Abstract
Background
Prematurely born individuals are usually of low or normal weight in childhood; in adulthood, however, their probability of being overweight is twice that of persons born at full term. There is not yet any way to predict the weight development of premature babies.
Methods
A polygenic BMI score (BMI = body-mass index), calculated from the often very small individual effects of more than 2 million genetic variants, was recently described for adults. We studied the possible association of this score with the course of BMI in premature babies over time, from infancy up to the age of 10–14 years.
Results
508 individuals were included in the study. At the age of 5–7 years, their mean body weight was 18.8 ± 3.3 kg. The difference between the highest and lowest deciles of the polygenic score was 3.3 kg. At age 10–14, the average body weight was 41.3 ± 11.3 kg, and the difference between the highest and lowest deciles had increased to 9.2 kg. In persons with birth weight under the 10th percentile (n = 68), the difference was 19.2 kg (30.9 kg vs. 50.1 kg). The polygenic BMI score was significantly associated with the BMI z-scores of the overall group and the subgroup of growth-retarded children.
Conclusion
Extreme values of a polygenic BMI score are strongly associated with the weight development of preterm infants as they develop into children aged 10–14. The large effect size implies that this score may aid in the counseling of prematurely born children and their parents.
Approximately 10% of children are born prematurely; prematurity is defined as a birth at a gestational age less than 37 weeks. In both prematurely born individuals and growth-retarded children with birth weights below the 10th percentile of their respective gestational age (small for gestational age, SGA), the risk of becoming overweight or developing a metabolic syndrome-related condition is twice and twice to four times, respectively, that of children born at full term (1, 2). The reasons why prematurely born persons are at a greatly increased risk of metabolic syndrome-related conditions are not yet fully understood; however, an increased body mass index (BMI) appears to be a key risk factor (2). The parental BMI is known to have an effect on the weight of the child and can be easily determined (3, 4). However, the majority of prematurely born individuals and SGA children are underweight before puberty and usually become obese not earlier than age 14–19 (5). Since it is not yet possible to predict the long-term risk of obesity, it is recommended that all premature babies and SGA children initially receive a high-calorie diet after birth during their hospital stay to ensure adequate catch-up growth (6). So far no evidence is available on the question of when and especially in which preterm infants a high-calorie diet should be discontinued to prevent the development of obesity.
In 2019, a polygenic risk score for the human BMI was published which was calculated on the basis of more than two million genetic variants (7). Adults in the highest decile, i.e. the highest 10%, had a mean body weight of 85.3 kg and a mean BMI of 30 kg/m2. These values were 13 kg (4.8 kg/m2) above the mean values for adults in the lowest decile, i.e. the lowest 10%, of the polygenic risk score (Box).The differences were particularly pronounced in the case of obesity with a BMI of over 40, which were found in only 0.2% of adults in the lowest decile, but in 5.6% of those in the highest decile of the score. The comparison of adults in the highest decile with those in the remaining 9 deciles showed a five-fold increased risk of bariatric surgery and a 72% increased risk of diabetes mellitus (7). The polygenic BMI risk score was associated with weight in children too. In newborns, the weight difference between the lowest and the highest decile was 60 g, in 8-year-olds it was 3.5 kg and in 18-year-olds it was 12.3 kg (7). Polygenic risk scores have not yet been adopted in routine clinical practice. However, first large clinical studies on the implementation of polygenic risk scores for ten common diseases, including obesity, in clinical practice are underway in the United States (8, 9).
Box. Polygenic risk score for BMI prediction.
Polygenic risk scores are calculated by multiplying common genetic variants, which are associated with a specific characteristic (such as BMI), by their effect size and then adding them up (7).
The effect sizes of individual genetic variants included in the score used in our study are generally very small (<1 g). By adding up the effects of more than two million variants, the body weight of adults in the highest and lowest decile (i.e., the highest and lowest 10%) of the score differs by 13 kg (7).
Despite this large difference, the score explains only 8.4% of the BMI variance in the general population. Likewise, the predictive power for obesity in the general population, measured as area under the curve (AUC), is only 0.64 (14). An AUC of 0.5 indicates a random distribution, an AUC of 1 an ideal prediction.
In populations at risk for obesity, e.g., adults who survived cancer in childhood, the predictive power of the score is considerably higher with an AUC of 0.75 (12).
In this study, we tested the hypothesis that in individuals born preterm with a birth weight of less than 1500 g the polygenic BMI risk score is associated with weight development in childhood. Provided the effect size is sufficiently large, such a score could be used as an aid in counseling prematurely born children and their parents to help prevent the development of obesity in puberty.
Methods
Patients and study design
Premature babies born at a gestational age of less than 37 weeks + 0 days and with a birth weight of less than 1500 g were included in a multicenter cohort study (German Neonatal Network, GNN) during their hospital stay (eList). Preterm infants with a birth weight below the 10th percentile of the respective gestational age were categorized as SGA preterm infants (10). Detailed information on participant selection and study design is provided in the eMethods section and in eFigure 1.
eMethods.
