Evidence for protein leverage in a general population sample of children and adolescents

Childhood overweight and obesity are major global health concerns due to their psychosocial and cardiometabolic consequences and tracking into adulthood. Total energy intake relative to expenditure drives weight gain, modulated by genetic, epigenetic, and physiological factors. Appetite systems for protein, carbohydrates, and fats regulate intake toward macronutrient targets in healthy food environments; in unbalanced environments (e.g., high ultra-processed food availability), relative macronutrient appetites can lead to dysregulated energy intake. The protein leverage (PL) concept, supported in animals and adult humans, proposes that protein intake is tightly regulated, so low-protein diets prompt increased total energy intake to meet protein targets. The protein leverage hypothesis (PLH) posits that this mechanism contributes to overconsumption and obesity in real-world settings. Evidence in children is limited, especially in population samples. This study aimed to test for PL and evaluate whether higher energy intake on lower-protein diets is associated with adiposity in a general population cohort of Finnish children followed into adolescence.

Prior research using the nutritional geometry framework shows protein intake is more tightly regulated than carbohydrates or fats in animals and adults, leading to PL on lower protein diets. Adult human trials and ecological analyses (e.g., ultra-processed food consumption linked to protein dilution and higher energy intake) support PL/PLH. Evidence in pediatric populations is sparse, with one cross-sectional study in youths with severe obesity indicating PL. The study builds on this literature by examining a population-representative pediatric cohort across three ages and accounting for macronutrient mixture effects.

Design and cohort: Data from the Physical Activity and Nutrition in Children (PANIC) study, an 8-year, single-center, controlled trial in a general population sample from Kuopio, Finland. Assessments at baseline (T0; age 6–8 years), 2-year follow-up (T1; ~10 years), and 8-year follow-up (T2; ~16 years). Ethics approved (Statement 69/2006); parental consent and child assent obtained. Participants: T0 n=422 (mean age 7.6 years), T1 n=387 (mean 9.8 years), T2 n=229 (mean 15.8 years). Measures: - Anthropometry and body composition: Weight, height, waist circumference, BMI, BMI z-score (Finnish references), DXA-derived total and percent fat and lean mass. Pubertal stage assessed by Tanner criteria. - Diet: Consecutive 4-day food records; total energy intake (TEI, kcal), absolute macronutrient intakes (g), and proportional energy from protein (%EP), carbohydrate (%EC), and fat (%EF) computed via Micro Nutrica v2.5. - Physical activity and sedentary time: Questionnaires and combined heart rate–movement sensor (Actiheart) in sub-analyses. - Total energy expenditure (TEE): Estimated from DXA lean mass (LM) and fat mass (FM) using Pontzer et al. equation: ln(TEE) = −0.121 + 0.696 ln(LM) − 0.041 ln(FM). Statistical analysis: - Descriptive statistics (means, SDs; counts, %). Group comparisons by Student’s t tests (intervention vs control). - Associations of TEI with adiposity (BMI z-score, WC, %LM, %BF) and with TEE assessed by multiple linear regression adjusted for age and sex; standardized beta coefficients with 95% CIs and P values reported. - Testing for protein leverage: Fitted power functions relating log(TEI) to log(%EP), with exponent L indicating leverage: log(TEI) = log(P) + L·log(p), where p is proportion of energy from protein; L between −1 and 0 indicates partial PL. Similar models for %EC and %EF. Models adjusted for potential confounders: fiber intake, physical activity, sedentary time, age, and TEE; variables normalized. - Effect modification/subgroup analyses: Tested interactive or additive effects of sex, pubertal stage, and study group (intervention vs control) on PL. - Mixture modeling: To account for covariance among macronutrients, applied mixture models (R mixexp package, MixModel) using %EP, %EC, %EF as compositional predictors of TEI, comparing five models to a null using AIC for selection; adjustments for normalized fiber, age, TEE, physical activity, sedentary time. Visualized with right-angled mixture triangles. - Additional analyses: Used Actiheart-derived activity data in adjusted power models; assessed impact of under/over-reporting by TEI vs TEE strata. - Two-tailed tests; significance P<0.05; RStudio v1.1.453.

