Similarities between acylcarnitine profiles in large for gestational age newborns and obesity

Large for gestational age (LGA) newborns have an increased risk of obesity, insulin resistance, and metabolic syndrome. Acylcarnitine profiles in obese children and adults are characterized by increased levels of C3, C5, and certain medium-chain (C12) and long-chain (C14:1 and C16) acylcarnitines. C2 is also increased in insulin-resistant states. In this 1-year observational study of 2514 newborns (246 LGA newborns, 250 small for gestational age (GA) newborns, and 2018 appropriate for GA newborns), we analyzed and compared postnatal acylcarnitine profiles in LGA newborns with profiles described for obese individuals. Acylcarnitine analysis was performed by tandem mass spectrometry on dried-blood spots collected on day 3 of life. LGA newborns had higher levels of total short-chain acylcarnitines (p < 0.001), C2 (p < 0.01) and C3 (p < 0.001) acylcarnitines, and all C12, C14, and C16 acylcarnitines except C12:1. They also had a higher tendency towards carnitine insufficiency (p < 0.05) and carnitine deficiency (p < 0.001). No significant differences were observed between LGA newborns born to mothers with or without a history of gestational diabetes. This novel study describes a postnatal acylcarnitine profile in LGA with higher levels of C2, C3, total acylcarnitines, and total short-chain acylcarnitines that is characteristic of childhood and adult obesity and linked to an unhealthy metabolic phenotype.

Large for gestational age (LGA) is defined as a birth weight above the 90 th percentile for the corresponding gestational age (GA) 1 . The maternal factors most closely associated with LGA are maternal obesity, excessive gestational weight gain, maternal gestational diabetes mellitus (GDM), pregestational obesity, and maternal stress 2-9 . Fetal factors associated with LGA consist primarily of genetic or chromosomal disorders. LGA prevalence is estimated at between 4.6% and 15.3% 10 and is influenced by ethnicity, with higher rates found in children born to African American and non-Hispanic Asian American women in U.S. studies 11,12 .
Excessive fetal growth has negative consequences that extend beyond the neonatal period and these include medium-and long-term neurological, behavioral, and cardiovascular impacts 13-16 . LGA newborns are also at an increased risk of obesity [17][18][19][20] , metabolic syndrome 21,22 , and insulin resistance 23 in later life. They have also been found to have elevated leptin and fasting insulin and homeostasis model assessment (HOMA) index levels 24 during childhood, in addition to elevated adiponectin levels 25 , despite previous reports to the contrary from other studies of insulin states. The risk of obesity in LGA newborns increases with co-occurrence of maternal overweight/obesity or diabetes mellitus 26 .
Dysregulation of fatty acid oxidation and subsequent lipotoxicity play an important role in the pathophysiology of obesity-induced insulin resistance 27,28 . Analysis of acylcarnitine profiles by tandem mass spectrometry (MS/MS) in dried-blood spots has been used to investigate fatty acid oxidation alterations in obesity and type 1 and type 2 diabetes mellitus in both human and animal models [29][30][31][32] . A recent systematic review, however, failed to identify a consistent metabolite profile in GDM 33 . Acylcarnitine profiles. All acylcarnitine values are expressed as medians and the corresponding 95% confidence interval is given in the tables. Compared with AGA newborns, LGA newborns had higher levels of FC, TC, tAC (p < 0.01), tACm, tACl, and, in particular, tACs (p < 0.01). As shown in Table 1, LGA newborns had the highest tAC/FC ratio (0.795) and the lowest FC/TC ratio (0.557). This was close to the cutoff for neonatal carnitine insufficiency (tAC/FC > 0.83) and carnitine deficiency (FC/TC < 0.54), suggesting reduced carnitine storage in this group (SGA, 0.603; AGA, 0.578; and LGA, 0.557) (p < 0.001). Separate analysis of the various acylcarnitines showed remarkably higher levels of C2 and C3 in LGA newborns (p < 0.01), although plasma concentrations for the majority of short-and long-chain acylcarnitines were also higher in the LGA group. The C8/C2 ratio was considerably lower in the LGA group (0.0078 vs. 0.0097 for the AGA group, p < 0.01). The most significant differences between LGA and AGA newborns are summarized in Fig. 1.
