Metabolic and hormonal serum markers in early childhood: Follow‐up of a randomized controlled trial during infancy evaluating a modified, low‐protein infant formula

Abstract

Objectives

High protein intake during infancy has been linked to accelerated weight gain and increased obesity risk. This study aimed to examine the effects of a low‐protein formula during the first 6 months of life on blood metabolic and hormonal markers during early childhood.

Methods

Formula‐fed infants (<45 days) were randomized to receive either a low‐protein formula with modified amino acid composition (mLP; n = 90; 1.7 g protein/100 kcal) or a control formula (CTRL; n = 88; 2.1 g protein/100 kcal) until 6 months of age. Breastfed infants served as a reference group (n = 67). Blood samples were collected in cooperating subjects at 1, 2, and 6 years. We measured insulin‐like growth factor‐1 (IGF‐1), IGF‐binding proteins (BPs), leptin, glucose, and insulin, and calculated Homeostatic‐Model‐Assessment‐of‐Insulin‐Resistance (HOMA‐IR). Data were analyzed using linear mixed models and linear regression, adjusting for confounders. In addition, results were correlated to priorly published body composition measurement.

Results

Venous blood was collected from 87 (36%), 77 (31%), and 63 (26%) participants at ages 1, 2, and 6 years, respectively. No differences were found in metabolic markers between the formula groups or compared to the breastfed group at any time point. Furthermore, at 6 years of age, a positive correlation was found between some biomarkers (IGF‐1, leptin, and HOMA‐IR) and body composition measurements, but not all biomarkers showed such an association.

Conclusions

In this relatively small study, providing a modified, low‐protein infant formula during the first months of life did not affect hormonal and metabolic markers during early childhood.

What is Known

• High protein intake in infancy is associated with accelerated weight gain and increased risk of obesity.

• Early nutrition influences metabolic and hormonal regulation later in life.

• Insulin‐like growth factor‐1 (IGF‐1) and leptin are important biomarkers in childhood growth and metabolism.

What is New

• A low‐protein infant formula with modified amino acid composition did not affect metabolic or hormonal markers up to age 6.

• At 6 years, IGF‐1 and leptin levels correlated with body composition, but other biomarkers did not.

• The relationship between metabolic markers and growth outcomes appears to change over time.

1. INTRODUCTION

Early‐life nutrition plays a pivotal role in shaping growth, body composition, and long‐term health outcomes. ¹ While breast milk is the preferred source of food in early life, research on formula feeding evolves to provide the best alternative source of nutrition when needed. As such, protein intake in formula‐fed infants, which is in general higher compared to that of breastfed infants, has garnered significant attention due to its regulatory influence on infant growth and metabolism. ² , ³ The early protein hypothesis suggests that a high‐protein intake during infancy leads to increased levels of branched chain amino acids and higher insulin and insulin‐like growth factor 1 (IGF‐1) secretion, which promotes growth but also programs a higher risk of fat accumulation later in life. ⁴ , ⁵ Following this hypothesis, lowering protein intake in early life by modifying infant formula composition, offers a potential strategy to influence metabolic markers and growth trajectories, thereby mitigating the long‐term risk of obesity.

Key hormonal and metabolic markers, such as IGF‐1, leptin, and insulin, provide valuable insights into how early nutrition impacts growth and metabolism. IGF‐1 and its binding proteins (BP) are critical regulators of cellular growth and differentiation during infancy. ⁶ , ⁷ Similarly, rapid weight gain and adiposity during infancy is often accompanied by higher concentrations of leptin, a hormone integral to appetite regulation and energy balance. ⁸ , ⁹ , ¹⁰ Insulin and glucose, along with the homeostasis model assessment of insulin resistance (HOMA‐IR), further illuminate the connections between early nutrition and metabolic regulation. ¹¹ , ¹² Unfavorable programming of these pathways during critical developmental periods may contribute to adverse health outcomes, including obesity and metabolic disorders.

