点击显示 收起
From the Departments of Human Nutritional Sciences (HAW, SCF-W, JMS, DEF, URM, RRV, and HRK), Pediatrics and Child Health (HAW and MMS), and Obstetrics and Gynecology (MMS), University of Manitoba, Winnipeg, Canada
2 Supported by grants from the Canadian Institutes of Health Research and the Manitoba Health Research Council; by a New Investigator Salary Award from the Canadian Institutes of Health Research (to HAW); by a postdoctoral fellowship (to URM); and by a graduate scholarship from the Manitoba Institute of Child Health (to HRK). The densitometer used was purchased and maintained by the Manitoba Institute of Child Health. 3 Address reprint requests to HA Weiler, School of Dietetics and Human Nutrition, McGill University, Macdonald Campus, 21, 111 Lakeshore Road, Ste-Anne de Bellevue, QC H9X 3V9, Canada. E-mail: hope.weiler{at}mcgill.ca.
ABSTRACT
Background: Evidence-based practice guidelines for aggressive nutritional intervention by using parenteral amino acids (AAs) and minimal enteral feeding (MEF) as early as the first day of life have not been tested for benefits to bone mass.
Objective: We investigated whether early introduction of parenteral AAs and MEF improves growth and bone mass achieved by term age in infants born prematurely.
Design: Twenty-seven infants who were 1200 g and 32 wk gestation at birth were randomly assigned by using a 2 x 2 design to treatment of either 1 g AAs/kg within the first 24 h or 12 mL MEF · kg1 · d1 within the first 72 h of life. Nutrition and growth were documented during hospitalization, and bone mineral content (BMC) of lumbar spine 14, femur, and whole body was measured at term age. Biomarkers of bone metabolism were measured at weeks 1, 3, and 5 and at discharge. Statistical analysis was conducted by using 2 x 2 analysis of variance for intent to treat and for infants receiving protocol nutrition.
Results: Over the first 14 d of life, a main effect of early AAs elevated total intake of protein, and a main effect of early MEF was a higher frequency of MEF volumes exceeding >12 mL · kg1 · d1. Main effects on growth were not evident. An interaction effect was observed for osteocalcin whereby early AAs or MEF alone elevated osteocalcin. A main effect of early MEF yielded higher BMC of spine and femur.
Conclusion: Early aggressive nutrition support with MEF enhances BMC in premature infants, but early MEF or AAs do not improve growth.
Key Words: Minimal enteral feeds parenteral amino acids aggressive nutrition therapy bone mineral content very-low-birth-weight infant premature infant
INTRODUCTION
Growth failure (13) and osteopenia (4, 5) continue to affect infants born prematurely and of very low birth weight (VLBW). By estimated term date, infants with VLBW typically have z scores > 2 SDs below the mean for growth (1, 2) and for whole-body (4, 5) and spine (6) bone mineral content (BMC). Some catch-up in growth and BMC appears to occur by age 2 y (6), but longitudinal data suggest that both remain low throughout childhood (7, 8) and to young adulthood (9).
Because of the longstanding existence of growth failure and osteopenia, many nutrition intervention studies have been conducted during the past 30 y. Intervention studies during the neonatal stay have tested for benefits of mineral supplementation of human milk (10), mother's milk fortifiers (11), and nutrient-enriched formula (12). All have shown some degree of success, particularly in the later study, in which whole-body BMC was elevated by 35% in the nutrient-enriched formula group compared with the control formula group at estimated term age (12). Although that study seems to be the most successful attempt to mimic intrauterine accretion of bone mineral, the intervention is not consistent with the recommendation to feed infants their own mother's milk (13), even if born prematurely (14). More nutrition intervention studies are needed to enhance BMC in the infant with VLBW fed their own mother's milk.
Early aggressive nutrition for preterm infants is practiced in many neonatal intensive care units and includes energy and amino acids (AAs) by parenteral nutrition (PN) plus enteral nutrition (EN) in the form of minimal enteral feeding (MEFs) within the first day of life (1518). The evidence behind these guidelines (1518) includes better nitrogen retention in infants provided with parenteral AAs within the first day of life than in infants provided with AAs later in the first 23 d of life (1922). Likewise, MEF within 35 d of life leads to earlier attainment of full feeds, nipple feeds, and discharge from the hospital (23). Introduction as early as 1 d also leads to earlier achievement of full feeds in infants < 1250 g at birth (24). In pioneering studies, MEF (day 4 compared with day 14) improved retention of calcium and other minerals, but enhanced radial bone mass, measured by using single-photon absorptiometry, was not evident (25).
