the completion of a certi®cate or diploma, 5 for a tertiary degree and 6 for a higher degree.

stimulation in the home (Table 1). The average duration of dependent. In contrast, levels of 22 : 5 n-3 in infant tissues breast feeding was 37 Æ 15 weeks and 20 infants were were negatively related to dietary DHA although the effect exclusively breast fed for 52 weeks (Table 1). There was was also small. a predominance of boys in the group that received the The increase in infant plasma and erythrocyte phospho- highest dose of DHA (Table 1). All infants grew well and lipid DHA was approximately matched by a decline in total were appropriate weight, length and head circumference at n-6 fatty acids from plasma and erythrocyte phospholipids birth, 12 weeks, one and two years of age. (Figure 1). Infant plasma and erythrocyte phospholipid arachidonic acid (AA, 20 : 4 n-6) was negatively associated with breast milk DHA (Figure 1) despite the relatively Plasma and erythrocyte fatty acids constant intake of AA from breast milk (Makrides et al, The mean fatty acid composition of plasma and erythrocyte 1996). There were also negative associations between phospholipids at 12 weeks of age for each dietary group are breast milk DHA and infant linoleic acid (LA, 18 : 2 n-6; listed in Tables 2 and 3, respectively. Infant plasma and r ˆ 70.33, P < 0.05 for plasma; r ˆ 70.37, P < 0.01 for erythrocyte phospholipid DHA were related to breast milk erythrocytes), 22 : 4 n-6 (r ˆ 70.65, P < 0.001 for plasma; DHA in a saturable curvilinear manner (r ˆ 0.89, P < 0.001 r ˆ 70.60, P < 0.001 for erythrocytes) and 22 : 5 n-6 for plasma; r ˆ 0.88, P < 0.001 for erythrocytes, Figure 1). (r ˆ 70.68, P < 0.001 for plasma; r ˆ 70.61, P < 0.001 The level of DHA in plasma and erythrocytes reached a for erythrocytes). plateau when breast milk DHA was  0.80% total fatty acids. We previously reported that increasing maternal dietary DHA had no effect on breast milk EPA levels VEP (Makrides et al, 1996). Despite this, infant EPA levels More VEP acuity results were excluded at 12 weeks (50%) rose slightly in plasma and erythrocyte and were diet than at 16 weeks (31%) (Table 4). Reasons for excluding

Table 2 Infant plasma phospholipids at 12 weeks post-partum (fatty acids expressed as a % of total fatty acids Æ s.d.)

Fatty acid 0.21 Æ 0.07* 0.35 Æ 0.04 0.46 Æ 0.16 0.86 Æ 0.24 1.13 Æ 0.33 (n ˆ 12) (n ˆ 10) (n ˆ 12) (n ˆ 10) (n ˆ 8)

Total sats 44.3 Æ 0.8 44.3 Æ 0.9 44.4 Æ 0.8 43.9 Æ 1.3 45.0 Æ 0.7 Total mono 12.0 Æ 0.7 12.0 Æ 1.2 12.2 Æ 1.1 12.5 Æ 0.8 12.0 Æ 1.1 18 : 2 n-6 19.9 Æ 2.1 20.0 Æ 2.3 18.6 Æ 2.2 19.4 Æ 4.5 17.4 Æ 2.1 18 : 3 n-6 ND ND ND ND ND 20 : 3 n-6 2.9 Æ 0.6 2.8 Æ 0.4 3.0 Æ 0.4 3.0 Æ 0.5 2.9 Æ 0.6 20 : 4 n-6 12.2 Æ 1.0a 11.3 Æ 1.4a,b 11.5 Æ 1.2a,b 9.6 Æ 1.3c 10.6 Æ 0.9b,c 22 : 4 n-6 0.5 Æ 0.1a 0.4 Æ 0.1b 0.4 Æ 0.1b 0.3 Æ 0.1b 0.3 Æ 0.1b 22 : 5 n-6 0.4 Æ 0.1a 0.2 Æ 0.1a,b 0.3 Æ 0.2a 0.2 Æ 0.1b 0.1 Æ 0.1b Total n-6 36.2 Æ 1.3a 35.0 Æ 1.6a 34.0 Æ 2.0b 32.8 Æ 3.1b,c 31.7 Æ 2.1c 18 : 3 n-3 ND ND ND ND ND 20 : 5 n-3 0.4 Æ 0.2a 0.4 Æ 0.1a 0.4 Æ 0.1a 0.7 Æ 0.2b 0.7 Æ 0.3b 22 : 5 n-3 1.1 Æ 0.2a 0.9 Æ 0.2b 0.8 Æ 0.2b 0.7 Æ 0.2b,c 0.6 Æ 0.1c 22 : 6 n-3 4.7 Æ 0.7a 6.3 Æ 0.9b 6.9 Æ 1.4b 8.1 Æ 1.9c 9.1 Æ 1.0c Total n-3 6.2 Æ 0.9a 7.6 Æ 1.1.b 8.1 Æ 1.3b 9.4 Æ 2.2c 10.4 Æ 1.2c