Although severe LGA newborns had higher levels of FC, TC, tAC, tACs, tACm, and tACl, the only significant difference compared with AGA  newborns were higher levels of C3 (p < 0.05) ( Table 2). In addition, no increase in carnitine deficiency or insufficiency markers was observed in this group.
The comparison between LGA-GDM and LGA-noGDM newborns revealed that a history of gestational diabetes was associated with higher levels of FC, TC, and short-chain acylcarnitines, including propionylcarnitine (LGA-GDM: 3.33 µmol/L; LGA-noGDM: 2.69 µmol/L), and lower levels of medium-and long-chain acylcarnitines, although the differences were not significant ( Table 3). Analysis of the individual acylcarnitines revealed no remarkable differences between the two subgroups.

Discussion
Disturbances in fetal nutrition, such as intrauterine growth restriction (IUGR) and macrosomia, can impact health during adolescence and adulthood [13][14][15][16]34 and are risk factors for later overweight [17][18][19][20]35,36 . Animal studies have shown that fetal overnutrition results in increased adiposity in newborns, leading to insulin resistance similar to that seen in cases of postnatal overnutrition 37 . Given the high worldwide prevalence of child obesity and overweight 38 , improved knowledge of metabolic homeostasis in higher-risk subgroups, such as LGA newborns, is essential for identifying possible treatment targets.
Recent investigations have identified abnormalities in the metabolic profiles of obese adults and children, such as increased plasma concentrations of branched-chain amino acids (BCAAs) (valine, leucine, and isoleucine), BCAA metabolism byproducts (e.g. alanine, glutamate/glutamine), C3 and C5 acylcarnitines, and aromatic amino acids (phenylalanine and tyrosine) 29,[39][40][41][42][43][44] . The increase in BCAA and short-chain acylcarnitine concentrations is linked to elevated protein intake 45 , and levels are positively correlated with adiposity 46,47 and strongly associated with IR 29,32,44,48 . The metabolic environment in LGA is less well-known. Higher levels of adipokines 49 , BCAAs, and other metabolites, such as alanine, glutamine, threonine, citric acid, glycerol, and glucose, have been detected in cord-blood samples of LGA newborns 50 , and similar findings have been reported for myo-inositol levels in urine samples 51 .
Ours is the first study to characterize acylcarnitine profiles in LGA newborns. In line with the results of obesity studies in children 41,43 and adults 29,31 , our data show that LGA newborns have increased postnatal levels of C3, a product of mitochondrial BCAA catabolism, and in particular of isoleucine and valine catabolism 29 . BCAAs act as signaling molecules of nutritional status 52 . Increased concentrations in obese and insulin-resistant humans may be caused by down-regulation of BCAA oxidation enzymes in adipose tissue 53,54 , as has been observed in animal models with genetic or diet-induced obesity 55 . Propionylcarnitine is a carnitine conjugate of propionyl-CoA, which has been identified as a potential substrate for odd-chain fatty acid synthesis 56 . Moreover, C3 and C5 levels are promising biomarkers for discriminating metabolic wellness in obese individuals 57 , as higher levels have been observed in metabolically unhealthy individuals, independently of body mass index 58 . Consistent with these observations, the LGA newborns in our study had higher C3 (p < 0.001) and C5 levels than AGA newborns, and the concentrations of C3 were even higher than those observed in obese children 41,43 and adults 29 . This could reflect an unhealthy metabolic status in LGA newborns with potential complications in later life. The LGA group also had higher C2 plasma levels, and these were particularly evident in severe LGA newborns. C2 may play a role in insulin resistance through its interaction with acetyl-CoA via carnitine acetyl-CoA transferase. This enzyme seems to act as a positive regulator of total body glucose tolerance and muscle activity of pyruvate dehydrogenase 59 . C2 levels are increased in prediabetic states 60 , diabetes mellitus 30,32,60 , and metabolic syndrome 32 , and are significantly and positively correlated with HbA 1c levels 30 . Interestingly, the levels in the LGA group were higher than those reported for diabetic adult patients 30,32,60 .