- Evidence for protein leverage at all ages: Proportional energy from protein (%EP) inversely associated with TEI following power functions with leverage exponents L: • T0 (~8y): L = −0.36 (95% CI −0.47 to −0.25), P<0.001 • T1 (~10y): L = −0.26 (−0.37 to −0.15), P<0.001 • T2 (~16y): L = −0.25 (−0.38 to −0.13), P<0.001 - Other macronutrients (power models): • %EC inversely associated with TEI at T0 (L = −0.19, P=0.029) and T1 (L = −0.17, P=0.045), not at T2 (P=0.31). • %EF positively associated with TEI at all ages: T0 L = 0.26 (P<0.001); T1 L = 0.22 (P<0.001); T2 L = 0.22 (P=0.002). - Mixture modeling: Variance in TEI primarily associated with %EP (inverse), not with %EC or %EF; response surfaces showed gradients aligned with the protein axis. - TEI and energy expenditure/adiposity: • TEI was positively associated with TEE at all ages (all P<0.001). • TEI was not associated with BMI z-score or waist circumference at any age. • At T2, TEI positively associated with %lean mass and inversely with %body fat; no associations at T0 and T1. - Cohort descriptives: • TEI increased with age: mean (SD) 1620 (303) kcal at T0; 1670 (341) kcal at T1; 1820 (539) kcal at T2. • TEE increased with age: 1660 (126) kcal at T0; 1860 (150) kcal at T1; 2730 (385) kcal at T2. • BMI z-scores near zero across ages (−0.19, −0.13, −0.03). - Subgroup findings: • Male sex associated with higher TEI (additive effect) at all ages; no interaction with %EP. • Pubertal status and study group (intervention vs control) not associated with TEI and did not modify PL. • Results robust when adjusting for Actiheart-derived activity and across TEI relative to TEE strata (under/over-reporting).

Findings support the presence of protein leverage in a general pediatric population: lower dietary protein proportion is associated with higher total energy intake, and mixture models confirm protein as the primary macronutrient explaining TEI variance. Despite higher TEI on lower-protein diets, adiposity did not increase, likely due to compensatory increases in total energy expenditure, indicating energy balance in this cohort. The lack of association between TEI and BMI z-score may also reflect underreporting of TEI, as suggested by TEE/TEI ratios >1, and the relatively healthy diet and weight status of the cohort. The results align with PL/PLH theory and imply that in environments with protein dilution (e.g., high ultra-processed food intake), PL could contribute to increased adiposity in youth. The study underscores the importance of considering macronutrient composition, not just calories, and using mixture modeling to disentangle compositional effects. Subgroup analyses show higher TEI in boys but no modulation of PL by sex, puberty, or intervention participation.

This population-based cohort study provides the first evidence of protein leverage among children and adolescents, with consistent inverse associations between dietary protein proportion and energy intake across three ages. In this generally healthy-weight cohort, increased energy intake on lower-protein diets was offset by higher energy expenditure and did not translate into greater adiposity. Future research should examine cohorts with higher exposure to ultra-processed, protein-diluted diets, employ longitudinal designs with shorter assessment intervals, and further assess developmental factors such as puberty in relation to PL.

- Cross-sectional analyses at each timepoint limit causal inference regarding changes in TEI and adiposity; more frequent longitudinal assessments are needed. - Cohort had relatively low prevalence of overweight/obesity and a comparatively healthy food environment, potentially limiting generalizability to settings with higher ultra-processed food exposure. - Underreporting of dietary intake likely (TEE/TEI >1), particularly in adolescents, may attenuate observed associations. - Sub-analyses for puberty effects involved smaller subsamples, reducing power.