* % total fatty acids in human milk as DHA Æ s.d., from the groups of women consuming supplements with 0, 0.2, 0.4, 0.9 and 1.3 g DHA per day. ND ˆ not detected, < 0.05% of total fatty acids. Values with different superscripts indicate signi®cant differences (P < 0.05). Increasing breast milk docosahexaenoic acid RA Gibson et al 581 Table 3 Infant erythrocyte phospholipids at 12 weeks post-partum (fatty acids expressed as a % of total fatty acids Æ s.d.)

Fatty acid 0.21 Æ 0.07* 0.35 Æ 0.04 0.46 Æ 0.16 0.86 Æ 0.24 1.13 Æ 0.33 (n ˆ 12) (n ˆ 10) (n ˆ 12) (n ˆ 10) (n ˆ 8)

Total sats 43.7 Æ 0.9 43.8 Æ 0.6 42.9 Æ 1.6 43.6 Æ 0.8 44.2 Æ 1.5 Total mono 16.2 Æ 0.9 16.5 Æ 1.1 17.3 Æ 2.3 16.3 Æ 0.7 16.6 Æ 0.9 18 : 2 n-6 8.5 Æ 1.0 8.4 Æ 0.8 7.5 Æ 0.7 7.9 Æ 2.1 7.2 Æ 0.8 18 : 3 n-6 ND ND ND ND ND 20 : 3 n-6 1.7 Æ 0.2 1.8 Æ 0.2 1.7 Æ 0.2 1.7 Æ 0.2 1.7 Æ 0.2 20 : 4 n-6 15.9 Æ 0.6a 15.1 Æ 1.0a,b 15.2 Æ 0.8b 14.6 Æ 0.7b 14.2 Æ 1.2b 22 : 4 n-6 3.6 Æ 0.5a 3.4 Æ 0.4a 3.4 Æ 0.5a,b 2.9 Æ 0.4b 2.9 Æ 0.5b 22 : 5 n-6 0.9 Æ 0.2a 0.7 Æ 0.1b,c 0.8 Æ 0.2b 0.6 Æ 0.1c 0.6 Æ 0.2b,c Total n-6 31.0 Æ 1.2a 29.7 Æ 0.8b 28.9 Æ 1.2b 28.0 Æ 2.3b,c 26.9 Æ 1.9c 18 : 3 n-3 ND ND ND ND ND 20 : 5 n-3 0.3 Æ 0.1a 0.4 Æ 0.1a,b 0.4 Æ 0.1a,b 0.6 Æ 0.2c 0.5 Æ 0.2b,c 22 : 5 n-3 1.8 Æ 0.3a 1.5 Æ 0.3b 1.5 Æ 0.2b 1.3 Æ 0.2b,c 1.1 Æ 0.2c 22 : 6 n-3 5.8 Æ 0.6a 7.2 Æ 0.8b 7.7 Æ 0.9b 9.0 Æ 1.5c 9.8 Æ 1.2c Total n-3 7.9 Æ 0.7a 9.2 Æ 1.0b 9.6 Æ 0.8b 10.9 Æ 1.9c 11.4 Æ 1.1c

* % total fatty acids in human milk as DHA Æ s.d., from the groups women consuming supplements with 0, 0.2, 0.4, 0.9 and 1.3 g DHA per day. ND ˆ not detected, < 0.05% of total fatty acids. Values with different superscripts indicate signi®cant differences (P < 0.05).