Combined parenteral AAs and MEF when initiated 48 h of birth is associated with less weight loss, more rapid recovery of birth weight, earlier attainment of full feeds, and reduced hospital stay (2628). In addition, weight is higher at estimated term age, although it remains <5th percentile (28). Although 2 studies (28, 29) included assessment of the incidence of osteopenia, no study has quantified the benefits of the aggressive early nutrition support on both growth and BMC. Therefore, the objective of this study was to investigate the effects of early introduction of parenteral AAs and MEF on growth and bone mass achieved by estimated term age in infants born with VLBW.
SUBJECTS AND METHODS
Preterm infants born at or between 24 and 32 wk gestational age (GA) and who were an appropriate weight for GA [as defined by >3rd percentile by using the 2001 Canadian size at birth standards (30)] with a weight 1200 g were eligible for the study; larger infants were not studied because it was felt that PN would not be required as long and the MEF would possibly delay progress to full EN. Infants were excluded if they had known congenital abnormalities, required any surgery, or were not likely to remain in the province for the hospital stay. Infants born of a mother who used alcohol or illicit drugs during pregnancy, had diabetes or other endocrine diseases known to affect bone metabolism, or took medications known to affect bone metabolism were also excluded from the study.
Infants were recruited from Winnipeg's Health Sciences Centre and St Boniface General Hospital between November 2001 and November 2003. No monetary remuneration was provided to the families; however, in efforts to support and encourage provision of breast milk to infants, mothers were provided with an electric breast pump and double pump kit for the duration of their infant's hospitalization.
During the 2-y recruitment period, 60 parents of infants who met inclusion criteria were approached within the first 24 h of the infant's delivery; of these parents, 25 declined participation. Three other infants met entry criteria during this period, but either the parent was not conscious (1) or a language barrier prevented ethical consent (2). Additionally, 4 other infants met entry criteria, but the time between delivery and first contact was too long (20 h) to obtain ethical and informed consent before 24 h of life. In total, 35 infants were recruited to the study, of which 1 declined after 24 h. Of the 34 participating infants, 7 were from multiple births, but only the firstborn was studied. Within the first 10 d of life, 7 infants died because of prematurity and did not receive the intervention. The remaining 27 infants survived beyond term age and are reported on herein.
Written informed consent was obtained from the children's parents before enrollment in the study. This study was reviewed by the University of Manitoba Ethics Committee, and ethical approval was obtained.
Medical and nutritional management
The interventions were designed in collaboration with the neonatal health care team. Of note, 5 neonatologists provided care to both neonatal intensive care units, making this two-site study a function of location rather than medical management philosophy and practice. The interventions were commencement of 1 g AA/kg parenterally within 24 h of life (AA group); MEF 12 mL · kg1 · d1 commencing within the first 72 h of life (MEF group); or both combined (AA + MEF group) by using a 2 x 2 design. The control group was represented by the standard of care that at the time of the study was defined by introduction of AAs parenterally on the second day of life after 1800 and to begin MEF after 72 h of life. Infants were randomly assigned to 1 of 4 study groups by using precoded envelopes stratified by gestational age by using 2-wk intervals beginning at 24 wk GA. The interventions were not blinded because use of a placebo for MEF such as water has been shown to alter gastric motility (31) and may have confounded the results.
Parenteral nutrition
All infants received 6080 mL fluid · kg1 · d1 on day 1, and that amount was increased to 150160 mL · kg1 · d1 within the first week. AminosynPF (Ross-Abbott, Saint-Laurent, Canada) was used in both neonatal intensive care units. A 30-mL/kg solution was designed by the pharmacy to provide 1 g AA/kg along with 0.5 mmol Ca/kg in a 10% dextrose solution and was made available at all times during the study. This solution was hung within the first 24 h of life for infants in the early AA group intervention. The addition of the calcium and dextrose was to standardize these nutrients across the groups because calcium was routinely administered within the first 24 h. On the second day of life, the PN orders were placed by using the following standard of care: carbohydrate was initiated at 4.55.5 mg · kg1 · min1 to a maximum of 1012 mg · kg1 · min1, AA began at 1 g · kg1 · d1 and increased to 3 g · kg1 · d1 by 1-g increments, and lipid was initiated on the third day of life at 1 g/kg and increased 1 g · kg1 · d1 to 3 g · kg1 · d1 as tolerated by using a 20% solution (Intralipid; Fresenius Kabi AB, Uppsala, Sweden). Lipid was discontinued when the infant received 50% of the volume of full enteral feeds. Mineral and vitamin supplementation was managed at the discretion of the health care team.