In our randomized controlled trial (RCT) (ProtEUs), we demonstrated that feeding a modified low‐protein infant formula (mLP; 1.7 g protein/100 kcal with an optimized amino acid profile) versus a conventional control formula (CTRL; 2.1 g protein/100 kcal) during the first 6 months of life was safe, supported adequate growth, but did not affect body composition up to the age of 6 years. ¹³ , ¹⁴ , ¹⁵ Furthermore, we observed no differences in metabolic and hormonal (IGF‐1 and its BP, leptin, insulin, glucose, and HOMA‐IR) markers at the age of 4 months between infants in the two formula groups. ¹⁶

Since the known programming character of early nutrition, the aim of the present study is to investigate the effects of early life protein intake on metabolic and hormonal profiles in early childhood. Specifically, we compare infants fed the mLP formula to those receiving the CTRL formula, by examining markers including IGF‐1, leptin, and other metabolic indicators until the age of 6 years. By investigating these potential underlying metabolic differences, we aim to determine whether the previously observed differences in early‐life protein intake influence metabolic programming through these hormones, ultimately altering the risk of obesity and metabolic disorders in later childhood.

2. METHODS

The original ProtEUs study investigated the safety of a low‐protein infant formula in healthy, term‐born infants. ¹³ In this RCT, formula‐fed infants were assigned within 45 days after birth to receive one of two noncommercial isocaloric formulas: mLP formula containing 1.7 g protein/100 kcal or CTRL infant formula containing 2.1 g protein/100 kcal. Apart from differences in protein content and amino acid composition, the infant formulas were identical, each containing 70% intact protein and 30% free amino acids. The precise amino acid composition is outlined in Table S1. A nonrandomized reference group of breastfed infants was included as well. The trial was conducted at Amsterdam UMC, VU University Medical Center, and the Dr. von Hauner Children's Hospital at Ludwig‐Maximilians‐Universität Munich. Between 2014 and 2016, 245 infants enrolled, of whom 178 were randomized to one of the formula groups, while 67 were in the breastfed reference group. The intervention period continued until the age of 6 months, during which safety, growth, and body composition were measured. Subsequently both groups received a standard follow‐on formula up to the age of 1 year. The study also incorporated long‐term follow‐up to assess growth outcomes at 1, 2, and 6 years of age, allowing for a comprehensive evaluation of the effects of early protein intake on longer‐term development. ¹⁴ , ¹⁵ Body composition was determined via air‐displacement plethysmography (ADP) using the BOD POD® Body Composition Systems (Cosmed). Detailed study procedures have been described previously. ¹³ , ¹⁴ , ¹⁵

Venous blood samples were collected in 2.5‐mL serum tubes and 0.5‐mL Heparin tubes at 4 months of age and during the follow‐up visits at 1, 2, and 6 years of age. The blood samples were collected in the morning after an overnight fast. After centrifugation (10 min at 1800 g, 20°C), each of the supernatants was divided into three aliquots. These aliquots were stored at −80°C and thawed only once, immediately prior to analysis. IGF‐1, IGF‐BPs, and leptin concentrations were measured in serum using ELISA (Mediagnost Inc.) according to manufacturer instructions. Glucose concentrations were assessed on a Beckman Coulter AU5800/AU680 (enzymatic UV‐assay) and insulin concentrations on a Roche Cobas 8000/e411 (electrochemiluminescence immunoassay [ECLIA]). Laboratory analyses were performed at the Institute of Laboratory Medicine, University Hospital, LMU Munich. The HOMA‐IR  ¹⁷ was calculated using glucose and insulin concentrations according to the following equation: (Glucose (mmol/L) × Insulin (μU/mL))/22.5. The same procedures were consistently followed at each follow‐up stage to ensure comparability of results.

Next to the aim of the assessment of potential differences in hormonal and metabolic markers between the study groups, we aimed to investigate potential associations between blood markers and growth outcomes at the different timepoints. Anthropometric measurements (weight, length, and BMI) and body composition assessments (fat mass [FM], fat‐free mass [FFM], and fat mass percentage [FM%]) were conducted during this study. Details regarding these procedures, results and differences between the feeding groups have been previously reported. ¹³ , ¹⁴ , ¹⁵

2.1. Ethics statement

The trial was approved by the Institutional Review Boards of the VU University Medical Center Amsterdam and the Medical Faculty of LMU Munich. Ethical approval was obtained from both participating institutions. The study adhered to the ICH‐GCP guidelines and the Declaration of Helsinki. The trial was registered in the Dutch Trial Register (Study ID NTR4829, trial number NL4677). Written informed consent was obtained from all parents or legal guardians for participation in both the intervention and follow‐up periods.

2.2. Statistical analyses

Normally distributed data were presented as mean ± SD and skewed data as medians (interquartile ranges; IQRs). As IGF‐1, IGF‐BP1, leptin, insulin, and HOMA‐IR were not‐normally distributed, concentrations were log transformed (natural log‐transformation) before statistical analysis.