Figure 1 Infant plasma (A) and erythrocyte (B) phospholipid fatty acids plotted as a function of breast milk docosahexaenoic acid (DHA).

Table 4 Visual evoked potential (VEP) acuity measured at 12 and 16 weeks of age (data expressed as mean Æ s.d.)

Parameter 0.21 Æ 0.07 a 0.35 Æ 0.04 0.46 Æ 0.16 0.86 Æ 0.24 1.13 Æ 0.33 (n ˆ 12) (n ˆ 10) (n ˆ 12) (n ˆ 10) (n ˆ 8)

VEP at 12 wks n recorded 12 10 12 10 8 n successful 7 4 4 5 6 Mean acuity 0.77 Æ 0.14 0.83 Æ 0.10 0.80 Æ 0.09 0.84 Æ 0.20 0.87 Æ 0.08 n excluded 5 6 8 5 2 n with no extrapolationb 4 5 6 5 2 n with dif®cult child 1 1 2 0 0 VEP at 16 wks n recorded 12 10 12 10 8 n successful 10 8 6 7 5 Mean acuity 0.72 Æ 0.09 0.69 Æ 0.03 0.70 Æ 0.09 0.69 Æ 0.05 0.78 Æ 0.09 n excluded 2 2 6 3 3 n with no extrapolationb 2 2 6 3 3 n with dif®cult child 0 0 0 0 0 a % total fatty acids in human milk as DHA Æ s.d., from the groups of women consuming supplements with 0, 0.2, 0.4, 0.9 and 1.3 g DHA per day. b reasons for exclusion included: only 2 points (check sizes) being available for extrapolation, a horizontal line or a non-signi®cant regression line. Increasing breast milk docosahexaenoic acid RA Gibson et al 582 Table 5 Associations between Bayley's MDI and environmental factors

1 Year, n ˆ 51 2 Year, n ˆ 49

Parameter r P r P

Gendera 0.04 0.77 70.27 0.06 Gestational age 0.10 0.48 0.15 0.30 Apgar at 5 minsa 70.02 0.87 70.10 0.50 Length of breast feeding 0.33 0.02 70.04 0.81 Birth weight 70.05 0.75 70.09 0.55 Birth head circumference 0.08 0.59 70.03 0.82 Maternal prestige scalea 70.12 0.39 70.07 0.61 Maternal level of educationa 0.16 0.26 0.05 0.72 Partner prestige scalea 70.27 0.06 70.42 0.00 Partner level of educationa 0.28 0.05 0.28 0.06 Siblings at testa 0.02 0.88 70.26 0.11 Home stimulation at test 0.34 0.02 0.30 0.04 Maternal smokinga 70.06 0.65 0.04 0.78 Partner smokinga 70.35 0.01 70.07 0.65 12 week breast milk DHA 0.29 0.04 0.09 0.55 12 week erythrocyte DHA 0.32 0.02 70.03 0.85 12 week plasma DHA 0.28 0.05 0.08 0.58

