Nurture growth: Ketogenic diet therapy and growth velocity in infants under 12 months with epilepsy (cid:0) A systematic review and infant data study

Ketogenic diet therapy (KDT) is well established for the treatment of early epileptic encephalopathies and specific aetiologies; however, the impact on growth in infancy remains controversial. Our aim was to examine the influence of early KDT on growth velocity and height percentiles completing two tasks. First, we systematically reviewed the literature on growth in infants younger than 12 months. Second, we analysed data from our prospective database, including infants < 12 months (n = 63) treated with KDT. The literature review (n = 7) remains descriptive and includes growth percentiles and z-scores as growth velocity was not described. Studies up to 2010 used fasting, calorie restrictions, and ratios > 3:1. In individual cases, significant growth delays were found; other authors did not find any changes in growth parameters. Study endpoints in our own cohort included z-scores of growth velocity, standard deviation (SD) of height, weight, BMI, deviation from individual height percentile, and daily macronutrient intake. The median z-score of growth velocity was 1.03 (first year of life). After three months, median daily intake of protein and energy was 1.68 g/kg and 85 kcal/kg. Until the age of one year, neither growth velocity nor individual growth percentiles decreased. Infants showed distinct growth improvements at three months, likely due to continuous nutritional monitoring and reduction in seizures. In the second year of life, z-scores of growth velocity decreased in patients still receiving KDT (from 1.03 at 12 months to (cid:0) 1.5 at 24 months). Furthermore, younger age at epilepsy onset and at KDT start correlated with slower growth velocities in the first year of life. With appropriate nutritional intake and monitoring, KDT does not reduce growth in the first year of life. Future directions might be to study the impact of KDT on growth velocity and growth hormones throughout childhood.


Introduction
Ketogenic diet therapy (KDT) is characterised by a high-fat, adequate protein, and low-carbohydrate intake, which is established for early epileptic encephalopathies and drug-resistant epilepsy [1][2][3].
Despite its proven effectiveness, the precise mechanisms of action by which KDT exerts its anticonvulsive and neuroprotective effects remain elusive [4].KDT likely acts through a combination of mechanisms that target fundamental biochemical pathways associated with cellular substrates such as ion channels and other mediators reducing neuronal hyperexcitability.
Due to the restrictive nature of KDT, supplementation of vitamins, minerals, and trace elements is mandatory, which overall places high demands on the interdisciplinary ketogenic diet team, infants, and their families, when calculating, preparing and adhering to ketogenic foods [5].
Data on growth during KDT in infants < 12 months are scarce and often come from paediatric age cohorts [16,17].Infancy, which lasts until the end of the 12th month, represents the fastest phase of postnatal growth: at five months, the birth weight typically doubles, at twelve months it triples, and the height increases by about 50 % [18].Dietary restrictions in the first two years of life can impair or delay an individual's physical and cognitive development and have long-term health consequences [18].Growth velocity, as an early marker of adequate nutrition for the first two years of life, and measured at least 12 months apart [19], was studied by two teams only, and in children on KDT and not in infants.These studies showed that KDT reduced growth velocity, especially when higher ratios were used [8,20].
The World Health Organization (WHO) introduced growth velocity standards in 2009 [19,21], allowing for scoring of weight and height gain to be assessed based on age and sex.
We therefore wanted to study comprehensively the effect of KDT on growth in the first two years of life and to identify possible changes in growth velocity in infants with KDT and their causes.The aim of the literature review was to clarify existing data on growth velocity, and the aim of our infant study was to study growth velocity in our own cohort.
Our study aims were: To conduct a systematic literature review of the literature to identify existing data on growth velocity.
To assess z-scores of growth velocity from birth to first birthday and from first to second birthday during KDT in our infant cohort.
To assess height, weight, body-mass-index (BMI), deviation from individual height percentile (>1SD), daily energy intake, daily macronutrient intake, fat/non-fat ratio, ß-Hydroxy-butyrate level, seizure reduction in % in our infant cohort.The hypothesis to be investigated in our cohort was, whether KDT in infants reduced growth velocity during the first two years of life and whether higher fat/non-fat ratios had a negative impact on growth.

Literature review
A systematic literature search was carried out according to the "Preferred Reporting Items for Systematic Reviews and Meta-Analyses" (PRISMA) guidelines [22].Pubmed and Scopus databases were searched using the following keywords: "ketogenic diet" AND "infant" AND "growth".Inclusion criteria were: 1) written in English language, and 2) KDT in infants including those under 12 months of age.No date restrictions were applied.The studies were screened for eligibility.Levels of evidence are displayed in Fig. 2. Case reports, reviews, duplicates, articles with insufficient data or not including infants < 12 months were excluded.The selection process for this research is shown in Fig. 2.

