点击显示 收起
From the Public Health Nutrition Unit, MRC International Nutrition Group, London School of Hygiene and Tropical Medicine, London (AC, SM, and AP); MRC Keneba, the MRC Laboratories, Banjul, The Gambia (AC, SM, and AP); and the Elsie Widdowson Laboratory, MRC Human Nutrition Research, Cambridge, United Kingdom (MO and CC)
2 Supported by the UK Medical Research Council, the Nestlé Foundation, and the Nutricia Research Foundation. 3 Address reprint requests to A Prentice, MRC International Nutrition Group, Public Health Nutrition Unit, London School of Hygiene and Tropical Medicine, Keppel St, London, WC1B 7HT, UK. E-mail: andrew.prentice{at}lshtm.ac.uk.
ABSTRACT
Background: Most human research on leptin has involved well-nourished subjects or clinical states such as anorexia nervosa or cancer cachexia.
Objective: We studied the development of leptin as a monitor of energy status in young African infants whose growth patterns probably reflect the evolutionary norm.
Design: We enrolled a prospective birth cohort of 138 rural Gambian mother-infant pairs. Plasma leptin was analyzed in maternal blood in late pregnancy, in cord blood, and at 8, 16, and 52 wk in the infants. Body mass index (BMI; in kg/m2) was used as a proxy for fatness. The mothers were lean (BMI: 21.6 ± 2.5), and the infants grew poorly compared with Western standards (average weight-for-age z score of –1.9 at 52 wk).
Results: Maternal and cord blood leptin and birth weight were all positively correlated. Throughout infancy, leptin was highly correlated with BMI. A strong sex difference existed at birth (ie, leptin concentrations were significantly higher in females than in males), disappeared at 8 wk, and reappeared at 16 and 52 wk. Absolute leptin concentrations declined by almost 90% from birth to 52 wk, but leptin's ability to discriminate across a range of BMI values improved with age. In early infancy, leptin concentrations were uncorrelated with recent changes in BMI, but, by 52 wk, leptin was able to assess both the size of energy stores and the direction of recent changes.
Conclusions: Leptin concentrations signal energy status from fetal life onward. As infancy progresses, leptin's power to discriminate both chronic and dynamic energy status increases, and this discrimination is achieved at much lower circulating peptide concentrations.
Key Words: Leptin • humans • pregnancy • infancy • adipose tissue • energy flux • Gambia • birth weight • growth
INTRODUCTION
Circulating plasma leptin concentrations are strongly correlated with adipose tissue mass (1, 2) and, in adult humans, are further modulated by the dynamic state of energy balance (3–5). Leptin therefore provides the higher centers of metabolic regulation with information about the status of peripheral energy stores in terms of both their absolute size and their direction of change. Although leptin may not be the only blood-borne signal of energy sufficiency (6), it clearly plays a major role in the regulation of numerous energy-sensitive physiologic functions, including appetite and energy expenditure, reproduction, inflammation and immunity, and organogenesis (2).
Almost all human research into the function of leptin has been conducted in well-nourished Western subjects, most of whom have adipose tissue stores considerably in excess of the global average and even further in excess of the human evolutionary norm. We have argued elsewhere that these differences in adipose tissue stores lead to some significant problems when investigators try to understand the true evolutionary origins of leptin and hence when they predict its normal behavior (7). For instance, the frequently made statement that obese subjects display "leptin resistance" makes the unjustified assumption that leptin has evolved to cope adequately as a metabolic signal in an organism whose fat mass is an order of magnitude greater than the historical norm.
We previously compared leptin concentrations in Gambian and Italian children aged 6–8 y (8). The Gambian children probably reflect the energy status that has been typical of humans over most of their evolution. Their BMI z scores ranged from –4 to 0 against the United Kingdom's 1990 reference standards (9). In contrast, the Italian children had BMIs ranging from 0 to 3. It was noticeable that the relation between leptin concentrations and BMI z scores differed profoundly between the 2 populations. The slope of the lines was very shallow in the Gambian children (ie, at BMI values less than the UK norm) and very steep in the Italian children. This suggests that the hormone will be operating in the 2 populations at very different regions of the dose-response curves that modulate its various physiologic actions (7).