a Spearman correlation coef®cients. Figure 2 P100 latency of the visual evoked (VEP) plotted as a function of check size at 12 and 16 weeks of age. showed that home stimulation (partial r2 ˆ 0.11; adjusted r2 ˆ 0.09, P < 0.05) and erythrocyte DHA (partial data included dif®cult behaviour (4 infants at 12 weeks r2 ˆ 0.10; adjusted r2 ˆ 0.09, P < 0.05) were the only only) or extrapolations not meeting our acceptance criteria. signi®cant predictors (r2 ˆ 0.22; adjusted r2 ˆ 0.18, We could detect no difference in VEP acuity between the P < 0.005). By 2 y of age the model only included dietary DHA groups at either 12 or 16 weeks of age gender plus the social score of the partner as predictors although numbers were limited in each treatment group. of Bayley's MDI (r2 ˆ 0.25; adjusted r2 ˆ 0.22, P < 0.005). P100 latency of all groups improved with age and increas- PDI was similar in infants between dietary groups at ing check size (Figure 2). VEP acuity also improved with both ages. There were no associations with any socio- age (0.83 Æ 0.13 logMAR at 12 weeks vs 0.73 Æ 0.09 log demographic variables at 1 y. The only association at 2 y MAR at 16 weeks, P < 0.01, n ˆ 19). was between PDI and the level of education of the partner There was no association between VEP acuity and the (r2 ˆ 0.10; adjusted r2 ˆ 0.08, P < 0.05). level of DHA in breast milk, infant plasma or erythrocytes. There was also no association between any socio-demo- Discussion graphic variables and VEP acuity. This study is unique because infants were fed breast milk with a range of DHA levels achieved by supplementing the Bayley's MDI and PDI diet of the mother with an oil that contained DHA with no On average, all infants scored approximately one s.d. above other PUFA. The study design had an advantage over the expected mean for MDI at one (116 Æ 12) and two directly supplementing the infants because the mother (115 Æ 17) y of age while PDI scores were similar to the was a biological ®lter that provided DHA in the form of expected mean of 100 (98 Æ 19 at 1 y; 100 Æ 17 at 2 y). breast milk fats rather than fats from algae. Furthermore, Bayley MDI at 1 y was found to correlate with DHA supplementing the maternal diet with only DHA did not indices of the infant's diet and status, although no associa- cause perturbations to other breast milk PUFA (Makrides et tions were found at two years (Table 5). Length of breast al, 1996). This enabled us to investigate the relationship feeding was also correlated with Bayley's MDI at 1 y but between breast milk and infant DHA status and between not at 2 y of age. The association between length of breast infant fatty acids and neural indices of a homogeneous feeding and Bayley's MDI, at 1 y, was not due simply to group of exclusively breast fed infants. the fact that those in the highest DHA intake groups breast Our data demonstrate that the level of DHA in infant fed longest. Length of breast feeding was collinear with plasma and erythrocytes is related to the level of DHA in indices of social status, education and home stimulation. breast milk in a curvilinear manner. The curve reaches a The social score and level of education of the partner as plateau when breast milk contains about 0.8% of total fatty well as the amount of stimulation children received in the acids (Figure 1). Our data provide the ®rst dose-tissue home were consistent predictors of Bayley's MDI at both response curves for dietary DHA in human infants and ages. Whether the partner smoked was also related to indicate that incorporation of DHA is saturable. A similar Bayley's MDI at 1 y but not 2 y of age (Table 5). curvilinear relationship between dietary n-3 LCPUFA and In post hoc analysis we investigated the relationship status of DHA in heart, liver and kidney has been reported between 1 and 2 y Bayley's MDI and DHA status at the in suckling mice (Huang et al, 1992). end of the supplementation period in a stepwise linear The increase in blood phospholipid DHA matched a regression model that also included gender, gestational reduction in total n-6 PUFA more closely than AA alone. age, apgar score at 5 mins, birth weight, birth head cir- Furthermore, the reduction in plasma and erythrocyte AA cumference, socio-economic status and level of education did not appear to be due to a direct exchange between AA of both parents, exposure to cigarette smoke, an index of and DHA such as has been reported for dietary EPA home stimulation, duration of breast feeding and number of (Gibson et al, 1992). EPA itself was not present in the siblings as independent variables. At 1 y, the model dietary supplement given to mothers and supplementation Increasing breast milk docosahexaenoic acid RA