The differences between the feeding groups in hormonal and metabolic markers were analyzed using linear mixed models (LMMs) and adjusted for sex and randomization hospital (Amsterdam or Munich). The markers measured with ELISA, IGF‐1, IGF‐BPs, and leptin, were additionally adjusted for the ELISA plate number, as three different batches were used. Since samples were randomly assigned to plates under blinded conditions, group allocation was not stratified by plate. As a result, some degree of inter‐assay variability may contribute to the observed variance. LMMs help dealing with longitudinal data by correlating measurements at different times within a subject via a random intercept at the subject level. LMMs also handle missing data by including all available data from the repeated measures (i.e., assessments at 4 months, 1 year, 2 years, and 6 years), even from participants who did not complete the full 6‐year follow‐up. We included sex as a covariate, as it is known to correlate with IGF‐1 response. ¹⁸ The hospital where participants were randomized was included as a covariate because location‐related factors, such as dietary habits and nutritional guidelines, could potentially influence the results.

Correlations between metabolic and hormonal marker concentrations at different time points (i.e., 4 months, 1 years, 2 years, and 6 years) were assessed using Pearson and Spearman correlation analyses. Pearson correlation was used for normally distributed variables (IGF‐BP2, IGF‐BP3, and glucose), while Spearman correlation was applied for non‐normally distributed variables (IGF‐1, IGF‐BP1, leptin, insulin, and HOMA‐IR).

The association between metabolic and hormonal markers at 1, 2, and 6 years of age and the anthropometric parameters and body composition were analyzed using linear regression analyses and adjusted for sex. As IGF‐1, IGF‐BP1, leptin, insulin, and HOMA‐IR were not‐normally distributed, concentrations were log transformed (natural log‐transformation) before statistical analysis. Statistical analyses were performed using IBM SPSS version 28 for Windows. ¹⁹ Two‐sided statistical significance was assumed at P‐values < 0.05.

3. RESULTS

In the current follow‐up study, blood samples were obtained from a total of 139 infants. Figure 1 illustrates the distribution of participants across the study groups and the number of successful blood withdrawals at each follow‐up time point. We found no differences between the feeding groups in baseline characteristics in infants in whom at least one blood sample was obtained, as shown in Table 1. Furthermore, the baseline characteristics of the subgroup of participants of whom at least one blood sample was obtained during the follow‐up period were not different from those of the entire sample during the intervention period (Table S2). Therefore, the follow‐up subgroup is regarded to be representative of the overall sample.

Figure 1.

Flowchart of the population with successful blood withdrawal.

Table 1.

Infant and maternal characteristics for participants with at least one blood sample collected during follow‐up.

	mLP (n = 56)	CTRL (n = 44)	BF (n = 39)
Male sex (%)	22 (40)	18 (42.9)	20 (50.0)
Gestational age (weeks)	39.3 ± 1.1	39.6 ± 1.1	39.8 ± 0.9
Birth weight (kg)	3.36 ± 0.34	3.36 ± 0.35	3.45 ± 0.30
Birth weight (SDS)	0.15 ± 0.70	0.15 ± 0.74	0.32 ± 0.58
Maternal BMI at study entry (kg/m2)	24.21 ± 4.20	25.05 ± 5.13	23.40 ± 3.64
Randomization Hospital Amsterdam (%)	41 (74.5)	36 (85.7)	24 (60)

Note: Infants in whom at least one blood sample was obtained during the follow‐up period were included in this table. Data are presented as mean SD or n (%). No differences are found between the feeding groups in characteristics.

Abbreviations: BF, breast‐fed; BMI, body mass index; CTRL, control formula; mLP, modified low‐protein formula; SDS, standard deviation score.

At ages 1, 2, and 6 years, no statistically significant differences were found between the two formula‐fed groups in any of the measured metabolic markers, IGF‐1, IGF‐BPs, leptin, glucose, insulin, or HOMA‐IR (Table 2).

Table 2.