To reach a true understanding of the functions and origins of genes and of their protein products, they must be studied in the context in which they evolved. To this end, we describe here the patterns of plasma leptin in late pregnancy, in cord blood, and in the first year of life in a cohort of rural Gambian mothers and infants living a traditional lifestyle, which is characterized by marginal nutritional conditions and an annual fluctuation in food availability between the hungry and harvest seasons.
SUBJECTS AND METHODS
Study population
This study was conducted in 5 villages in the West Kiang region of The Gambia. Its main purpose was to investigate seasonal influences on the early-life development of immune function. The current analyses used residual plasma samples. Detailed descriptions of environmental and nutritional conditions in this region were provided elsewhere (10, 11). Briefly, this rural subsistence-farming lifestyle is associated with frugal diets that lead to widespread undernutrition as judged by Western norms. The area experiences a profound seasonality revolving around the annual rains (July–October) that coincide with a "hungry season," when weight loss occurs in all adults, including pregnant women, and which is associated with a seasonal reduction in birth weights (10–12). Many other markers of nutritional status also show a marked seasonal variation (13).
Pregnant women (
View this table:
TABLE 1. Descriptive statistics at birth and during infancy1
At 8, 16, and 52 wk, a blood sample was collected from each infant by venisection, and the plasma was stored for analysis. All plasma samples were kept at –80 °C until transportation on dry ice to the United Kingdom for analysis.
Ethical approvals for both the original study and the subsequent analysis of leptin in residual plasma were granted by the joint Gambian Government/Medical Research Council Ethics Committee. Written informed consent was obtained from the families of all infants before recruitment.
Laboratory analysis
Plasma leptin concentrations were measured by enzyme-linked immunosorbent assay at MRC Human Nutrition Research, Cambridge, United Kingdom, with the use of a commercially available kit (Quantikine ELISA; R&D Systems, Minneapolis). Intraassay and interassay CVs for quality control (QC) standards run during each assay series were 46-80 pg/mL = 0.0–11.9% and 7.4–9.8%, respectively, for QC1; 116-170 pg/mL = 0.3–6.3% and 11.0–11.5%, respectively, for QC2; and 496-678 pg/mL = 0.2–6.1% and 10.0–11.9%, respectively, for QC3. The enzyme-linked immunosorbent assay technique has greater sensitivity than does the frequently used radioimmunoassay and was used because of the low leptin concentrations in this population. -1-Antichymotrypsin (ACT) was assayed in duplicate by using a nephelometric assay on a Cobas Bio centrifugal analyzer (Roche, Rotkreuz, Switzerland).
Statistical analysis
Anthropometric z scores were calculated with reference to the UK Reference Standards (15). Values at birth were calculated for the appropriate gestational age. The hungry season was defined as extending from July through October. All remaining months were categorized as the harvest season. Comparisons among group means were made by using two-sample t tests. Leptin concentrations were loge transformed to normalize the distribution before analysis. Associations between leptin concentrations and the measured variables were tested by analysis of variance or analysis of covariance, as appropriate, and multiple regression analysis. P < 0.05 was considered significant for main effects and P < 0.1 was considered significant for interactions. All statistical analyses were performed with the use of DATADESK for WINDOWS software (version 6; DataDesk, Ithaca, NY).
RESULTS
Nutritional status of the population
The women in this study were of average height (159.2 ± 5.5 cm) but low body weight (55.2 ± 6.9 kg), especially considering that these values were obtained in late gestation. Their mean BMI of 21.6 ± 2.5 reflects their leanness. Twenty-three percent had a BMI <20, and only 5% had a BMI >25.
Birth weights were low by Western norms (2867 ± 333 g for boys and 2840 ± 329 g for girls), and they represent z scores of –0.94 for boys and –0.66 for girls (Table 1). Early postnatal growth was excellent when the infants were fully breastfed (and largely free from infections), as shown in Figure 1. By age 8 wk, the infants' weight and BMI z scores were slightly greater than zero, which indicates that the infants had fully caught up to the UK standards. However, after age 8 wk, there was a progressive and severe failure to thrive, which is caused by the introduction of nutritionally inadequate and bacterially contaminated complementary foods. At age 52 wk, the weight-for-age z score was close to –1.9, and the BMI z score was close to –1.5. Weight-for-age growth curves for the girls were 0.5 z scores lower than those for the boys from 8 wk onwards when expressed as sex-specific z scores; this indicated either that the girls were more vulnerable to the environmental stresses than were the boys or that the girls were less well cared for than were the boys.