Gibson et al 583 caused no increase in the level of EPA normally found in Conclusion breast milk (Makrides et al, 1996). Furthermore, because This study has characterised the plasma and erythrocyte EPA remained below 1% of total fatty acids in all infants it fatty acid response of exclusively breast fed infants fed a suggests that little or no retroconversion of DHA to EPA range of DHA levels that cover all reported ranges (Innis, occurred in these infants. Our results are therefore easier to 1992). From our secondary end points there was no interpret than earlier studies in which ®sh oils rich in both evidence that such an intervention affected the develop- EPA and DHA were fed, because any reduction in n-6 ment of VEP acuity, while there was some suggestion that PUFA in the current study can be attributed to increased early DHA may be positively related with Bayley's MDI at incorporation of DHA alone (Henderson et al, 1992). 1 y of age. It is, however, likely that the effect of a single A secondary aim of this study was to address the dietary factor will be small compared with social and assumption that the level of DHA in breast milk was environmental in¯uences in infants that are healthy and limiting with regard to development of neural function. born at term. Large studies will be necessary to delineate Our results indicate that there was no relationship between such subtle effects. VEP acuity and dietary grouping or the level of DHA in breast milk fed to individual infants. However, the large number of exclusions from the VEP acuity determinations reduced the power of the study raising the possibility of a AcknowledgementsÐWe thank J Osmond, M Wooden, C Blank, B Paine, B Jeffrey, D Dixon and E Zielinski for administrative and technical type II error. The number of exclusions was greater than assistance, and acknowledge helpful discussions with Dr K Simmer and expected at 12 weeks of age which may have been due to Dr D Kyle. the fact that some infants may not have developed binocu- larity at this age so that some children may have had dif®culty focusing on the monitor (Skoczenski & Norcia, 1994). Despite this, we were able to demonstrate all anticipated maturational changes in the VEP. For example, References P100 latency improved with check size and age (Figure 2) Agostoni C, Trojan S, BelluÁ R, Riva E & Giovannini M (1995): and VEP acuity also improved with age (Table 4). These Neurodevelopmental quotient of healthy term infants at 4 months and maturational changes were similar to reported reference feeding practice: The role of long-chain polyunsaturated fatty acids. ranges (McCulloch & Skarf, 1991; Sokol, 1978). Pediatr. Res. 38, 262±266. Auestad N, Montalto MB, Hall RT, Fitzgerald KM, Wheeler RE, Connor Our results suggest that increasing DHA in breast milk WE, Neuringer M, Connor SL, Taylor JA & Hartmann EE (1997): during the ®rst 12 weeks of life has a small but signi®cant Visual acuity, erythrocyte fatty acid composition, and growth in term effect on MDI scores measured at 1 y but this effect was infants fed formulas with long chain polyunsaturated fatty acids for one not evident at 2 y. It is important to note however that our year. Pediatr. Res. 41, 1±10. study had limited power. Such ®ndings provide interesting Bayley N (1969): Manual for the Bayley Scales of Infant Development. USA: The Psychological Corporation. pilot data and are an impetus for much larger studies with Birch DG, Birch EE, Hoffman DR & Uauy RD (1992a): Retinal devel- increased power. The data did demonstrate that indices of opment in very-low-birth-weight infants fed diets differing in omega-3 socio-economic status, parental education and the degree of fatty acids. Invest. Ophthalmol. Vis. Sci. 33, 2365±2376. stimulation the child receives in the home were all con- Birch EE, Birch DG, Hoffman DR & Uauy R (1992b): Dietary essential fatty acid supply and visual acuity development. Invest. Ophthalmol. sistently related to performance on the mental scale of the Vis. Sci. 32, 3242±3253. Bayley's test at both ages. These data underscore the Bligh EG & Dyer WJ (1959): A rapid method of total lipid extraction and powerful nature of such environmental factors in affecting puri®cation. Can. J. Biochem. Physiol. 37, 911±917. development even in our relatively advantaged sample of British Nutrition Foundation Task Force (1992): Recommendations for children who, on average, had parents who completed high intakes of unsaturated fatty acids. In Unsaturated Fatty Acids: Nutri- tional and physiological signi®cance. The report of the British Nutrition school, had professional/non-manual jobs and provided Foundation Task Force, ed. The British Nutrition Foundation, pp 152± stimulating home environments (Table 1). 