Enteral nutrition
The intervention MEF was 12 mL · kg1 · d1 given as 1 mL/kg every 2 h. The first milk was mother's milk unless none was available or the mother elected to have her infant fed formula. In those situations, a 68 kcal/100 mL standard formula (Enfalac Premature Formula; MeadJohnson, Ottawa, Canada) was used for priming until mother's milk was available or until the infant could be fed a higher-density formula. The MEF was to be delivered over 35 d before advancing volumes for nourishment purposes. Nurses were asked to not aspirate the MEF and to only interrupt MEF on physician request on the basis of abdominal distension, bloody stools, bilious aspirates, or abnormal abdominal X-ray. Nourishment feeds were advanced by 30 mL · kg1 · d1, and human milk fortifier (MeadJohnson) was added or the formula was changed to a higher density (81 kcal/100 mL) when the feed volume reached 150 mL · kg1 · d1 for 24 h.
Aside from the timing for introducing the parenteral AAs and MEF, the nutritional and medical management of all infants was not altered at any point during the course of the study. Provision of breast milk or formula type was determined by the parents and the availability of breast milk. Vitamin and mineral supplementation was managed by the health care team and included folate; vitamins A, C, and D; and iron on the basis of total EN intake. Data on feeding type and volume and other medical information were collected from hospital charts of the preterm infants. Nutrient intake from PN and EN was recorded daily until full EN (150 mL · kg1 · d1 or 100 kcal · kg1 · d1) was established, after which intakes were recorded twice weekly. Medications administered to infants, including vitamin and mineral supplements, were recorded daily for the full duration of hospitalization. Duration of ventilation and oxygen support was recorded, and bronchopulmonary dysplasia (BPD) was defined as oxygen requirements continuing beyond 36 wk GA (32).
Growth
Weight and GA at birth (based on ultrasound dating between 8 and 13 wk) were obtained from the medical record. Weight, length, and head circumference were measured at the end of each week while infants were hospitalized. Weight was measured to the nearest gram (SB32000; Mettler Toledo, Hightstown, NJ). Infant crown-heel length was measured to the nearest millimeter by 2 examiners by using a Plexiglas recumbent preemie, newborn, or pediatric stadiometer (Ellard Instruments, Inc, Seattle, WA). Head circumference was measured to the nearest millimeter by using a nonstretch tape measure placed at the most prominent part above the supraorbital ridges and over the part of the occiput that gave the maximum circumference.
Dual-energy X-ray absorptiometry
At term age, the BMCs of infant whole body, whole left femur, and lumbar spine 14 were measured by using software version 11.2 and a Hologic 4500 Acclaim Series Elite dual-energy X-ray absorptiometer (Hologic Inc, Bedford, MA). Infants were clothed in gowns without snaps, zippers, or buttons and were swaddled in a thin cotton blanket to minimize movement. All infants were scanned with clean dry diapers and while sleeping. Scans were analyzed by a single investigator (JS). The CV for quality control measurements of BMC by using a spine phantom (Phantom no. 8832; Hologic) over the course of the study was <1.0%.
Sample collection and biochemistry
While the infants were in the hospital, 4 blood samples (500 µL) were obtained from preterm infants on weeks 1, 3, and 5 of life and at the time of discharge by using a heel prick and heparin-coated microtainers. Blood samples (with the exception of cord blood) were collected between 0400 and 0900 to control for diurnal variations in the biomarkers of bone metabolism. Blood samples were centrifuged at 2000 x g at 4 °C for 10 min, and plasma was removed and stored under nitrogen at 80 °C until analysis.
Infant urine samples were obtained between 0200 and 0900 to control for diurnal rhythms. At birth, urine samples were obtained within the first 48 h and then on days 7, 21, and 35 after birth. Urine was collected by using a pediatric urine collection bag that was placed in the infant's diaper. All urine samples were stored at 80 °C until analysis.