FIGURE 1.. Mean (±SEM) weight (), length (), head circumference (), and BMI z scores (•) from birth to age 52 wk in boys (n = 75) and girls (n = 63) combined.
Maternal leptin in late gestation
Eighty-two maternal plasma samples were available from late gestation (36 ± 2 wk). Leptin concentrations were strongly skewed (geometric
The sex of the infant was the dominant predictor of cord blood leptin concentrations [boys (n = 55): 3.17 ng/mL; 95% CI: 2.52, 3.98; girls (n = 45): 6.10 ng/mL; 95% CI: 4.73, 7.88; P = 0.0002]. All subsequent analyses of cord blood leptin were performed with adjustment for sex and are summarized in Table 2. There was a strong positive association with birth weight and with the derived variables that act as measures of body fatness (BMI or ponderal index). The lack of association with midupper arm circumference may reflect a lack of discriminatory power in this variable when assessed in neonates. Low-birth-weight babies (<2500 g; n = 19) had mean plasma leptin concentrations of 2.44 ng/mL (95% CI: 1.66, 3.44), whereas the concentrations in normal-weight babies (n = 81) were 5.05 ng/mL (95% CI: 4.00, 5.92; P = 0.0006).
View this table:
TABLE 2. Bivariate analysis of factors associated with cord blood leptin1
In the subset for whom maternal leptin values were available (n = 60), there was a modest correlation between late gestation maternal leptin and cord blood leptin (r = 0.51, P = 0.027) after adjustment for birth weight and sex. This correlation probably reflects that fact that maternal BMI was associated with both maternal leptin and birth weight and hence indirectly with cord blood leptin.
Cord blood leptin concentrations as a predictor of infant growth
Ong et al (16) reported that cord blood leptin concentrations predicted subsequent weight, length, and head circumference growth up to 4 and 12 mo in 197 British infants and that low cord blood leptin concentrations predicted catch-up growth. We retested this within the current data. After adjustment for sex, there was no relation between cord blood leptin concentrations and any measures of growth to 16 or 52 wk (n = 100). Additional adjustment by season of birth made no difference in this result.
Leptin concentrations during infancy
The average plasma leptin concentrations between birth and age 52 wk are shown in Figure 2. The outstanding features are the decline by approximately 90% in mean concentrations as infancy progresses and the fact that the strong sex difference at birth had disappeared by age 8 wk and remained absent thereafter when viewed as absolute concentrations. In fact, it is shown in Table 3 that, at weeks 16 and 52, there is a strong sex effect (ie, higher concentrations in girls thanin boys) after adjustment for BMI z scores. The similarity of the absolute values was caused by the fact that the girls had lower BMIs than did the boys.
FIGURE 2.. Geometric
View this table:
TABLE 3. Sex differences in leptin concentrations at birth and in infancy1
At weeks 8, 16, and 52, leptin concentrations were very strongly associated with BMI z score (Table 4). The correlations were stronger than those at birth. The upper half of Figure 3 shows absolute concentrations of leptin by quartiles of BMI z score (where the interquartile boundaries and assignment to quartiles were calculated for the sexes separately before the data were repooled). The graph shows strong positive relations at all ages, but at lower absolute concentrations as infancy progressed. It appears from the upper half of this graph that the slope of the relation becomes attenuated with age. However, the lower half of the graph, which replots the values as a percentage of the global mean for the age group, shows that this is not the case. In fact, the expression of the mean leptin concentration in the highest quartile of BMI as a multiple of the concentration in the lowest quartile, as seen in Figure 4, indicates that the ability of leptin to discriminate between the thinnest and fattest children became progressively better from cord blood onward.