Patients and study design
Once the written informed consent of the parents was obtained, the clinical data were collected by staff of the participating hospitals over the course of the inpatient stay and monitored and checked by a team from the central study office (University of Lübeck). In addition, a DNA sample of the child was obtained during the inpatient stay. At the ages of 5–7 and 10–14 years, -follow-up examinations were performed by a team of the central study office; the children were usually examined in the rooms of the pediatric department of the hospital where there were born. For the analyses in this study, we selected children who had undergone height measurement and weight measurement at both ages 5–7 and 10–14 years, who had fat mass and grip strength of the right hand determined at age 10–14 years, and who had the polygenic Khera BMI risk score calculated based on SNP array genotyping data. Since the GNN study had started to enroll patients in January 2009 and the children should be 10 years of age at the time of the second follow-up examination, only children born between January 2009 and April 2014 were eligible for inclusion in the analysis.
The GNN study was evaluated and approved by the ethics committees of the University of Lübeck and the ethics committees of all participating hospitals.
SNP array genotyping and calculation of the -polygenic BMI risk score
SNP array genotyping was carried out at the Cologne Center for Genomics using AXIOM CEU chips (Affymetrix, Santa Clara, USA) and at the Institute of Clinical Molecular Biology at Kiel University using Global Screening Array chips (Illumina, San Diego, USA). After imputation of additional single nucleotide polymorphisms (SNPs) on the basis of the 1000 Genomes database (1000 G Phase 3) and quality control, the polygenic scores according to Khera were calculated on the basis of the data published in the Polygenic Score Catalog https://www.pgscatalog.org/score/PGS000027/). Even though most of the genetic variants in the score described by Khera and used by us have no causal effect on the BMI, the predictive power of the score is significantly higher than the predictive value of genetic scores confined to a few variants with a functional effect on the BMI.
Follow-up examinations and determination of the -endpoints
The central study office invited the parents and children to -attend the follow-up visits. The invitation process was the same for all participating hospitals and designed to invite a representative sample to the follow-up examinations. For both follow-up visits, all required instruments were brought along by the GNN team to the respective hospital. The height of the child was -determined using a Harpenden Portable Stadiometer (Holtain Inc. UK) or a Seca 217 Stadiometer (Seca GmbH, Germany) and the body weight was measured using an ADE MAK T7336 scale (ADE GmbH, Germany). The BMI was calculated as follows: BMI (kg/m2] = body weight (kg)/height2 (m). The BMI z-scores were calculated based on the reference values by Kromeyer–Hauschild (13). We also used the threshold values defined there for severe underweight (z-score < –1.88 or <3rd percentile), underweight (z-score –1.88 to < –1.28 or 3rd to <10th percentile), normal weight (z-score –1.28 to +1.28 or 10th to 90th percentile), overweight (z-score above 1.28 to 1.88 or above 90th to 97th -percentile) and obesity (z-score over 1.88 or over the 97th percentile). Strength was determined using a digital hand dyna-mometer (Jamar plus, Lafayette Instrument, USA) and the fat mass was determined using bioelectrical impedance analysis (Nutribox, Data Input GmbH, Germany). Information on parental height and parental weight was obtained at the follow-up at age 5 to 7 years. The parental BMI was calculated in the same way as for the children and the parents were categorized into the following groups:
Normal weight (BMI of both parents <25 kg/m2)
Obesity (BMI of father or mother >30 kg/m2)
All other BMI combinations.
Statistical analysis
The basic data and the follow-up data at the ages 5–7 and 10–14 years were first descriptively grouped into children with a low polygenic BMI risk score (first decile), an intermediate polygenic BMI risk score (deciles 2–9) and a high polygenic BMI risk score (tenth decile). The basic data and all secondary endpoints at age 5–7 years (body weight, height, BMI) and age 10–14 years (body weight, height, BMI, grip strength, absolute fat mass, proportion of underweight and overweight children) were analyzed using a linear or logistic regression model where the polygenic BMI risk score decile was used as continuous independent variables. For continuous data, the effect size per decile of the polygenic risk score was described. For describing the effect sizes of categorical data, we used the odds ratios per decile of the polygenic BMI risk score, each with 95% confidence intervals [95% CI]. Since the confidence intervals of these analyses were not -adjusted for multiple testing, some of the secondary analyses may not be reproducible. To estimate the sensitivity and speci-ficity of the genetic score, we performed a receiver operating characteristic (ROC) analysis with the endpoint obesity at age 10–14 years as the dependent variable and the raw score values of the polygenic BMI risk score or the maternal and paternal BMI as the test variables.
eFigure 1.
Patient selection
GNN, German Neonatal Network; SNP, single-nucleotide polymorphism
Genotyping and calculation of the polygenic BMI risk score
DNA extraction, genotyping using a single nucleotide polymorphism (SNP) array (Affymetrix AXIOM or Illumina Global Screening Array) and imputation of additional polymorphisms in the GNN have already been described elsewhere (11). We used the BMI risk score describe by Khera et al. which includes 2 100 302 single-nucleotide polymorphisms (7, 12), of which 2 064 658 (98.3%) were available in the genotyping data of the GNN study and could be used for the calculation.