Plasma 25-hydroxyvitamin D was measured by using an equilibrium radioimmunoassay (DiaSorin, Stillwater, MN) to ensure that potential differences in bone mass were not due to vitamin D status. CV 20% or a difference of 500 cpm between duplicates was considered acceptable precision. This assay had a minimum detectable limit of 10 nmol/L. Plasma osteocalcin was measured by using radioimmunoassay (DiaSorin) as an indicator of osteoblast activity. This assay had a minimum detectable limit of 0.04 nmol/L. Urinary N-telopeptide (NTx) was determined by enzyme-linked immunosorbent assay (Osteomark, Seattle, WA) and was used as a marker of osteoclast activity. Infant urine samples were diluted 10-fold with deionized water. The CV for reproducibility was 20% for all sample duplicates. This assay had a minimum detectable limit of 20 nmol/L bone collage equivalents. Because the urine samples were spot samples, all NTx values were corrected to creatinine. Creatinine was measured by using an adapted microassay based on the Jaffe reaction (Sigma Diagnostics, Inc, St. Louis, MO). The CV for reproducibility was 10% and an assay minimum detectable limit of 44.2 µmol/L. Calcium and phosphorous were measured in urine (0.25 mL) combined with 0.5 mL concentrated HNO3 (trace metal grade; Fisher Scientific, Ottawa, Canada) and allowed to dissolve overnight in glass test tubes. Once dissolved, deionized water was added to the test tubes to reach a concentration of 5% HNO3. Dissolved and diluted urine was then transferred to a scintillation vial and analyzed by emission spectrometry (Varian Liberty 200 ICP; Varian Canada, Mississauga, Canada). The minimum detectable limit for calcium was 2.5 x 104 mmol/L, and for phosphorous it was 6.5 x 104 mmol/L. All values were corrected to creatinine just as for NTx.
Sample size and statistical analysis
Because there is no study of the benefits of such early combined nutrition intervention on BMC and because many studies do not show improved growth at term age, the sample estimate was based on a 30% elevation of BMC values measured in 7 infants (27 wk GA, 950 g at birth) participating in a descriptive study at our facility 6 mo before inception of this study. The 30% elevation was used because Schanler et al (25) observed a 33% elevation in calcium retention with earlier MEF and is reasonable because enriched formula lead to a 35% elevation in whole-body BMC than standard formula (12). On the basis of a difference of 0.225 g with a SD of 0.216 for lumbar spine, the estimate was 11 per group; based on a difference of 0.343 g with SD of 0.327 for femur, the estimate was 15 per group; and for whole body a difference of 13.178 g with SD of 12.674 gave 16 as the estimate. The total sample size of 34 (ie, n = 17 per main effect) was used to account for protocol violations.
The data were analyzed in terms of both intent-to-treat (ITT) numbers and treatment according to protocol by using SAS for WINDOWS software (version 8; SAS Institute Inc, Cary, NC). Potential reasons for not adhering to protocol included either medical instability that prevented the intervention from being implemented or protocol violations. No differences were observed in outcomes between the 2 neonatal units; thus, the data were combined for analysis. Main effects of the interventions on bone outcomes were identified by using a 2 x 2 analysis of variance (ANOVA) with P < 0.05 accepted as significant. For repeated measurements of the biomarkers of bone metabolism, a repeated-measures 2 x 2 ANOVA was used with post hoc testing by using Bonferroni's multiple comparisons test where appropriate. Data are expressed as weighted means ± SDs unless otherwise indicated. Weighted means were selected for presentation because the sample size behind the mean is considered in calculations of main effects. This is important because the sample size was unequal among the treatment groups, particularly in the protocol analysis.
RESULTS
By design, GA and weight at birth did not differ significantly among groups (Table 1). Infants from multiple births were equally represented among the groups, as was sex (Table 1). All infants had respiratory distress syndrome at birth and required ventilation and oxygen therapy, except one infant in the AA group who did not have respiratory distress syndrome. Apgar scores at 5 min were 7 ± 2 for each group. On average, early introduction of parenteral AAs occurred at 0.9 d and early MEF occurred at 3.2 d of life in their respective groups. However, not all infants received protocol nutrition. Reasons for infants to be excluded from the protocol analysis were as follows: 2 infants in the standard-of-care group were not studied because 1 infant received AAs and MEF too early (ie, within first 24 h of life) and the other was too critical to receive the standard of care, and 3 infants in the early MEF group and 2 infants in the early AA + MEF group did not receive the protocol intervention because of medical instability.