View this table:
TABLE 4. Associations between plasma leptin concentrations in infancy, BMI, and antecedent change in BMI
FIGURE 3.. Geometric
FIGURE 4.. Geometric
To test whether leptin concentrations provide a "readout" of recent changes in energy status as well as of absolute status, the change in BMI z score since the anthropometric measurements 4 wk earlier was entered as a variable. This test showed that, as infancy progressed, leptin concentrations were increasingly influenced by the change in BMI z score (Table 4) and that there was quite a strong influence at 52 wk (P = 0.009). The data at 52 wk are illustrated in Figure 5; the infants are divided into 2 groups according to their energy status (BMI z score above or below the median) and then further divided according to their BMI trajectory (gainers compared with losers since the previous measurement). An infant's BMI was the dominant determinant of plasma leptin, but recent change in BMI was also a significant modifier of leptin concentrations across the whole BMI range. The fattest children who were gaining weight had circulating leptin concentrations 4 times those of the thinnest children who were losing weight.
FIGURE 5.. Geometric
In this population, the trajectory of weight (and hence BMI change) is strongly influenced by infections. Therefore, the plasma ACT concentration was also introduced as a variable. At age 8 and 52 wk (but not at 16 wk), ACT concentration was a significant negative predictor of plasma leptin, and, at 52 wk, it displaced the change in BMI.
DISCUSSION
A key feature of these data is that they originate from rural African children exposed to the sorts of dietary inadequacy and infectious disease stresses that probably dominated human evolutionary selection for 12 000 y—ie, since the start of the agricultural era—and probably for much longer. The children are born rather small (which indicates maternal uterine constraint), they grow extremely well when fully breastfed (which indicates a genetic potential similar to that of Western children), and then they deteriorate progressively (compared with the UK reference norms) during the remainder of infancy. At age 12 mo, their weight-for-age z scores average –1.9, their length-for-age z score is –1.1, and their BMI z score is –1.5. The pattern of birth weight and growth in infancy is also affected by the annual hungry season, which coincides with a time of increased incidence of infectious diseases, particularly in the children (17).
A disadvantage of this study, shared with most others in the literature, is that we have no direct measures of body composition in the neonates or infants. BMI was the best available proxy measure. Although it gives only a crude measure of body fatness, BMI has yielded results in reference to leptin concentrations that are remarkably robust. However, in interpreting the results, it must be remembered that BMI may have a variable correlation with body fat at different ages and in males and females.
It was shown that there is a dramatic increase in fetal leptin concentrations during the end of gestation, the time when the development of fetal adipose tissue begins (18). This increase suggests that leptin is synthesized and secreted by the fetus in an autonomous fashion (19). Our observations at birth accord qualitatively with those of most other published studies, although the average leptin concentration of 5.9 ng/mL in these thin mothers was lower than values reported from Western populations [eg, 22.6 ng/mL (20) and 13.5 ng/mL (21)]. We found a strong relation with birth weight and with measures likely to be related to fatness (BMI and ponderal index), but not with other measures of body size, such as length or head circumference. In numerous other studies, leptin concentrations were also shown to be positively correlated with birth weight and with measures of body fat such as BMI or ponderal index (16, 18–30). We found no significant associations with gestational age, prematurity (gestation <37 wk), or intrauterine growth retardation (defined as <10th centile of standard), but this may have been due to the relative homogeneity of the sample. We found a modest but significant association between maternal and cord blood leptin concentrations (P = 0.027). The literature is inconsistent on this point (31, 32). Our finding may be due to the facts that maternal uterine constraint in this environment has a greater effect on birth weight than it has in better-nourished populations, and maternal BMI was consequently correlated with birth weight, which in turn was related to increased fetal leptin concentrations.
We found a significant sex difference in cord blood leptin, even after adjustment for fetal weight and BMI: the girls had leptin concentrations double those of the boys (6.10 and 3.17 ng/mL, respectively; P = 0.0002). A sex difference in favor of higher concentrations in female neonates has also been reported in most other studies with adequate sample size (16, 18, 23–25, 30). A review stating that there is no sex effect at birth has cited several studies with very small sample sizes (26). Where a sex difference has been found, it has been shown, in common with our result, to be independent of growth status, which suggests that sexual dimorphism already exists in utero (16, 18, 25). Androgens are known to suppress, and estrogens have been reported to promote, leptin expression both in vitro and in vivo (26), although it is not known whether prenatal sex hormone exposure drives the sex difference.