163. London: Chapman & Hall. It is important to note that the study was not designed to Broekhuyse RM (1974): Improved lipid extraction of erythrocytes. Clin. test whether preformed DHA is a conditionally essential Chim. Acta 51, 341±343. Carlson SE, Werkman SH, Rhodes PG & Tolley EA (1993): Visual-acuity nutrient for term infants since our study did not include a development in healthy preterm infants: effect of marine-oil supple- formula (no DHA) reference group. All the infants in this mentation. Am. J. Clin. Nutr. 58, 35±42. study had a minimum mean erythrocyte phospholipid DHA Choue RW, Park HS, Hong JY & Chung KS (1995): The effects of DHA level of 5.8 Æ 0.6% total fatty acids. In our earlier cross supplementation in maternal diet on fatty acid composition of plasma sectional study of breast and formula fed infants where we lipid and human milk. p 351 (Abstract). Coons C, Gay E, Fandal A & Frankenburg W (1981): The Home correlated VEP acuity with erythrocyte phospholipid DHA, Screening Questionaire Reference Manual. Denver, Colorado: JF Ken- the range of erythrocyte DHA values was between 2 and nedy Child Development Centre. University of Colorado Health 7% total fatty acids (Makrides et al, 1993). Randomised Science Centre. studies comparing infants fed placebo (no DHA) formulas Daniel A (1983): Power, Privilege and Prestige: Occupations in Australia. Melbourne: Longman-Cheshire. with those fed DHA supplemented formulas have demon- Gibson RA, Neumann MA, James MJ, Hawkes JS, Hall C & Cleland LG strated that it is possible to improve erythrocyte DHA (1992): Effect of n-3 and n-6 dietary fats on the lipoxygenase products levels but have reported con¯icting results (Makrides et from stimulated rat neutrophils. Prostaglandins Leukot. Essent. Fatty al, 1995; Agostoni et al, 1995; Auestad et al, 1997). These Acids 46, 87±91. studies require careful interpretation since each study has Harris WS, Connor WE & Lindsey S (1984): Will dietary omega-3 fatty acids change the composition of human milk?. Am. J. Clin. Nutr. 40, compared a mixture of LCPUFA, from different sources, in 780±785. a variety of background formula fat blends. Although the Henderson RA, Jensen RG, Lammi-Keefe CJ, Ferris AM & Dardick KR level of DHA supplementation in the current study was (1992): Effect of ®sh oil on the fatty acid composition of human milk high, it remains unique because we were able to vary a and maternal and infant erythrocytes. Lipids 27, 863±869. Holman RT, Johnson SB & Ogburn PL (1991): De®ciency of essential single fatty acid, namely DHA, in the breast fed infant's fatty acids and membrane ¯uidity during pregnancy and lactation. Proc. diet. Natl. Acad. Sci. USA 88, 4835±4839. Increasing breast milk docosahexaenoic acid RA Gibson et al 584 Huang YS, Wainwright PE, Redden PR, Mills DE, Bulman-Fleming B & Makrides M, Neumann MA & Gibson RA (1996): Effect of maternal Horrobin DF (1992): Effect of maternal dietary fats with variable n-3/n- docosahexaenoic acid (DHA) supplementation on breast milk composi- 6 ratios on tissue fatty acid composition in suckling mice. Lipids 27, tion. Eur. J. Clin. Nutr. 50, 352±357. 104±110. McCulloch DL & Skarf B (1991): Development of the human visual Innis SM (1992): Human milk and formula fatty acids. J. Pediatr. 120, system: monocular and binocular pattern VEP latency. Invest. Ophthal- S56±S61. mol. Vis. Sci. 32, 2372±2381. Kneebone GM, Kneebone R & Gibson RA (1985): Fatty acid composition Morrow-Tlucak TM, Haude RH & Ernhart CB (1988): Breastfeeding and of breast milk from three racial groups from Penang, Malaysia. Am. J. cognitive development in the ®rst 2 years of life. Social Sci. Med. 26, Clin. Nutr. 41, 765±769. 635±639. Lucas A, Morley R, Cole TJ, Lister G & Leeson-Payne C (1992): Breast Nettleton JA (1993): Are n-3 fatty acids essential nutrients for fetal and milk and subsequent intelligence quotient in children born preterm. infant development?. J. Am. Diet. Assoc. 93, 58±64. Lancet 339, 261±264. Rogan WJ & Gladen BC (1993): Breast-feeding and cognitive develop- Makrides M, Simmer K, Goggin M & Gibson RA (1993): Erythrocyte ment. Early Hum. Dev. 31, 181±193. docosahexaenoic acid correlates with the visual response of healthy, Skoczenski AM & Norcia AM (1994): VEP measurements of form term infants. Pediatr. Res. 33, 425±427. discrimination in human infants. Invest. Ophthalmol. Vis. Sci. 35, Makrides M, Neumann M, Simmer K, Pater J & Gibson R (1995): Are p 2028 (Abstract). long-chain polyunsaturated fatty acids essential nutrients in infancy?. Sokol S (1978): Measurement of infant visual acuity from pattern reversal Lancet 345, 1463±1468. evoked potentials. Vision Res. 18, 33±39.