Metabolic and hormonal markers in each feeding group at 1, 2, and 6 years of age and the differences between the feeding groups.

	mLP		CTRL		BF		mLP minus CTRL		mLP minus BF		CTRL minus BF
	n	Mean ± SD or median (IQR)	n	Mean ± SD or median (IQR)	n	Mean ± SD or median (IQR)	Diff (95% CI)	p	Diff (95% CI)	p	Diff (95% CI)	p
1‐year
IGF‐1 (ng/mL)	35	115 (75.2, 242)	22	110 (67.9, 237)	30	89.4 (58.5, 147)	−0.03 (−0.10 to 0.15)	0.693	−0.002 (−0.14 to 0.14)	0.982	−0.03 (−0.18 to 0.13)	0.741
IGF‐BP1 (ng/mL)	35	10.7 (6.83, 15.2)	22	8.33 (6.7, 13.5)	30	9.81 (6.94, 17.0)	0.03 (−0.15 to 0.22)	0.726	−0.11 (−0.32 to 0.11)	0.327	−0.14 (−0.37 to 0.09)	0.240
IGF‐BP2 (ng/mL)	35	636 ± 144	22	640 ± 176	30	661 ± 155	−0.08 (−70.3 to 70.2)	0.998	−26.4 (−106 to 52.3)	0.514	−26.3 (−114 to 61.0)	0.553
IGF‐BP3 (ng/mL)	35	2379 ± 400	22	2366 ± 443	30	2288 ± 470	−15.6 (−252 to 221)	0.897	74.4 (−195 to 344)	0.587	89.9 (−205 to 385)	0.549
Leptin (ng/mL)	35	1.79 (1.40, 2.33)	19	2.11 (1.44, 2.56)	30	1.69 (1.03, 2.39)	−0.08 (−0.23 to 0.07)	0.307	0.04 (−0.12 to 0.20)	0.632	0.12 (−0.06 to 0.30)	0.203
Glucose (mmol/L)	35	4.61 ± 0.37	22	4.65 ± 0.44		4.55 ± 0.39	−0.02 (−0.25 to 0.21)	0.841	0.09 (−0.12 to 0.30)	0.392	0.11 (−0.12 to 0.35)	0.339
Insulin (µU/mL)	35	3.90 (2.50, 6.90)	22	3.75 (2.30, 8.55)	30	3.8 (1.75, 6.1)	0.003 (−0.18 to 0.19)	0.974	0.06 (−0.11 to 0.23)	0.483	0.06 (−0.14 to 0.25)	0.554
HOMA‐IR	35	0.72 (0.49, 1.38)	22	0.68 (0.49, 1.73)	30	0.76 (0.35, 1.27)	0.002 (−0.20 to 0.20)	0.986	0.07 (−0.11 to 0.25)	0.451	0.07 (−0.14 to 0.28)	0.513
2 years
IGF‐1 (ng/mL)	27	223 (111, 332)	25	159 (95.7, 302)	24	113 (92, 171)	0.03 (−0.10 to 0.16)	0.634	−0.04 (−0.20 to 0.11)	0.595	−0.07 (−0.23 to 0.09)	0.378
IGF‐BP1 (ng/mL)	27	7.55 (4.40, 20.2)	25	6.02 (3.87, 11.1)	24	6.46 (5.0, 12.4)	0.07 (−0.13 to 0.26)	0.499	−0.02 (−0.26 to 0.21)	0.846	−0.09 (−0.33 to 0.16)	0.472
IGF‐BP2 (ng/mL)	27	446 ± 111	25	497 ± 111	24	468 ± 120	−41.1 (−113 to 31.1)	0.263	−18.0 (106 to 70.4)	0.689	23.1 (−68.4 to 115)	0.619
IGF‐BP3 (ng/ml)	27	2428 ± 405	25	2500 ± 493	24	2416 ± 484	−14.4 (−257 to 228)	0.907	14.6 (−283 to 312)	0.923	29.0 (−279 to 337)	0.853
Leptin (ng/mL)	26	1.63 (1.27, 2.40)	23	2.11 (1.12, 2.57)	24	1.77 (1.28, 2.70)	0.01 (−0.15 to 0.16)	0.911	−0.07 (−0.26 to 0.11)	0.433	−0.08 (−0.27 to 0.11)	0.398
Glucose (mmol/L)	28	4.72 ± 0.42	25	4.86 ± 0.68	24	4.73 ± 0.54	−0.17 (−0.40 to 0.07)	0.160	0.02 (−0.22 to 0.25)	0.899	0.18 (−0.06 to 0.42)	0.141
Insulin (µU/mL)	28	4.30 (2.95, 8.53)	25	5.40 (2.75, 14.50)	24	5.6 (3.6, 8.48)	−0.11 (−0.29 to 0.08)	0.274	−0.02 (−0.21 to 0.18)	0.873	0.09 (−0.11 to 0.29)	0.370
HOMA‐IR	28	0.90 (0.58, 1.85)	25	1.15 (0.53, 3.39)	24	1.15 (0.71, 1.65)	−0.12 (−0.32 to 0.09)	0.258	−0.01 (−0.22 to 0.19)	0.898	0.10 (−0.11 to 0.32)	0.336
6 years
IGF‐1 (ng/mL)	26	378 (255, 506)	17	435 (255, 496)	20	301 (214, 429)	−0.06 (−0.19 to 0.08)	0.423	−0.06 (−0.20 to 0.09)	0.440	0.00 (−0.15 to 0.16)	0.999
IGF‐BP1 (ng/mL)	26	11.5 (6.70, 15.7)	17	10.3 (5.40, 16.2)	18	11.6 (6.17, 14.3)	0.01 (−0.20 to 0.23)	0.899	0.05 (−0.17 to 0.28)	0.636	0.04 (−0.20 to 0.288)	0.745
IGF‐BP2 (ng/mL)	26	392 ± 113	17	399 ± 128	20	370 ± 118	−18.0 (−98.0 to 62.1)	0.659	17.8 (−64.0 to 99.5)	0.668	35.8 (−53.1 to 125)	0.429
IGF‐BP3 (ng/mL)	26	3289 ± 504	17	3331 ± 553	20	3432 ± 578	−58.2 (−327 to 210)	0.670	−123 (−398 to 152)	0.379	−65.0 (−364 to 234)	0.669
Leptin (ng/mL)	25	2.30 (1.71, 2.99)	17	1.22 (0.83, 2.82)	19	3.11 (0.84, 3.81)	0.17 (0.00 to 0.33)	0.050	0.004 (−0.17 to 0.18)	0.965	−0.16 (−0.35 to 0.02)	0.082
Glucose (mmol/L)	26	4.59 ± 0.34	17	4.71 ± 0.39	20	4.60 ± 0.25	−0.05 (−0.32 to 0.21)	0.686	0.58 (−0.20 to 0.31)	0.650	0.11 (−0.16 to 0.39)	0.423
Insulin (µU/mL)	26	4.20 (2.90, 7.65)	17	6.20 (2.65, 14.5)	20	4.80 (2.35, 9.83)	−0.03 (−0.25 to 0.18)	0.756	0.02 (−0.18 to 0.23)	0.825	0.06 (−0.17 to 0.28)	0.618
HOMA‐IR	26	0.86 (0.57, 1.51)	17	1.30 (0.53, 2.33)	20	1.00 (0.49, 1.95)	−0.04 (−0.27 to 0.19)	0.729	0.03 (−0.20 to 0.25)	0.811	0.07 (−0.18 to 0.31)	0.582