Study design of the infant cohort
The study was a prospective cohort study of all consecutive infants receiving KDT from September 1999 to July 2022.Data were retrieved from a longitudinal and prospective database, including all children who were consecutively treated with KDT at the Department of Paediatrics of the Medical University of Vienna; the study was approved by the institutional ethics committee (No. 1391/2013).
Data includes patients' demographics, medical history, seizure outcomes, growth, and nutritional outcomes, as well as tolerability of KDT and laboratory examinations.

Study population
Inclusion criteria were (1) infants younger than 12 months with an ascertained diagnosis of epilepsy treated with KDT, and (2) written informed consent of legal guardians.

KDT treatment protocol and follow-up visits
KDT was introduced gradually according to the Johns Hopkins protocol at a fat/non-fat ratio (ratio) of 1:1 without fasting and fluid restriction and increased individually up to a maximum of 3:1 ratio (ratio was modified when beta-hydroxybutyrate reached levels > 5 mmol/l).Age-appropriate protein intake was considered [23].For the energy and protein intake, guideline values for the German-speaking countries were taken into account: (for the first month of life 91-94 kcal and 2.5 g of protein per kg of bodyweight; from second to the fourth month recommended intake is 90-91 kcal and 1.3 g of protein per kg of bodyweight, and from five to twelve months intake 72-91 and 1.3 g of protein per kg of bodyweight) [24].
At baseline, all infants underwent a medical and metabolic work-up.Follow-up examinations were performed at one, three, six, nine, 12 and 24 months.At each visit, Video-EEG, laboratory parameters, paediatric, nutritional, and neurological examinations were performed, and seizure diaries were kept.
In addition, seizure frequency, age at KDT start, age at epilepsy onset, and duration of KDT during the first year and the second year of life (if used) were documented.The effectiveness of KDT was assessed as 1) seizure reduction > 50 % (responder) and 2) seizure freedom compared to baseline.

Anthropometric measurements
Height, weight, and daily nutritional intake were assessed at each time point.Growth velocity was calculated in centimetres from birth to first birthday and from first to second birthday.To calculate growth velocity, reference values from the German-speaking countries were used [25,26].The calculation of the z-score is shown in Fig. 1 − as standard deviation (SD) of growth velocity [25,26].
Primary endpoint was the median z-score of growth velocity in the first year of life compared to reference values from the literature.
Secondary endpoints were: z-scores of growth velocity during the second year of life for those still receiving KDT, height, weight, and BMI, growth deviation from individual height percentile, daily energy intake (kcal/kg bodyweight), daily macro-nutrient intake (g/kg bodyweight), fat/non-fat ratio, ß-Hydroxy-butyrate level, and seizure frequency at each time point.

Data analysis
Data were analysed using the IBM Statistical Package for Social Fig. 1.Calculation for z-score as standard deviation (SD) of growth velocity Calculation for growth velocity (SD=standard deviation of growth velocity; GV=growth velocity; cm = centimetres; y = year; M (t) = median of age and S (t) = standard deviation [24,25].
A. Maass et al.

Science (SPSS Statistics Version 25
).The significance level was set at p ≤ 0.05.For descriptive statistics, absolute numbers, percentages, median, minimum, and maximum values were used.For comparison of mean z-scores between different time points student's t-test was used and unpaired t-test for group comparison (sex).Further, correlation analyses between z-score of growth velocity of the 1st year of life with ß-Hydroxy-butyrate, daily macronutrient intake, and fat/non-fat ratio at each time point were performed.For nominal and ordinal parameters, the eta coefficient was analysed.

Literature review
Fig. 2 shows the literature screening process.A total of 142 publications were identified in Pubmed and Scopus and screened for eligibility (n = 1.228 patients).42 studies had to be excluded because no data on growth were reported.A further 18 studies were excluded as they were case reports, as well as a further 22 reviews and 13 duplicates.Of the remaining 47, further studies had to be removed: n = 15 did not contain sufficient data on KDT initiation and KDT management, n = 9 did not contain data on energy intake.The remaining 23 studies on growth in early child hood were screened for eligibility as they reported z-scores of height and weight.However, protein intake, epilepsy syndromes, seizure types, anti-seizure medications (ASM), and level of ketosis were not available for all.Eight of the 23 studies reported significant growth delays due to KDT including infants and children > 12 months [6][7][8][10][11][12][13][14]: Vining and Neal both reported an initial decrease in weight z-scores followed by a slowing of growth.Although Peterson observed weight and height decreases over the same period, only individual cases of infants younger than 12 months were included.As n = 16 did not include infants < 12 months, final data included seven studies.These are displayed in Table 1 (infants < 12 months and growth data; n = 345 infants).Table 1 shows data from two retrospective studies and five prospective studies (including 2 RCTs).13/345 (3.7 %) showed growth deficits in all studies.When calculating the odds ratios for studies up to 2010, the risk of a growth deficit was slightly higher albeit not statistically significant (4.5 % vs. 3.2 %, OR 1.415; CI 95 % 0.46-4.30).However, studies up to 2010 included protocols with fasting, calorie restriction, and fat/non-fat ratios > 3:1.Overall, most infants (96.3 %) showed adequate growth.Growth velocity has not been reported in literature.