Statistical analysis
Associations of the polygenic BMI risk score decile with the BMI z-scores according to Kromeyer-Hauschild (13) measured at age 10–14 years were determined using linear regression for the overall group and for the following subgroups:
SGA children
Normal parental weight (BMI of both parents < 25)
Parental obesity (BMI of father or mother > 30 kg/m2).
This resulted in four linear regression analyses, whereby the type-one error level according to Bonferroni was reduced from 0.05 to 0.0125 in order to achieve a global significance level of 0.05.
Results
A total of 23 652 premature babies with a birth weight of less than 1500 g were included in the GNN between 2009 and 2023. By April 2024, 13 150 of these children had already been genotyped using the SNP array. Of these, 12 628 (96%) had been discharged alive. The BMI risk score was not associated with in-hospital mortality. By April 2020, a total of 508 children had taken part in both the follow-up at 5 to 7 years and the follow-up at 10 to 14 years of age (including assessment of fat mass and grip strength) and thereby fulfilled the inclusion criteria.
The mean birth weight was 1002 ± 283 g. The mean gestational age at birth was 28.1 ± 2.3 weeks; it was slightly lower among premature babies with a higher BMI risk score. Other than that, no differences were found in the overall group with regard to the basic data or the weight gain achieved until discharge from hospital (Table 1).
Table 1. Basic data after birth.
|
Decile of the polygenic
BMI risk score |
Decile 1
(n = 50) |
Deciles 2–9
(n = 408) |
Decile 10
(n = 50) |
| Gestational age (weeks) | 28.3 (2.5) | 28.1 (2.3) | 27.6 (2.2) |
| Birth weight (g) | 1014 (311) | 1003 (281) | 976 (283) |
| Boys (n, %) | 22 (44) | 211 (52) | 27 (54) |
| Multiples (n, %) | 28 (56) | 153 (38) | 18 (36) |
| SGA (n, %) | 9 (18) | 52 (13) | 7 (14) |
| Length of hospital stay after birth (days) | 75 (39) | 74 (31) | 71 (30) |
| Weight on discharge (g) | 2567 (801) | 2635 (548) | 2571 (603) |
| Weight gain per day (g) | 22 (6) | 23 (5) | 22 (6) |
All figures are mean values (standard deviation) or n (%).
SGA, small for gestational age
The mean age at the time of the follow-up examination at 5 to 7 years and at 10 to 14 years of age was 5.8 ± 0.4 years and 12.0 ± 1.2 years, respectively. The BMI z-score according to Kromeyer-Hauschild showed a strong correlation with the polygenic BMI risk score (+0.12 Z scores per additional decile of the polygenic BMI risk score; 95% confidence interval: [+0.08; +0.16] p = 7.1 × 10–11). Likewise, all other weight variables, such body weight in kilogram, BMI and fat mass in kilogram, were highly associated with the polygenic BMI risk score, both at the follow-up examination at 5–7 years and at 10–14 years of age (Table 2). At the 5– to 7-years follow-up, the difference in mean weight between children of the first and tenth decile of the polygenic BMI risk score was 3.3 kg. At age 10 to 14, the difference between the two deciles had increased to 9.2 kg (+0.71 kg [+0.37; +1.04 kg] per additional decile of the polygenic BMI risk score). At age 5 to 7, the children of the highest decile of the polygenic BMI risk score were slightly taller, but this difference was no longer observed at the 10– to 14-years follow-up. The polygenic BMI risk score was not associated with grip strength (Table 2).
Table 2. Growth data by polygenic BMI risk score deciles.