Note: Values are presented as mean ± SD or median (IQR). Differences are tested with linear mixed models and adjusted for sex, randomization hospital and ELISA plate number. Differences in insulin, HOMA‐IR, Leptin, IGF‐1, and IGF‐BP1 were log‐transformed before analysis and are presented as ratios.

Abbreviations: BF, breastfed; BPs, binding proteins; CI, confidence interval; CTRL, control formula; HOMA‐IR, Homeostatic‐Model‐Assessment‐of‐Insulin‐Resistance; IGF‐1, insulin‐like growth factor‐1; IQR, interquartile range; mLP, modified low‐protein formula.

Moreover, when comparing the formula‐fed groups with the reference breastfed group, no significant differences in any of the assessed metabolic parameters were observed at ages 1, 2, or 6 years either (Table 2).

IGF‐1 and its BPs showed consistent correlations across all time points (p < 0.001 and p < 0.05, respectively; data not shown), suggesting some stability over time. In contrast, leptin, insulin, glucose, and HOMA‐IR values did not show significant correlations between the different ages.

3.1. Association with anthropometry and body composition

At ages 1 and 2 years old, there were no associations between IGF‐1 and any of the assessed anthropometric outcomes. However, at the age of 6 years, IGF‐1 was positively correlated with all anthropometric (weight, length, and BMI) measurements (Table 3). Similarly, at the age of 2 years, there were no correlations between IGF‐1 and body composition outcomes, but at the age of 6, IGF‐1 was positively correlated with all body composition measurements (FM, FFM, and FM%) (Table 3).

Table 3.

Metabolic and hormonal markers at 1, 2, and 6 years in relation to anthropometric and body composition parameters measured at the same age.