View this table:
TABLE 1. Characteristics of infants at birth and anthropometric measurements in infants randomly assigned to receive early parenteral amino acids (AA) or early minimal enteral feeds (MEF) reported as intent-to-treat (ITT) or according to protocol treatment by using a 2 x 2 design1
After removal of the protocol violations, a main effect of AAs in the infants in the protocol group resulted in receipt of AAs significantly earlier than in the standard-of-care control group (Table 2). A main effect of AAs was observed to also result in higher intake of proteins during the first 14 d (1.9 ± 0.2 compared with 2.3 ± 0.4 g/kg; P = 0.013). A main effect of MEF in the protocol group resulted in significantly earlier introduction of MEF than in the standard-of-care control group (Table 2) and resulted in a higher number of days that EN volume exceeded 12 mL/kg (10.6 ± 2.3 compared with 6.6 ± 2.7 d; P = 0.008) during the first 14 d of life. Overall, the average postnatal age at which 100 kcal/kg from EN and PN was achieved did not differ significantly among the respective groups (Table 2) and remained stable thereafter (data not shown). In the protocol analysis but not in the ITT analysis, the number of days without oral intake was reduced by both interventions, whether alone or combined (Table 2). Calcium intake after full EN feeds were established in all infants did not differ significantly between the groups regardless of ITT or protocol analysis (Table 2). The proportion of infants fed mother's milk, at >80% of feeds, from birth to discharge also did not differ significantly between groups, regardless of ITT or protocol nutrition (Table 2).
View this table:
TABLE 2. Nutritional and clinical characteristics of infants randomly assigned to receive early parenteral amino acids (AA) or early minimal enteral feeds (MEF) reported as intent-to-treat (ITT) or according to protocol treatment by using a 2 x 2 design1
During the course of the hospitalization, no significant differences were observed among groups in the duration of PN, duration of oxygen therapy, the incidence of BPD or use of dexamethasone, or duration of hospitalization regardless of ITT or protocol management (Table 2). However, under ITT analysis only, a main effect of MEF resulted in a higher proportion of infants having proven sepsis by way of positive culture (Table 2). No necrotizing enterocolitis (NEC) was observed.
For all outcomes except biochemistry, no interactions between the interventions were identified by using 2 x 2 ANOVA. This is not surprising, because the power to detect a significant interaction was low. For example, for the primary BMC outcomes, the power ranged from 0.126 for femur to 0.45 for spine. Regarding growth in response to the nutritional interventions, the days to recover birth weight did not vary as a function of either intervention (Table 1). In addition, weight, length, and head circumference did not differ significantly among the groups at any time in the study, yielding similar size at the term age assessment (Table 1).
Because there was no difference in body size, BMC values were expressed only as absolute values. In ITT analysis, a main effect of MEF resulted in significantly higher BMC in the spine (1.057 ± 0.281 compared with 1.259 ± 0.290 g; P = 0.014) and femur (1.252 ± 0.389 compared with 1.595 ± 0.415 g; P = 0.015) but not whole body (45.905 ± 11.821 compared with 50.298 ± 13.404 g; P = 0.31). In ITT analysis, a main effect of AAs resulted in lower femur BMC (1.566 ± 0.232 compared with 1.257 ± 0.538 g; P = 0.029) but not lower spine or whole-body BMC. After protocol analysis, the main effects of early MEF on BMC did not differ significantly from that found with the ITT analysis, with 16.7% higher BMC in the spines and 37.1% higher BMC in the femurs of the early compared with standard MEF groups (Figure 1). Because the sample size in the protocol analysis was small with low power to detect differences among groups, the 95% CI was calculated for the difference in BMC observed because of a main effect of MEF. The difference was estimated to range from 0.042 to 0.298 g for spine BMC, from 0.246 to 0.624 g for femur BMC, and from 1.292 to 10.894 g for whole-body BMC. The negative main effect of AAs on femur disappeared after analysis by protocol nutrition.