Prospective study 3.2.1. Patient characteristics
From March 1999 to January 2023, 247 children started KDT at our centre.23 of them (9.3 %) dropped out due to non-compliance with the dietary intervention within the first two weeks.Of 98 children under months of age, 63 infants had complete data sets and could be analysed (Fig. 3).An overview of baseline characteristics is presented in Table 2.The median age at KDT start was six months (range from 0 to 11), the median age at epilepsy onset was 2.9 months (range 0 to 9.6), and n = were males.Five infants (8 %) with difficulty in oral feeding were supplied with feeding tubes.At KDT start, the median seizure frequency was 120 seizures per month (range from 1 to 3600).

Primary endpoint (Table 4)
The median z-score of growth velocity was 1.03 for the first year of life (range − 7.01 to 6.2) (Table 4).In our cohort, z-scores of growth velocity were higher than those reported in literature [25,26].

Secondary endpoint (Table 4)
In the second year of life, z-scores of growth velocity were lower than those reported in literature.However, the difference between minimum and maximum was smaller and less heterogeneous compared to the first   year of life.

Correlation analysis (Table 5A and 5B)
In the first year of life, significant positive correlations with growth velocity were observed for age at KDTstart, age at epilepsy onset, protein intake at start, and protein intake at three months, whereas a negative correlation was observed for energy intake at six months.For the second year of life, no significant correlations were observed.

Median z-scores of height, weight, and BMI during the first 12 months of KDT (Fig. 4)
A significant increase in height was observed from KDT start to third month of therapy.A slight increase in weight was also observed during this period which stabilised over time.

Median z-scores of height, weight, and BMI from birth to second birthday (Fig. 5)
In the first and second years of life, the median z-scores of height, weight, and BMI increased slightly (not statistically significant).No significant correlation between KDT duration was observed in either the first (n = 53) or the second (n = 35) year of life.

Deviation from height percentile during KDT in percent (Fig. 6)
Fig. 6 shows the percentage of infants who demonstrated a deviation from their individual height percentile during KDT.The majority remained at their individual percentile between different measurements.At three months (n = 63), seven infants (11 %) demonstrated a decline from their height percentile.None of the infants left the normal range.In one infant, the value decreased by more than 2SD at baseline and normalised during KDT.Six infants started and remained within the normal range.At six months (n = 62), nine infants (14 %) demonstrated a decline from their height percentile.Three infants were outside the normal range, and two of them showed inadequate weight despite adequate energy and protein intake.Three infants whose height was higher than 2SD at three months, decreased to the normal range at six months.In addition, three infants showed a slight decrease of 1SD and remained within the normal range.At 12 months (n = 60), 10 infants (17 %) demonstrated a decrease in their height percentile, with one infant outside the normal range.Of four infants who had shown an increase above 2SD at six months, two remained above 1SD, and two normalised at 12 months.In five infants the decrease of 1SD remained within the normal range.
All infants who had a decrease in height suffered from a genetic or metabolic encephalopathy.

Macronutrients at start, 3, 6, and 12 months of KDT (Supplemental Fig. 1)
Macronutrients at KDT start, after three, six, and 12 months were calculated by g/kg of bodyweight.Over the course of KDT, protein intake decreased slightly, whereas fat and carbohydrate intake increased with age (as appropriate).

Discussion
The systematic literature review aimed to summarise the available knowledge on growth of infants under 12 months on KDT, which yielded mixed results.Studies up to 2010 still used fasting, calorie restrictions, and higher ketogenic ratios.Growth delays were reported in individual patients and has decreased in recent studies.However, the majority did not detect any changes in growth parameters.Growth velocity was not examined.Due to insufficient data on influencing factors such as protein intake, aetiology, and degree of ketosis, only a few of studies could be included and no definite conclusion could be drawn on growth of infants on KDT.
The aim of the prospective study was to assess growth velocity in infants younger than 12 months receiving KDT in the treatment of early epileptic encephalopathies.Growth velocity increased during the first year of life under KDT, but decreased slightly after the first birthday.Serious growth delays or weight loss were rare, except in one infant who died due to unknown epileptic encephalopathy.Moderate weight due to metabolic or genetic causes affected 16 % of infants.After three months of KDT, height increased significantly, but no significant correlations were found between growth velocity and KDT duration, level of ketosis, fat/non-fat ratio, or macronutrient intake.Younger age at epilepsy onset and a deviation from the individual growth percentile correlated with reduced growth.Overall, nutritional status was maintained or improved one year after KDT start.
When calculating growth velocity, which provides information about