One of the most notable differences between LGA and AGA newborns in our study was the significantly higher tACs levels in the LGA group (p < 0.0001). This observation is consistent with reports of increased tACs in obesity, impaired glucose tolerance, and diabetes mellitus 61 .
The acylcarnitine pattern of increased tAC, tACs, C2, and C3 levels was more evident in the SGA group, supporting previous observations in animal models 62 and neonatal studies 63 , and suggesting an impaired fatty acid metabolism in both fetal growth disorders.
In agreement with the profile described for overweight adults 29 and children 64 , LGA newborns also showed higher (though not significantly so) concentrations of medium-and long-chain acylcarnitines than AGA newborns.
A strong correlation was recently demonstrated between carnitine and body composition 65 . Although we observed slightly higher plasma concentrations of free carnitine in LGA newborns, our data also showed a greater tendency towards carnitine deficiency (p < 0.05) and insufficiency (p < 0.001) in this group. Increases in tAC/FC ratio precede a decrease in total plasma carnitine and indicate low tissue bioavailability of FC 66 . The higher FC levels in LGA contrast with the carnitine depletion reported for diet-induced obesity 67 Table 1. Acylcarnitine profiles (µmol/L) according to birth weight based on the 10 th and 90 th percentiles. p 1 , comparison between SGA and AGA newborns; p 2 , comparison between AGA and LGA newborns; 95% CI, 95% confidence interval; FC, free carnitine; TC, total carnitine; tAC, total acylcarnitines; tACs, total short-chain acylcarnitines; tACm, total medium-chain acylcarnitines; tACl, total long-chain acylcarnitines; SGA, small for gestational age; AGA, appropriate for gestational age; LGA, large for gestational age; NS, p not significant.
metabolomic studies of obesity have also shown higher levels of carnitine in obese children 43 . Carnitine insufficiency in our cohort appears to be unrelated to antenatal exposure to GDM. This is relevant given the proposed causative role of carnitine insufficiency in mitochondrial dysfunction and obesity-related impairments in glucose tolerance 68 . It is also consistent with reports that document that GDM in pregnant women does not negatively affect the efficiency of the carnitine system 69 . We did not find any postnatal differences between acylcarnitine profiles in LGA-GDM and LGA-noGDM newborns in our study, although the former had higher concentrations of FC, TC, tAC and tACs. In line with this observation, higher FC and TC levels have been reported in pregnant women with GDM versus healthy pregnant women at 30-33 weeks of gestation 69 . We do not consider that the absence of significant differences between the LGA-GMD and LGA-no GDM groups is due to sample size, as the minimum detectable effect sizes for the samples used in each comparison (with 5% significance and 80% statistical power) were 0.2 for AGA vs LGA and AGA vs SGA (Table 1), 0.25 for SGA vs LGA (Table 1), 0.33 for AGA vs LGA (Table 2), and 0.48 for LGA-noGMD vs LGA-GMD (Table 3). This means that, even in the worst-case scenario (Table 3), we are able to detect true between-group differences of higher than 50% of the SD, which are considered medium effect sizes 70 . Our findings describe a postnatal acylcarnitine profile in LGA newborns that is characteristic of obesity and associated with the development of insulin resistance and prediabetic states, supporting the view that early imbalance in metabolic homeostasis in LGA newborns could contribute to deleterious effects in the long term. Identification of this profile, linked to an unhealthy metabolic phenotype, in the postnatal period could help to establish early dietary intervention and follow-up to reduce the risk of overweight and metabolic syndrome in later life.  Patients. This study was conducted at Hospital Clínico Universitario de Santiago de Compostela, a tertiary hospital in north-west Spain. All newborns born in or referred to our hospital during the first 48 hours of life over the period of 1 year (2015) were included in the study. Informed consent was obtained from parents or legal guardians. Infants with an established diagnosis of an inborn error of metabolism known to alter acylcarnitine profiles (primary systemic carnitine deficiency, mitochondrial fatty acid ß-oxidation defects, ß-ketothiolase deficiency, propionic, methylmalonic or isovaleric acidemia, 3-methylcrotonyl-CoA carboxylase deficiency, 3-hydroxy-3-methylglutaryl-CoA lyase deficiency, or 3-methylglutaconic aciduria) were excluded from the analysis. The following variables were recorded at birth: sex, gestational age, weight (g), length (cm), head circumference (cm), history of GDM, and treatments received (dietary and/or insulin treatment).