|
Decile of the polygenic
BMI risk score |
Decile 1
n = 50 |
Deciles 2–9
n = 408 |
Decile 10
n = 50 |
Effect size per decile
[95% CI] |
p |
| Follow-up at 5 to 7 years of age | |||||
| – Weight (kg) | 17.7 (2.8) | 18.7 (2.3) | 21.0 (4.0) | +0.24 +[0.14; +0.34] | |
| – Height (cm) | 112 (7) | 113 (6) | 116 (6) | +0.26 [+0.07;+ 0.44] | |
| – BMI (kg/m2) | 14.1 (1.2) | 14.6 (1.5) | 15.6 (6) | +0.11 [+0.07; +0.16] | |
| – BMI z-score* | –1.1 (1.0) | –0.7 (1.0) | –0.1 (1.1) | +0.07 [+0.04; +0.10] | |
| Follow-up at 10 to 14 years of age | |||||
| – Weight (kg) | 38.0 (9.0) | 41.0 (11.0) | 47.2 (13.3) | +0.71 +[0.37; +1.04] | |
| – Height (cm) | 152 (10) | 151 (10) | 152 (10) | +0.09 [−0.22; +0.40] | |
| – BMI (kg/m2) | 16.3 (2.6) | 17.8 (3.1) | 20.2 (4.5) | +0.29 [+0.19; +0.39] | |
| – BMI z-score* | –1.2 (1.2) | –0.3 (1.1) | +0.5 (1.2) | +0.12 [+0.08; +0.16] | 7.1 × 10−11 |
| – Body fat (kg) | 7.8 (5.3) | 9.4 (5.4) | 12.5 (7.7) | +0.40 [+0.23; +0.57] | |
| – Grip strength (kg) | 20 (6) | 20 (6) | 20 (6) | −0.01 [−0.19; +0.18] | |
*The BMI z-score indicates the deviation from the median of the general population in standard deviations. At the age of 12, the BMI for boys is 17.99 kg/m2 and for girls 18.19 kg/m2. At this age, the threshold for obesity (which was exceeded by only 2.8% of the prematurely born individuals in our study) is for boys at a BMI of 25.44 kg/m2 and for girls at a BMI of 25.47 kg/m2 (13). Prematurely born individuals in decile 10 with a BMI of 20.2 kg/m2 are +0.5 standard deviations above the expected average, while prematurely born individuals in decile 1 are -1.2 standard deviations below it. All figures (except for effect size) are mean values (± standard deviation).
BMI, body mass index
The weight development in the three subgroups (SGA, parents with BMI <25, at least one parent with BMI >30) from birth to age 10–14 years is summarized in Table 3. The most remarkable group were the SGA children. At birth, the mean weight of the 68 SGA children was 728 ± 319 g and there were no differences between the deciles of the polygenic BMI risk score. At the time of discharge from hospital, the preterm infants in the highest deciles of the polygenic BMI risk score had already gained slightly more weight, while the SGA preterm infants in the first decile of the polygenic BMI risk score were discharged with a weight that was about 300 g below the weight of the non-SGA preterm infants. This trend of very strong weight gain in the 10th decile of the polygenic BMI risk score and low weight gain in the first decile was maintained throughout the follow-up period and led to a marked weight difference of 19.2 kg between the 1st decile and 10th decile at the follow-up examination after 10–14 years (+1.4 kg [+0.4; +2.3 kg] per additional decile of the polygenic BMI risk score).
Table 3. Weight development of prematurely born individuals of different subgroups by polygenic BMI risk score deciles.
| Subgroup | Decile 1 | Deciles 2–9 | Decile 10 |
Effect size per decile
[95% CI]* |
p |
| SGA preterm infants | n = 9 | n = 52 | n = 7 | ||
| – Birth (g) | 717 (279) | 736 (325) | 681 (367) | −11 [−37; +15] | |
| – Discharge (g) | 2 287 (280) | 2 437 (539) | 2 538 (577) | +47 [+6; +88] | |
| – 5–7 years (kg) | 14.8 (2.6) | 16.9 (2.8) | 20.1 (4.1) | +0.4 [+0.16; +0.64] | |
| – 10–14 years (kg) | 30.9 (7.0) | 37.2 (10.9) | 50.1 (17.4) | +1.4 [+0.4; +2.3] | |
| – 10– to 14-year BMI z-score | −2.5 (0.7) | −0.7 (1.3) | +0.7 (1.6) | +0.24 [+0.12; +0.34] | 0.00006 |
| Normal-weight parents | n = 11 | n = 87 | n = 5 | ||
| – Birth (g) | 959 (337) | 1 044 (265) | 1 046 (292) | +5 [−15; +24] | |
| – Discharge (g) | 2 752 (1 054) | 2 590 (435) | 2 678 (485) | +5 [−33; +44] | |
| – 5–7 years (kg) | 16.8 (2.1) | 17.9 (2.6) | 21.4 (4.6) | +0.15 [−0.05; +0.36] | |
| – 10–14 years (kg) | 35.4 (5.6) | 39.1 (9.0) | 40.7 (7.4) | +0.32 [−0.3; +0.9] | |
| – 10– to 14-year BMI z-score | −1.7 (0.7) | −0.8 (1.0) | −0.2 (0.8) | +0.1 [−0.001; +0.13] | 0.052 |
| Parents with obesity | n = 4 | n = 106 | n = 17 | ||
| – Birth (g) | 1 076 (300) | 974 (280) | 982 (241) | −4 [−22; +13] | |
| – Discharge (g) | 3 041 (921) | 2 581 (594) | 2 374 (688) | −21 [−61; +20] | |
| – 5–7 years (kg) | 19.3 (1.2) | 19.5 (3.8) | 21.1 (4.7) | +0.4 [+0.2; +0.7] | |
| – 10–14 years (kg) | 39.6 (11.3) | 43.2 (12.7) | 46.8 (13.3) | +0.8 [−0.02; +1.6] | |
| – 10– to 14-year BMI z-score | −0.25 (1.4) | +0.03 (1.2) | +0.6 (1.2) | +0.06 [−0.006; +0.13] | 0.07 |
*The effect size was calculated using a linear regression model; it indicates the mean effect of an additional decile. An effect size of +1.4 kg for the body weight of SGA preterm born individuals aged 10–14 thus means that the children at this age were on average 1.4 kilograms heavier per additional decile. In the available analyses, however, the differences between the lowest and highest deciles of the risk score were always particularly large, while the middle deciles showed less marked differences. This is also obvious in eFigure 2. All figures (except for effect size) are mean values (± standard deviation).