FIGURE 1.. Weighted mean (± 2SD) bone mineral content (BMC) of lumbar spine 14, femur, and whole body at estimated term age in infants randomly assigned to receive early minimal enteral feeding (MEF) or the standard of care according to the protocol nutrition analysis. n = 8 [n = 4 MEF and n = 4 amino acid (AA) + MEF] infants for the early MEF main effect and n = 12 (n = 5 control and n = 7 AA infants) for the standard MEF. Main effects of AAs and interaction effects between AAs and MEF were not observed. Main effects were identified by using 2 x 2 ANOVA. Data for healthy term infants from the same geographic region are mean (± 2 SD), and the horizontal lines indicate the use of identical methods (33).
No main effects or interaction effects were observed between treatments or with time for plasma 25-hydroxyvitamin D, urinary calcium:creatinine, phosphorus:creatinine or NTx:creatinine (data not shown by intervention groups). An overall effect of time (P < 0.001) was observed on osteocalcin, NTx:creatinine, and calcium:creatinine regardless of ITT or protocol analysis (Table 3: ITT data shown). In addition, an AA x MEF interaction was observed such that osteocalcin was higher in the intervention groups alone than in the combined group (P = 0.01). In the protocol analysis, the same interaction was observed (Figure 2; P = 0.018); data are shown for the protocol nutrition to complement the BMC figures that are based on the protocol nutrition. No interaction effects were observed between the treatments and time for osteocalcin.
View this table:
TABLE 3. Plasma and urinary biomarkers of bone formation in prematurely born infants over time by using intent-to-treat analysis1
FIGURE 2.. Interaction effect of early amino acids (AAs) and early minimal enteral feeding (MEF) on plasma osteocalcin in infants. Mean (±SD) osteocalcin values were analyzed by using two-way repeated-measures ANOVA in the protocol group. An interaction (P < 0.0118) between early AAs and MEF was observed such that either AAs or MEF alone resulted in higher values than when the 2 were combined. No interactions between the interventions with time were observed. Osteocalcin values (pooled) were measured at 12, 7, 21, and 35 d of age plus discharge per diet group. Bars with different superscript letters are significantly different (AA compared with MEF + AA, P = 0.041; MEF compared with MEF + AA, P = 0.017) by using Bonferroni's multiple comparisons test.
DISCUSSION
Osteopenia and growth delay continue to challenge infants born prematurely and with VLBW. This study was undertaken to evaluate aggressive nutrition support as a strategy to prevent osteopenia in the most vulnerable of prematurely born infants. The infants participating in this study represented the most critical at birth, as shown by the high incidence of respiratory distress and the subsequent development of BPD and confirmed sepsis. During the course of the study, growth was similar among treatment groups. This was not surprising because enhanced growth has only been seen in studies that compared early aggressive nutrition with delayed nutrition support (23). Nonetheless, main effects of the MEF intervention resulted in higher BMC of spine and femur by term age. After conducting analysis on only those infants receiving the protocol intervention, the benefits incurred by MEF remained, and no effect of early AAs was observed.
Few studies of MEF or aggressive nutrition support have included assessment of bone. Schanler et al (25) studied slightly larger and more mature infants and intervened with MEF (20 mL/kg) at day 4 compared with day 14. In that study, calcium retention was elevated 33% above the standard-of-care group, but no significant change to bone mass of the radius was observed. However, measurement of radial bone mass is a cross-sectional measurement and is not representative of a whole bone or series of bones in the case of lumbar spine. Two other groups of researchers studied alkaline phosphatase in infants receiving early MEF. In 1988, Dunn et al (29) showed that MEF (1520 mL/kg) beginning at 48 h of life, compared with nothing by mouth for the first 9 d, protected against metabolic bone disease as indicated by lower alkaline phosphatase. Wilson et al (28) conducted a prospective, randomized, controlled trial of aggressive nutrition support in infants who weighed < 1200 g at birth and included both small and appropriate size for GA infants. The basis of the intervention was similar with introduction of AAs (0.5 g/kg) and MEF (12 mL/kg) on the first day of life compared with introduction of AAs later at day 3 and MEF once clinically stable. In Wilson et al (28), the infants in the intervention group had greater morbidity yet consumed more total energy and gained more weight. No differences were observed in the number of infants with osteopenia on the basis of plasma alkaline phosphatase or evidence of bone demineralization on radiography. Defining the success of nutritional interventions by incidence of osteopenia is not likely to fairly show the improvements as a result of nutrition because almost all prematurely born infants are discharged with osteopenia. In fact, most infants in the current study were osteopenic by using the definition of a BMC > 2 SD below the mean of values for healthy term infants in the same geographic region (33). Nonetheless, BMCs of the spine and femur were significantly higher as a main effect of MEF. Collectively, the work by Dunn et al (29) and our study showed that early MEF is beneficial to bone mass.