Table 4
Primary endpoint median Z-Scores of growth velocity (1st and 2nd year of life) *.  regular growth, at least 12 months should be taken into account for infants [19].This study is the first to focus on growth velocity during KDT in the first 24 months.To date, only changes in z-scores or in absolute numbers have been reported [27][28][29].El Rashidy reported increased height when switching from the classic KDT to the modified Atkins diet (MAD) [30].Armeno observed no significant changes in height and weight development in most children, nutritional status was maintained or improved, including all paediatric age groups (mean age 6.6 years; range 0.8-17. 3) [16] and in younger ones (mean age 12.2 months, range 1.7-25.9months) [17].Lu found that height z-scores decreased with a longer duration of KDT, which was even more pronounced when children were younger, started a classic KDT early, and were non-ambulatory [15], although the number of infants younger than 12 months was not reported.With our study, we confirm these data: Growth velocity decreased slightly when KDT was used also in the second year of life, and a younger age at epilepsy onset was also  associated with reduced growth.In contrast to literature, we also analysed timelines in terms of infants' age not just KDT duration in months, and included deviation from individual height percentiles.Growth mediators were analysed in two studies [31,20]: Spulber reported reduced growth hormones (IGF-1), reduced growth velocity, and reduced weight gain during KDT, which correlated to the degree of ketosis (n = 22).Despite adequate growth, Svedlund observed a reduction in growth hormones (IGF-1, IGF-BP3) [31].A possible mechanism could be that KDT reduces growth hormones by reducing ghrelin levels [32], but no infants were analysed.
After one year of KDT, nine children showed lower energy intake than recommended, but only one child had a reduced height of − 2SD, and all of them were affected by genetic or metabolic encephalopathies.Energy intake is difficult to compare, because most studies did not report daily calorie intake, but rather percentages of recommendations which have changed frequently over the past decades [27][28][29][30]33].
The lowest required levels of ketosis (>2 mmol/l − 3 mmol/l) and lower fat/non-fat ratios (<3:1) have been reported [2,16,23] which did not reduce effectiveness.We suggest that this is protective factor for adequate growth via the ghrelin and growth hormone pathway [32] and a possible future direction.
Our study highlights that well-calculated KDT improves nutritional status in infants with severe and difficult to control epilepsy in the first year of life and that growth velocity is not reduced when KDT is used only in the first year of life.Implications are that we should tailor our therapies as already shown in infantile epileptic spasm syndrome (IESS) where KDT is effectively used for a shorter period of six to nine months [34], and that we should use lower ratios [23].
The tremendous improvement in growth after three months of KDT is probably due to the reduction in seizures, as well as to close dietetic monitoring and exact nutritional calculation [16], but also to high protein intake.Energy and protein intake in our work was slightly higher than in the infantile guidelines [1] and within recommendations [23].
Therefore, younger infants receiving KDT need to be monitored even more closely to avoid negative effects on growth.Future studies need to examine the role of modified and individualised KDT including lower ratios and lower levels of ketosis, as well as to analyse growth hormones and ghrelin.
One limitation is that commonly used data to calculate growth velocity are WHO data [21], which are based on a relatively small cohort of diverse ethnic backgrounds.Therefore, we used in our study two comparable reference datasets of the German speaking countries [25,26], which included neuro-typical children, but only a small number of infants below 12 months.
The strengths of our study were a large group of infants younger than 12 months, reporting growth velocity, deviation of individual height percentiles, z-scores, ketosis, fat/non-fat ratios, detailed macro-nutrient and energy intake in g/ kg of bodyweight, and timelines for the first and the second year of life.

Conclusions
A well-calculated ketogenic diet therapy does neither reduce growth velocity nor individual height percentiles during the first year of life, especially if protein and energy intake are closely monitored.
Infants with severe and difficult to control epilepsy benefit from close nutritional monitoring during the first three months of KDT showing a tremendous improvement in growth, possibly due to the cessation of seizures.However, younger age at epilepsy onset, and KDT use beyond the first year of life, however, appear to slow down growth.

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

*
Data are displayed at first and at second birthday; z-scores of growth velocity were calculated according to Brandt and Hesse.**median, minimum-maximum, (n = 63) for the 1st year and 8 N=35) for the 2nd year of life.

Fig. 5 .
Fig. 5. Mean z-scores of height, weight, and BMI from birth to second birthday The medium line of the boxplot represents the median, boxes represent 25% and 75% percentiles, respectively.Whiskers display the minimum and the maximum.Circles display outliers.

Table 2
Baseline characteristics at KDT-start.

Table 5A
Correlation analysis of growth velocity throughout the 1st year of life.

Table 5B
Correlation analysis of growth velocity throughout the 2nd year of life.Z-scores of growth velocity were calculated according to Brandt and Hesse. *