Markers	Age (Years)	Weight, kg			Length, cm			BMI, kg/m2
		β	95% CI	p	β	95% CI	p	β	95% CI	p
IGF‐1 (ng/mL)
	1	0.46	−0.03 to 1.18	0.211	0.71	−1.19 to 2.61	0.457	0.44	−0.55 to 1.44	0.377
	2	0.61	−0.69 to 1.92	0.354	−0.68	−2.87 to 1.51	0.537	0.79	−0.56 to2.13	0.249
	6	8.69	5.53 to 11.84	<0.001	14.77	9.93 to 19.62	<0.001	2.00	0.43 to 3.58	0.014
Leptin (ng/mL)
	1	1.47	0.84 to 2.11	<0.001	2.32	0.50 to 4.16	0.015	1.62	0.70 to 2.53	<0.001
	2	1.34	0.21 to 2.47	0.021	‐0.59	−3.05 to 1.87	0.632	1.94	0.80 to 3.07	0.001
	6	8.53	6.15 to 10.9	<0.001	7.43	2.73 to 12.13	0.003	3.70	2.64 to 4.75	<0.001
HOMA‐IR
	1	0.06	−0.48 to 0.60	0.822	0.48	−0.94 to 1.90	0.501	−0.13	−0.84 to 0.58	0.711
	2	0.29	−0.43 to 1.00	0.428	−0.65	−2.07 to 0.77	0.363	0.60	−0.15 to 1.35	0.116
	6	3.47	0.88 to 6.06	0.010	5.65	1.64 to 9.67	0.007	0.86	−0.32 to 2.04	0.149

			Fat mass, kg			Fat free mass, kg		Fat mass percentage, %
IGF‐1 (ng/mL)
	2	−0.39	−1.79 to 1.02	0.583	0.19	−0.94 to 1.33	0.734	−2.91	−13.3 to 7.44	0.574
	6	2.16	0.74 to 3.56	0.004	5.98	3.77 to 8.18	<0.001	6.80	1.56 to 12.04	0.012
Leptin (ng/mL)
	2	0.71	−0.80 to 2.22	0.348	‐0.45	−1.66 to 0.77	0.461	5.53	−5.54 to 16.59	0.319
	6	2.88	1.81 to 3.94	<0.001	4.62	2.49 to 6.75	<0.001	8.19	3.74 to 12.64	<0.001
HOMA‐IR
	2	0.31	−0.43 to 1.05	0.403	0.20	−0.38 to 0.78	0.495	1.41	−4.03 to 6.86	0.604
	6	0.16	−0.91 to 1.24	0.759	3.01	1.20 to 4.82	0.002	−0.95	−4.89 to 3.00	0.632

Note: Values are associations between metabolic and hormonal markers and growth and body composition using linear regression analysis adjusted for sex and ELISA plate number. Bold values indicate statistically significant p‐values. β, coefficients.

Abbreviations: BMI, body mass index; CI, confidence interval; HOMA‐IR, Homeostatic‐Model‐Assessment‐of‐Insulin‐Resistance; IGF‐1, insulin‐like growth factor‐1.

Leptin concentrations at the ages of 1, 2, and 6 years were positively correlated with all anthropometric measurements (weight, length, and BMI), except for length at the age of 2 years (Table 3). At the age of 2 years, leptin was not correlated with body composition outcomes, but at the age of 6, leptin was positively associated with FM, FFM, and FM% (Table 3).

At the ages 1 and 2 years old, Homa‐IR was not correlated with any of the anthropometric or body composition outcomes. However, at the age of 6 years, HOMA‐IR was positively associated with weight, length, and FFM (Table 3).

4. DISCUSSION

With the degree of differences in protein intake and composition provided during the first 6 months of life in infants fed one of the two study formulas tested here, our study found no clear differences in concentrations of metabolic and hormonal markers during the follow‐up visits at 1, 2, and 6 years of age. This is in line with our previously reported findings, which showed no significant differences in growth and body composition outcomes between the formula groups at 17 weeks, 6 months, 1 years, 2 years, and 6 years of age. ¹³ , ¹⁴ , ¹⁵

As previously reported, metabolic and hormonal blood markers were analyzed at 4 months of age, during the intervention period. At that time point, no significant differences were observed between the mLP‐fed and CTRL‐fed infants. ¹⁶ However, in comparison to the breastfed reference group, both formula‐fed groups showed higher concentrations of IGF‐1, insulin, and HOMA‐IR. As demonstrated in the current study, these differences did not persist at later follow‐up time points, suggesting a transient effect rather than a lasting alteration in IGF‐1 regulation. This may, in part, be attributable to the smaller sample sizes at later follow‐up timepoints.