The volume of MEF used in the present study was only 12 mL · kg1 · d1 on the basis of that reported as safe in infants who weighed < 1200 g (26, 28). Typical volumes of MEF range from 12 to 24 mL · kg1 · d1 (23) and are initiated as early as the first day after birth and in older reports after the second week of life (25). The positive main effects of MEF in our infants likely arose from improved gastrointestinal development and nutrient retention, because no differences were observed in nutrient intake aside from initial volume of milk during the first 14 d of life. The basis for this speculation is that amniotic fluid and mother's milk contain trophic factors for sustained intestinal health and development (34). Restricting the oral intake of infants at birth may not only delay development but may also reduce intestinal function as a result of lack of stimulus. In animals, the minimal proportion of feed volume given enterally that increases intestinal mucosal mass is 30% in neonatal dogs (35) and 40% in piglets (36). In neonatal dogs, only 10% of feed volume by the enteral route is required to enhance intestinal motility (35). The piglet study used elemental nutrients; thus, the trophic factors in human milk may have stimulated intestinal development to an even greater degree (34). In human infants, 3040% of full feeds translates into 3042 kcal/kg or 5060 mL/kg, depending on the definition of full feeds (100105 kcal · kg1 · d1 and 150 mL · kg1 · d1). The infants in the MEF intervention group treated according to the protocol reached 5060 mL/kg 4 d earlier (data not shown) than did infants in the standard-of-care group. Whether such differences in EN could lead to enhanced intestinal development and function in prematurely born infants is unclear, but it is realistic, because, in neonatal dogs, only 45 d is required to observe benefits of small volumes of EN to induce mucosal growth maturation of motor function (35). The result in the current study was improved bone mass as a main effect of MEF over the standard of care. Further research is required to learn whether higher volumes of MEF (12 compared with 24 mL · kg1 · d1) could improve bone mass to a greater degree than was observed in the current study.
Whether MEF protects against NEC is still greatly debated (23) and cannot be addressed with small sample sizes as used in the current study. Infants in our study received MEF early in life (typically between 2 and 6 d), which is thought to protect against intestinal permeability (37). Feeds were advanced slowly (after 68 d; Figure 2) as opposed to aggressive advancement of feeds to 140 mL · kg1 · d1 during the first 10 d, which is associated with an elevated risk of developing NEC (38). Therefore, it is not surprising that no diagnosis of NEC was made. In contrast to NEC, sepsis was commonly diagnosed in the infants in the current study, as is common in such small infants (39). However, the incidence of sepsis was not higher in the infants receiving the MEF or AAs + MEF according to protocol. The small sample of 20 infants receiving the interventions according to protocol was not large enough to truly detect differences among groups with respect to sepsis.
Whether aggressive MEF is suitable for all critically ill infants is not certain, because many of the infants in this study were not fed according to the MEF or the AA + MEF protocol because of medical instability. Thus, the sample size was smaller than desired, and the data could be limited by type I errors. Nevertheless, the outcomes complement existing reports and provide the first assessment of neonatal bone mass in such vulnerable infants. The elevations in bone mass for femur were as high as 36% for femur and 16% for lumbar spine with early MEF. These increments are in line with elevations in bone mass in larger, more mature, and healthy premature infants fed nutrient-enriched formula (12). However, even with these elevations, bone mass was far from that expected had the infants been born at term, especially when lumbar spine is examined (Figure 1). The early parenteral AAs did not confer any benefit to bone mass, but perhaps the initial dosage was too low. Further research is required to fully evaluate the benefits of MEF at 12 mL · kg1 · d1 or at higher volumes with respect to both growth and bone mineralization during the hospital stay anc after discharge from the hospital. Additionally, combining MEF with other strategies to enhance bone mass, such as earlier introduction of mother's milk fortifiers or nutrient-enriched formulations, may support bone mineralization and growth parallel to that achieved by full-term gestation.
ACKNOWLEDGMENTS
Each author was involved in some aspect of the design, whether in relation to conception of the study (HW and MS) or to method development (all authors), and was fully involved in study conduct, data analysis, and the final manuscript. None of the authors had a conflict of interest.
REFERENCES