BMI, body mass index; SGA, small for gestational age
Maternal and paternal BMI values of 416 prematurely born individuals were available at the follow-up at 5 to 7 years of age. The mean maternal BMI was 25.3 ± 5.7 kg/m2, the mean paternal BMI was 27.2 ± 4.2 kg/m2. The polygenic BMI risk score was associated with child BMI z-scores, even when the parental BMI values were included in the linear regression analysis (eTable 1). As expected, the BMI of the parents was associated with a corresponding shift in the child‘s weight (eFigure 2, Table 3).
eTable 1. Association of the polygenic BMI risk score and the parental BMI with the child BMI at the follow-up at 10 to 14 years of age.
| Influencing factor | Effect size*1 | 95% confidence interval | p-value |
| Maternal BMI*2 | +0.024 | [+0.004; +0.044] | 0.017 |
| Paternal BMI*2 | +0.064 | [+0.037; +0.091] | 4.1 ×10−6 |
| Child’s polygenic BMI score | +0.094 | [+0.057; +0.131] | 7.6 ×10−7 |
Multiple linear regression; dependent variable: child BMI z-scores at the follow-up at 10 to 14 years of age. *1The effect size is given per additional BMI (kg/m2) or per additional decile of the polygenic score. *2 The maternal and paternal BMI values were obtained at the follow-up at 5 to 7 years of age. Complete values from both parents were available for 416 children.
eFigure 2.
BMI z-scores in prematurely born individuals at age 10–14 years by decile of the polygenic BMI risk score in the overall group (a); in SGA preterm born individuals (b); in prematurely born individuals with a parental BMI <25 kg/m2 (c), and in prematurely born individuals with a parental BMI >30 kg/m2 (d). The figures show the means as dots and the 95% confidence intervals of the means.
The analysis of the follow-up BMI z-scores of the prematurely born individuals based on the reference values by Kromeyer-Hauschildt, which are recommended in Germany, found mean z-scores of –0.7 ± 1.0 and –0.3 ± 1.2 for the overall group at age 5–7 and age 10–14, respectively. This means that the prematurely born individuals were underweight on average in both follow-up age groups, but approached the mean z-score of normal-weight children (which is 0). As a result, however, the proportion of overweight prematurely born individuals also increased. At age 5–7, only 0.8% of the children were obese and 2.4% overweight. At age 10–14, 69% of the children were of normal weight, 22% were underweight and 9% were overweight or obese. The distribution was particularly extreme in the extreme deciles of the polygenic BMI risk score in children who were born preterm and small for gestational age. All SGA preterm infants in the first decile of the polygenic BMI risk score were underweight at age 10–14 (the majority were severely underweight). In the 10th decile, 3 of 7 children were overweight or obese (Figure). For the endpoint “obesity at age 10–14“, the area under the curve (AUC) in the ROC diagram was 0.74 [0.61; 0.87] for the overall study population. In the subgroup of children for whom we had information on parental BMI available, the polygenic risk score achieved a higher AUC compared to the maternal and paternal BMI (eFigure 3). The sex of the children was not associated with the frequencies of underweight and obesity, both at the age 5–7 and age 10–14 follow-ups (eTable 2).
Figure.
Association of the polygenic BMI risk score with the weight categories of prematurely born individuals at age 10–14 years in the overall group (left) and in the group of SGA preterm born individuals (right). SGA, small for gestational age
eFigure 3.
Receiver Operating Curve (ROC) for the endpoint “Obesity at age 10–14 years“
eFigure 3a shows the ROC for the polygenic BMI risk score in the overall group (n = 508, 14 children with obesity, AUC = 0.739 [0.61; 0.87]). eFigure 3b shows ROC curves (n = 416, 7 children with obesity) for the maternal BMI (green, AUC = 0.768 [0.63; 0.90]), the paternal BMI (red, AUC = 0.748 [0.59; 0.91)] and the polygenic BMI risk score (purple, AUC = 0.805 [0.67; 0.94]). Since a higher AUC indicates a higher prognostic value, eFigure 3b shows that the polygenic BMI risk score used by us is more strongly associated with obesity in prematurely born individuals at age 10–14 years than to the paternal or maternal BMI.
eTable 2. Weight categories of prematurely born individuals by sex.
|
Boys
n = 260 |
Girls
n = 248 |
Total
n = 508 |
|
| Follow-up at 5 to 7 years of age | |||
| Severely underweight | 41 (16) | 33 (13) | 74 (15) |
| Underweight | 30 (12) | 32 (13) | 62 (12) |
| Normal weight | 181 (70) | 175 (71) | 356 (70) |
| Overweight | 7 (3) | 5 (2) | 12 (2) |
| Obesity | 1 (0.4) | 3 (1.2) | 4 (0.8) |
| Follow-up at 10 to 14 years of age | |||
| Severely underweight | 30 (12) | 15 (6) | 45 (9) |
| Underweight | 33 (13) | 35 (14) | 68 (13) |
| Normal weight | 173 (67) | 176 (71) | 349 (69) |
| Overweight | 18 (7) | 14 (6) | 32 (6) |
| Obesity | 6 (2.3) | 8 (3.2) | 14 (2.8) |
All figures are n (%).