To further explore potential tracking of metabolic markers over time, we assessed correlations between marker concentrations across time points using both Pearson and Spearman analyses. Results revealed that not all markers were consistently correlated across ages, indicating limited stability of individual marker levels. For example, IGF‐1 and its BPs showed strong to moderate correlations between time points (e.g., 4 months, 1 year, 2 years, and 6 years), while others, such as leptin and insulin, showed weak or no correlations. However, it is important to note that available samples at each time point did not always come from the same participants. As a result, the correlations do not fully reflect within‐individual tracking over time but rather group‐level trends, which limits interpretation of individual‐level longitudinal consistency.

Our previously reported associations between metabolic and hormonal markers and growth and body composition outcomes at 4 months of age revealed that IGF‐1 and IGF‐BPs were primarily linked to weight, length, and FFM, suggesting their potential role in early growth regulation. ¹⁶ While these associations were not evident at 1 and 2 years of age, they were observed again at 6 years of age, with IGF‐1 showing a positive association with weight, length, FM, and FFM. Additionally, leptin displayed a similar positive association with these growth outcomes at 6 years. These markers may play a role in early growth regulation, but our follow‐up data did not show any significant differences in anthropometry, body composition, or metabolic and hormonal markers between feeding groups. Therefore, our findings do not show a direct long‐term relationship between the differences in protein intake and composition during the first 6 months of life as tested here, and these markers at the later ages of 2–6 years.

All ELISA measurements were performed using a 96‐well plate format, distributed across three plates (batches). Each plate included pooled quality control samples to assess analytical precision through coefficients of variation. Although some inter‐assay variability is expected, we observed substantial batch effects that could bias the results. As samples were randomly assigned to plates under blinded conditions, group allocation was not stratified by plate. To minimize potential bias, we statistically adjusted for plate number in all ELISA‐based analyses. This adjustment strengthens the validity of our findings.

Previous studies have also tested the effects of lowering the protein content in infant formulas on metabolic markers. For instance, a follow‐up study from the European Childhood Obesity Project (CHOP) demonstrated that infants fed a high‐protein formula (2.9 g/100 kcal for infant formula and 4.4 g/100 kcal for follow‐on formula) had significantly higher IGF‐1 concentrations at 6 months of age compared to those fed a low‐protein formula (1.8 g/100 kcal for infant formula and 2.2 g/100 kcal for follow‐on formula). ²⁰ The study also reported that IGF‐BP2 concentrations were lower in the high‐protein group, while IGF‐BP3 remained unaffected. Furthermore, it was shown that weight gain during the first 6 months of life was associated with IGF‐1 plasma concentrations, though this association was no longer evident after 6 months. ²⁰ On the other hand, the Early Protein and Obesity in Childhood (EPOCH) study investigated the effect of dietary protein on IGF‐1 concentrations, growth, and body composition in healthy term infants in a RCT and found contrary results. ²¹ Protein intake did not significantly influence IGF‐1 concentrations during the first 12 months of life. However, it did affect growth parameters such as length and head circumference, but not FM. Similarly, our results do not show indications that long‐term metabolic and hormonal markers may be significantly influenced by the differences in early protein intake tested here, although the differences in protein intake between the formula groups in our study were not as large as in the CHOP and EPOCH studies. Earlier studies have shown that IGF‐1 concentrations are higher in formula‐fed infants compared to breastfed infants during the feeding period. ²² , ²³ While elevated IGF‐1 levels in infancy have been linked to early obesity, this relationship appears to be age‐dependent and complex. ⁶ Further studies and reviews have suggested that early dietary exposures can program the IGF axis, with higher IGF‐1 levels in early life potentially associated with lower levels in adulthood, a pattern that may influence the risk of non‐communicable diseases later in life. ⁶

In the current RCT (n = 139), we observed that leptin concentrations were positively associated with growth and body composition parameters at 4 months, ¹⁶ 1 year, and 6 years of age, independent of group allocation. Similarly, a study in healthy term‐born infants (n = 197) investigating serum concentrations of appetite‐regulating hormones, found that formula‐fed infants had significantly higher leptin concentrations at 3 months of age compared to breastfed infants. ¹⁰ These leptin concentrations correlated positively with FM% at both 3 and 6 months of age. Additionally, a recent study by de Fluiter et al. examined the trajectories of appetite‐regulating hormones and their relationship with FM development in term‐born infants during the first 6 months of life. They observed that leptin was positively associated with FM percentage and its increase over time, suggesting its involvement in early adiposity programming. ²⁴ In our study, we observed a positive correlation between leptin and FM at 6 years of age.