Patients and Methods
Newborns were classified into the following groups according to their birth weight: AGA (≥10 th percentile and ≤90 th percentile of birth weight for GA), SGA (<10 th percentile of birth weight for GA), and LGA (>90 th percentile of birth weight for GA). For additional analyses we also classified newborns with a birth weight greater than the 97 th percentile for GA into a severe LGA group.
Birth weight percentiles and z-scores for GA were calculated using the online nutritional assessment tool of the Spanish Society of Gastroenterology, Hepatology and Nutrition (www.gastroinf.es), which is based on Spanish neonatal growth curves 71 .
GDM was defined according to the criteria established in the 2016 Guidelines of the American Diabetes Association 72 using the two-step diagnostic strategy: 1) if plasma glucose is ≥140 mg/dL (7.8 mmol/L) in the 1-hour glucose loading-test, pregnant women must 2) undergo a glucose tolerance test (administration of 100 g of glucose after 8 hours of fasting with subsequent sequential blood sampling). Diagnosis is confirmed when two or more of the following glucose criteria are fulfilled: fasting, ≥105 mg/dL (5.8 mmol/L); 1 hour, ≥190 mg/dL (10.6 mmol/L); 2 hours, ≥165 mg/dL (9.2 mmol/L); and 3 hours, ≥145 mg/dL (8.0 mmol/L).
Based on the corresponding obstetric history of gestational diabetes, LGA newborns were sub-classified as LGA-GDM (co-occurrence of gestational diabetes) or LGA-noGDM (absence of GDM).

Tandem mass spectrometry and study parameters. Analyses of free carnitine and acylcarnitines
were performed on a tandem mass spectrometer coupled to a triple quadrupole analyzer (ESI-MS/MS API 2000; Applied Biosystems Sciex, Toronto, Canada) following an established methodology 73 .
The plates were prepared using the following protocol. The paper blood sample disks and patterns were cut using a BSD 700 automatic drill (BSD tech., Brisbane, Australia) and hand drills followed by microplate placement and addition to each well of methanol. Acylcarnitines were purified with methanol and stable isotope-labeled patterns were used to determine their respective concentrations. The acylcarnitines were then extracted by vortex shaking for 25 minutes. Subsequently, all the methanol was transferred to another plate to distil the blood disks and then evaporated in a gas extractor. The acylcarnitines were derivatized with butanol to their butyl-esters in an acid medium to increase the selectivity of the technique. This was done with the addition of 3N HCl in n-butanol followed by heating at 65 ± 5 °C for 20 minutes and cooling for 5 minutes in a freezer. Excess butanol was evaporated to dryness and once the evaporated plates were at room temperature, a new solution was prepared with 100 μL of the mobile phase of the chromatograph (acetonitrile: water, 1:1). The plates were then covered with foil, vortexed for 5 minutes, and finally analyzed (precursor m/z 120-280 amu).
The reagents were prepared using water purified with a Milli-Q system (Millipore) and the mobile phase was composed of acetonitrile (LiCrosolv Merck, ref. 00030) and formic acid 0.005% (Merck, ref.02264).