Discussion
The polygenic BMI risk score that we used in our analysis reflects part of the genetic predisposition to develop a high or low BMI in the overall population. The predictive power of the score in the overall population for the outcome “obesity” is moderate with an AUC of 0.64 (14). However, it has already been shown for specific populations at risk of obesity, for example children who survived cancer, that the score is predictive for the development of obesity in adulthood (AUC = 0.75) (12).
The data of our study revealed a significant association of the polygenic BMI risk score with the BMI of prematurely born individuals when they were 5–7 years old and 10–14 years old. Only 14 of the 508 (2.8%) prematurely born individuals were obese at the age of 10–14 years. The predictive power of the polygenic BMI risk score was in the clinically relevant range with an AUC of 0.74 in the overall group of prematurely born individuals and with an AUC of 0.78 already at age 10–14 years in the group of individuals born as SGA preterm babies. Given that obesity in preterm infants usually does not develop until puberty (5), it is to be expected that a large proportion of premature babies in the highest range of the score will be obese as adults.
Worldwide, about 13 million children are born prematurely and 23 million children are born small for gestational age every year (15). Since both being underweight and overweight have long-term negative effects on health (16, 17), the data from our analysis may have important implications for the counseling of preterm infants, SGA infants and their parents.
A long-term study from the UK evaluated the effect of breastfeeding (five months or longer) on the weight development of children with various genetic risks. The analyses were controlled for gestational age, maternal BMI, education, family income, and smoking. Newborns with high polygenic BMI risk scores, who had been breastfed for at least five months, had a significantly lower body weight as adults compared to non-breastfed children (18). According to our data, SGA preterm infants with a high polygenic BMI risk score and possibly also prematurely born children of obese parents can in particular benefit from breastfeeding counseling and the resulting success in breastfeeding.
According to recent data from the German Health Interview and Examination Survey for Children and Adolescents (KiGGS) study, the prevalence of overweight and obesity in children aged 3–6 years is 10.8% for girls and 7.3% for boys. In adolescents aged 11–13 years, the prevalence of overweight and obesity was 20% in girls and 21% in boys (19). In both age groups, overweight and obesity was considerably less common among the prematurely born individuals in our study (eTable 2). According to epidemiological data, many prematurely born individuals experience extreme weight gain during puberty and are at particularly high risk of cardiovascular disease as obese adults (20). We suspect that prematurely born individuals with a high polygenic BMI risk score are at extremely high risk of becoming obese during puberty. The costs of determining a polygenic BMI risk score are low. However, there is uncertainty as to whether the knowledge of an increased risk of obesity, combined with counseling on preventive measures such as exercise and a calorie-reduced diet, is enough to prompt a change in the behavior of the affected prematurely born individuals during puberty. Only a randomized trial could provide answers to this question..
High values of the polygenic risk score used in our study are associated with increased appetite in childhood. Yet, appetite explains only about 11% of the effect size of the polygenic score (21). Drugs such as semaglutide have a very strong appetite-reducing effect (22). However, only 25.4% of obese adolescents who were treated with semaglutide experienced a normalization of their body weight (23, 24). This shows how difficult it is to treat overt obesity in adolescents and highlights the great potential for prophylactic measures in prematurely born individuals, as only 14 children in the cohort we studied were obese at the 10– to 14-years follow-up.
As reported in numerous publications, intrauterine malnutrition in combination with postnatal overnutrition leads to overweight and a shortened lifespan, in both animals and humans (25). The current treatment of SGA preterm infants matches this constellation. However, the data of our study show that, besides intrauterine deficiency, the individual predisposition to a high or low body weight also has a decisive effect on weight development from infancy up to the age of 10–14 years. Furthermore, the results published here confirm recent reports on the impact of paternal BMI on the weight development of children (4).
The most important limitation of our study is the lack of data on the weight development after puberty. Additionally, there are currently still very few cases in the subgroups of SGA children and children of obese or normal-weight parents.
In summary, we were able to show that a polygenic BMI risk score developed in adulthood is strongly associated with the weight development of preterm infants as they develop into adolescents aged 10–14 years. In this age group, only 2.8% of the prematurely born individuals already suffer from overt obesity so that it would be possible to offer targeted preventive genetic counselling of prematurely born individuals and their parents, especially in SGA preterm infants. However, only prospective scientific studies can show whether the knowledge of a high genetic risk for obesity actually leads to a reduction in weight gain in predisposed prematurely born individuals during puberty.