While protein intake in infancy appears to affect short‐term growth and metabolic markers, such as IGF‐1, as observed in other studies, its long‐term impact on body composition and metabolic health remains less clear and may involve additional regulatory pathways. These findings highlight the need for further research to clarify the mechanisms linking early nutrition, hormonal regulation, and growth trajectories.

Due to attrition over time, the number of participants with available biomarker data declined at each follow‐up point, limiting the statistical power of our analyses. Post‐hoc calculations showed low power to detect small effects (10%–12%) and moderate power (39%–52%) for medium effects, depending on the time point. Thus, the study was underpowered to detect subtle group differences, particularly at later stages such as 6 years. While the original trial was adequately powered for early outcomes, reduced follow‐up sample sizes constrained detection of longer‐term effects. However, to address missing data, we used LMMs to efficiently use available data assuming values are missing at random. Although multiple imputation could be considered, it does not improve power and is less stable in small samples. These limitations should be considered when interpreting non‐significant results. Future studies with larger follow‐up cohorts are needed to confirm these findings.

A strength of this study is that all samples collected at 1, 2, and 6 years were analyzed simultaneously, minimizing variation in measurement procedures and enhancing data consistency and reliability. The long follow‐up period further strengthens the study by enabling a more comprehensive evaluation of long‐term outcomes. However, several limitations must be acknowledged. The sample size was relatively small, with considerable dropout due to COVID‐19‐related restrictions in both hospitals and the community, potentially introducing attrition bias and limiting generalizability. Difficulties in obtaining blood samples during infancy and early childhood, such as parental refusal, child distress, or procedural resistance, also contributed to the reduced sample size. To minimize participant burden, no second blood withdrawal attempts were made following an initial failure. Consequently, not all children provided samples at each time point, and samples were not always consecutive or from the same participants, limiting the ability to assess individual‐level longitudinal trends. However, LMMs handles missing data by including all available data from the repeated measures, even from participants who did not have all samples collected. An additional limitation is the absence of data on nutritional status and dietary intake during follow‐up, which precluded adjustment for these potential confounders.

5. CONCLUSION

Children who were fed a low‐protein infant formula with a modified amino acid composition during the first 6 months of life did not show substantial alterations in their metabolic profile at follow‐up ages up to 6 years.

CONFLICT OF INTEREST STATEMENT

Chris H. P. van den Akker reports receipt of speakers and consultancy honoraria from Nestlé Nutrition Institute, Nutricia, and Baxter; used as research funds. Johannes B. van Goudoever is founder and director of the National Human Milk Bank, member of the National health Council. RIVM (National Institute for Public Health and the Environment) funds a project determining PFAS levels in human milk and serum of the lactating women. Stefanie M. P. Kouwenhoven reports receipt of speakers and consultancy honoraria from Nestlé Nutrition Institute and Nutricia; used as research funds. Marieke Abrahamse‐Berkeveld is an employee of Danone Research & Innovation. She had no role in the execution of the study or in the statistical analyses of the results. LMU and its employees Nadja Antl and Berthold Koletzko benefitted from support for scientific and educational activities from Bayer, Barilla, Danone, DGC, DSM, Hipp, Nestlé, and reckitt. The remaining authors declare no conflicts of interest.

Supporting information

Supplemental Table S1. Amino acid composition of the study formulas. Values are as mg either per 100 mL/per 100 kcal. mLP, modified low‐protein; CTRL, control. *Conditionally essential.

Supplemental Table S2. Patient characteristics of the original cohort and infants with at least one blood sample collected during follow‐up. Data are presented as mean SD or n (%). No differences are found between the populations.

Muts J, Kouwenhoven SMP, Antl N, et al. Metabolic and hormonal serum markers in early childhood: follow‐up of a randomized controlled trial during infancy evaluating a modified, low‐protein infant formula. J Pediatr Gastroenterol Nutr. 2026;82:1108‐1118. 10.1002/jpn3.70341

DATA AVAILABILITY STATEMENT

Data are described in the manuscript, code book and analytic code will be made available upon request.