By age 8 wk, leptin concentrations had declined markedly, especially in the girls, and the sex difference had disappeared. At age 8 wk, there was no suggestion of a sex difference, but it reappeared at ages 16 and 52 wk after adjustment for the fact that girls had lower average BMI z scores than did boys. We reported elsewhere that, in 6–8-y-old children from this population, the sex difference in leptin concentrations was strong (8), and others reported similarly higher leptin concentrations in girls during childhood (33, 34). There is evidence from dual-energy X-ray absorptiometry studies of children aged 3 y that girls have a higher proportion of body fat than do boys (34, 35).
There are few published data from other populations with which to compare our current observations of changes in leptin concentrations during infancy, although it was previously reported that leptin concentrations decrease markedly during very early postnatal life (23, 25). In the current study, circulating leptin concentrations decreased by a factor of 7 in the boys and a factor of 10 in the girls between birth and age 12 mo. The fact that this decrease occurred in all quartiles of the BMI distribution suggests that it is a normal developmental change rather than a result of the deteriorating nutritional status of the children. However, data from well-nourished, uninfected infants are required to confirm this.
Despite the decrease in absolute leptin concentrations in late infancy, leptin's ability to provide afferent signals about the energy status in central and peripheral adipose tissue stores appeared to be at least as effective as at other times. In fact, the ratio of leptin concentrations in the highest and lowest quartiles of BMI, a measure of leptin's discriminatory power, increased progressively from birth onwards (Figure 4), which may suggest that the system is becoming better tuned as infancy progresses. Some caution should be added to this interpretation, however, because it could also result from the fact that BMI may be a better measure of adipose tissue mass in older infants. Direct assessments of fat mass would be required to test this possibility.
Another element of improved sensitivity is that, by infant age of 52 wk, leptin was responding to both chronic energy status (assessed by BMI) and acute modulation of energy status (assessed by antecedent change in BMI). This finding matches data from adults, in whom leptin is responsive both to static energy stores and to the direction and rate of energy flux (3–5). The fact that ACT concentrations, a measure of inflammation, were a stronger predictor than was change in BMI may result from ACT's correlation with the inanition of disease and its provision of a more immediate measure of the trajectory of energy status than does the change in BMI over the 4 wk before the blood sample.
Surprisingly, we could detect no significant influence of season of the year on BMI or leptin concentrations. This contrasts with the typical pattern of quite pronounced seasonal swings in nutritional status between hungry and harvest seasons; it may be that the year when this study was conducted was a benign year, which was made possible by an increased reliance on a cash economy in these villages that helps to smooth out the previous seasonal variations of the subsistence economy. Alternatively, it may reflect the fact that young infants are shielded from seasonality when fully breastfed.
In summary, these new data from infancy support the accepted view that leptin concentrations provide a robust measure of medium- and long-term energy status. The signal develops prenatally and appears to become more finely tuned as infancy progresses. The evidence for this comes from the fact that there is a decline of 70–90% in circulating concentrations over the first year of life, and yet there is an improvement in leptin's discriminatory power, as well as a developing ability to monitor both static concentrations and dynamic trends in energy stores. We believe that the low BMIs seen in this population, and the consequent low concentrations of leptin, probably reflect the sort of ecologic setting in which leptin (and its dose-response characteristics with reference to the regulation of metabolic functions) must have evolved. Data obtained from Western populations should accordingly be interpreted with great care when investigators try to impute evolutionary significance to the "normal" functions of leptin.
ACKNOWLEDGMENTS
We are grateful to all of the subjects who participated in this study, as well as to the field staff (Baba Jobarteh, Ebrima Camara, Kebba Bajo, Sana Fabureh, Yusufa Darboe, and Lamin Njie) and midwives (Frances Foord, Fatou Sossay, and Ndeye Bah).
AC was principal investigator of the main immunologic study from which the residual samples used in this analysis were obtained. SM and AP were closely involved in all aspects of the original study. MO and CC performed the leptin analyses. AC, SM, and AP analyzed the data and wrote the manuscript. None of the authors had any conflict of interest.
REFERENCES