Acknowledgments
Acknowledgement: We would like to thank all parents and prematurely born children who made this study possible by participating in the GNN. We would also like to thank all physicians in the GNN network. A list of participating centers is provided in the supplement.
Translated from the original German by Ralf Thoene, M.D.
References (abbreviated)
1. Ashorn P, et al.: Lancet 2023; 401: 1692–1706.
2. Sipola-Leppänen M, et al.: Am J Epidemiol 2015; 181: 861–73.
3. Lichtwald A, et al.: Arch Gynecol Obstet 2024; 309: 105–18.
4. Tomar A, et al.:. Nature 2024; 630: 720–7.
5. Vinther JL, et al.: PLoS Med 2023; 20: e1004036.
6. Embleton ND, J et al.: J Pediatr Gastroenterol Nutr 2023; 76: 248–68.
7. Khera AV, et al.: Cell 2019; 177: 587–96.
8. Lennon NJ, et al.: Nat Med 2024; 30: 480–7.
9. Sabatello M, et al.: HGG Adv 2024; 5: 100281.
10. Voigt M, et al.: Z Geburtshilfe Neonatol 2012; 216: 212–9.
11. Göpel W, et al.: Arch Dis Child Fetal Neonatal Ed 2020; 105: 184–9.
12. Sapkota Y, et al.: Nat Med 2022; 28: 1590–8.
13. Kromeyer-Hauschild K, et al.: Monatsschr Kinderheilkd 2001; 149: 807–18.
14. Loos RJF, et al.: Nat Rev Genet 2022; 23: 120–33.
15. Lawn JE, et al.: Lancet 2023; 401: 1707–19.
16. Tumas N, et al.: Lancet 2024; 403: 998–9.
17. NCD-RisC: Lancet 2024; 403: 1027–50.
18. Wu Y, et al.: PLoS Genet 2020; 16: e1008790.
19. Schienkiewitz A, et al.: J Health Monit 2018: 22.
20. Markopoulou P, et al.: J Pediatr 2019; 210: 69–80.e5.
21. Renier TJ, Y et al.: Int J Obes (Lond) 2024; 48: 71–7.
22. Müller TD, et al.: Nat Rev Drug Discov 2022; 201–23.
23. Weghuber D, et al.: N Engl J Med 2022; 387: 2245–57.
24. Kelly AS, et al.: Obesity (Silver Spring) 2023: 2139–49.
25. Skowronski AA, et al.: Endocr Rev 2024; 45: 253–80.
Footnotes
Funding: German Federal Ministry of Education and Research (BMBF, Bundesministerium für Bildung und Forschung) (BMBF 01ER0805 und BMBF 01ER1501), German Research Foundation (DFG, Deutsche Forschungsgemeinschaft) (SFB 1665–515637292, Sexdiversity).
Conflict of interest statement
WG received reimbursement of travel expenses and payment of congress fees from GNPI.
The remaining authors declare no conflict of interest.
References
- 1.Ashorn P, Ashorn U, Muthiani Y, et al. Small vulnerable newborns—big potential for impact. Lancet. 2023;401:1692–1706. doi: 10.1016/S0140-6736(23)00354-9. [DOI] [PubMed] [Google Scholar]
- 2.Sipola-Leppänen M, Vääräsmäki M, Tikanmäki M, et al. Cardiometabolic risk factors in young adults who were born preterm. Am J Epidemiol. 2015;181:861–873. doi: 10.1093/aje/kwu443. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Lichtwald A, Weiss C, Lange A, et al. Association between maternal pre-pregnancy body mass index and offspring‘s outcomes at 9 to 15 years of age. Arch Gynecol Obstet. 2024;309:105–118. doi: 10.1007/s00404-023-07184-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Tomar A, Gomez-Velazquez M, Gerlini R, et al. Epigenetic inheritance of diet-induced and sperm-borne mitochondrial RNAs. Nature. 2024;630:720–727. doi: 10.1038/s41586-024-07472-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Vinther JL, Cadman T, Avraam D, et al. Gestational age at birth and body size from infancy through adolescence: an individual participant data meta-analysis on 253,810 singletons in 16 birth cohort studies. PLoS Med. 2023;20 doi: 10.1371/journal.pmed.1004036. e1004036. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Embleton ND, Jennifer Moltu S, Lapillonne A, et al. Enteral nutrition in preterm infants (2022): a position paper from the ESPGHAN committee on nutrition and invited experts. J Pediatr Gastroenterol Nutr. 2023;76:248–268. doi: 10.1097/MPG.0000000000003642. [DOI] [PubMed] [Google Scholar]
- 7.Khera AV, Chaffin M, Wade KH, et al. Polygenic prediction of weight and obesity trajectories from birth to adulthood. Cell. 2019;177:587–596. doi: 10.1016/j.cell.2019.03.028. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Lennon NJ, Kottyan LC, Kachulis C, et al. Selection, optimization and validation of ten chronic disease polygenic risk scores for clinical implementation in diverse US populations. Nat Med. 2024;30:480–487. doi: 10.1038/s41591-024-02796-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Sabatello M, Bakken S, Chung WK, et al. Return of polygenic risk scores in research: stakeholders‘ views on the eMERGE-IV study. HGG Adv. 2024;5 doi: 10.1016/j.xhgg.2024.100281. 100281. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Voigt M, Rochow N, Guthmann F, Hesse V, Schneider KT, Schnabel D. Birth weight percentile values for girls and boys under consideration of maternal height. Z Geburtshilfe Neonatol. 2012;216:212–219. doi: 10.1055/s-0032-1316324. [DOI] [PubMed] [Google Scholar]
- 11.Göpel W, Müller M, Rabe H, et al. Genetic background of high blood pressure is associated with reduced mortality in premature neonates. Arch Dis Child Fetal Neonatal Ed. 2020;105:184–189. doi: 10.1136/archdischild-2019-317131. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Sapkota Y, Qiu W, Dixon SB, et al. Genetic risk score enhances the risk prediction of severe obesity in adult survivors of childhood cancer. Nat Med. 2022;28:1590–1598. doi: 10.1038/s41591-022-01902-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Kromeyer-Hauschild K, Wabitsch M, Geller F, et al. Perzentile für den Body Mass Index für das Kindes- und Jugendalter unter Heranziehung verschiedener deutscher Stichproben. Monatsschr Kinderheilkd. 2001;149:807–818. [Google Scholar]
- 14.Loos RJF, Yeo GSH. The genetics of obesity: from discovery to biology. Nat Rev Genet. 2022;23:120–133. doi: 10.1038/s41576-021-00414-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Lawn JE, Ohuma EO, Bradley E, et al. Small babies, big risks: global estimates of prevalence and mortality for vulnerable newborns to accelerate change and improve counting. Lancet. 2023;401:1707–1719. doi: 10.1016/S0140-6736(23)00522-6. [DOI] [PubMed] [Google Scholar]
- 16.Tumas N, López SR. Double burden of underweight and obesity: insights from new global evidence. Lancet. 2024;403:998–999. doi: 10.1016/S0140-6736(24)00051-5. [DOI] [PubMed] [Google Scholar]
- 17.NCD Risk Factor Collaboration (NCD-RisC) Worldwide trends in underweight and obesity from 1990 to 2022: a pooled analysis of 3663 population-representative studies with 222 million children, adolescents, and adults. Lancet. 2024;403:1027–1050. doi: 10.1016/S0140-6736(23)02750-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Wu Y, Lye S, Dennis CL, Briollais L. Exclusive breastfeeding can attenuate body-mass-index increase among genetically susceptible children: a longitudinal study from the ALSPAC cohort. PLoS Genet. 2020;16 doi: 10.1371/journal.pgen.1008790. e1008790. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Schienkiewitz A, Brettschneider AK, Damerow S, et al. Overweight and obesity among children and adolescents in Germany. Results of the cross-sectional KiGGS Wave 2 study and trends. J Health Monit. 2018:22. doi: 10.25646/6797. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Markopoulou P, Papanikolaou E, Analytis A, Zoumakis E, Siahanidou T. Preterm birth as a risk factor for metabolic ayndrome and cardiovascular disease in adult life: a systematic review and meta-analysis. J Pediatr. 2019;210:69–80. doi: 10.1016/j.jpeds.2019.02.041. e5. [DOI] [PubMed] [Google Scholar]
- 21.Renier TJ, Yeum D, Emond JA, et al. Elucidating pathways to pediatric obesity: a study evaluating obesity polygenic risk scores related to appetitive traits in children. Int J Obes (Lond) 2024;48:71–77. doi: 10.1038/s41366-023-01385-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Müller TD, Blüher M, Tschöp MH, DiMarchi RD. Anti-obesity drug discovery: advances and challenges. Nat Rev Drug Discov. 2022:201–223. doi: 10.1038/s41573-021-00337-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Weghuber D, Barrett T, Barrientos-Pérez M, et al. Once-weekly semaglutide in adolescents with obesity. N Engl J Med. 2022;387:2245–2257. doi: 10.1056/NEJMoa2208601. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Kelly AS, Arslanian S, Hesse D, et al. Reducing BMI below the obesity threshold in adolescents treated with once-weekly subcutaneous semaglutide 2.4 mg. Obesity (Silver Spring) 2023:2139–2149. doi: 10.1002/oby.23808. [DOI] [PubMed] [Google Scholar]
- 25.Skowronski AA, Leibel RL, LeDuc CA. Neurodevelopmental programming of adiposity: contributions to obesity risk. Endocr Rev. 2024;45:253–280. doi: 10.1210/endrev/bnad031. [DOI] [PMC free article] [PubMed] [Google Scholar]