Literature Search Strategy
We conducted a systematic literature review, with eligibility criteria and search strategy created a priori based on The Cochrane Handbook of Systematic Reviews.6 The databases searched included the Cochrane Library, MEDLINE, PubMed, and the Web of Science, and the time period was from the time of database inception to the present. The search question was the following “How are allostatic load biomarkers utilized to predict objective growth patterns in children?” Each individual biomarker associated with AL was also queried. For example: “How are C-reactive protein measurements utilized to predict objective, physical health outcomes across pediatric populations?”
The search was conducted by a reviewer who assessed the abstract for inclusion criteria, and then examined each full-text report for quality assessment and data extraction. Inclusion criteria included: English language; all ages, races, ethnicities, and genders; studies that used biomarkers linked with objective health outcome data; randomized controlled trials (RCTs), meta-analysis, longitudinal studies, cross-sectional studies, prospective and retrospective review studies; and those studies published within the past decade (2000–2010). Failure to meet one of these criteria resulted in study exclusion. When in doubt, the complete article was screened using the same criteria. Information extracted during the literature review for each individual AL parameter included AL parameter measured; age of pediatric population studied; and method of measurement or collection of the AL parameter.
We reviewed 1,362 abstracts: 569 from MEDLINE and 741 from Web of Science. Of those abstracts, 137 met inclusion criteria and were reviewed. From the articles reviewed, an additional 19 articles were added from article references. We examined each of the major AL coordinated systems, specifying the disease stage captured (primary, secondary, or tertiary), indicating the percent of studies that included children, and the association of each categorical biomarker with growth (Table 2, see page 254).
Table 2. Frequency of Pediatric Studies Measuring Allostatic Load Markers
Sympathetic Nervous System
In the current medical literature, there are few pediatric studies that examine the direct correlation between urinary or plasma catecholamines and growth pattern measurements, and within those few studies there are conflicting data. One study found no association between BMI z-score and urinary catecholamine excretion, but examined a population of children with obstructive sleep apnea.7 The other study did find a relationship between the growth measurements fat mass and BMI with 24-hour excretion rates of catecholamines.8 Both studies looked at a population ranging from 5 to 19 years of age. The studies that were excluded from this review examined the association of urinary and plasma catecholamine levels with other outcomes or mediators, including hypertension, exercise, caffeine administration, and sleep apnea, and not with growth pattern measurements. No recent studies have investigated plasma catecholamine levels or looked at the preschool (below age 5 years) population.
In the adult population, high cortisol levels are associated with obesity, especially abdominal obesity, but in the pediatric population the data are conflicting. Current literature exploring the correlation between cortisol and growth measurement in children looks at salivary cortisol, serum cortisol, urinary cortisol, and dexamethosone suppression testing. Our review found that even within each methodology of cortisol collection there is variability in conclusions.9,10 Barat and colleagues11 examined both plasma cortisol and salivary cortisol within the same population of prepubertal (6 to 12 years) children and found that truncal distribution of fat mass, measured by a dual-energy X-ray absorptiometry (DEXA) scan, correlated positively with morning plasma cortisol but correlated negatively with rise of salivary cortisol. The range and variability of conclusions within the pediatric population shows the need for further research as well as the need for consistent collection methods across studies.
Dehydroepiandrosterone sulfate (DHEAS) is an endogenous natural steroid hormone and one of the most abundant circulating androgens in men and women. The majority of current medical literature in pediatric populations shows that peripubertal obesity is associated with hyperandrogenemia, with elevated DHEAS. All the studies in this review showed a direct correlation between BMI or weight and DHEAS level in older children. However, one study that used waist-to-hip ratio (WHR) as a marker of abdominal obesity showed a negative correlation between WHR and DHEAS level.12
Metabolism and Lipid Profiles
The fasting lipid panel, or at least one or more lipid components, was collected in almost 25% of all pediatric studies that correlated to growth predictor measurements. The most commonly analyzed lipid was triglycerides, with 26% of studies examining this parameter. The majority of triglyceride studies revealed high plasma triglycerides associated with all measures of adiposity, including BMI, sum of four skinfold thicknesses, waist circumference, and waist-to-height ratio.
In Freedman and colleagues’13 cohort study, the only adult lipid marker with a positive correlation to childhood BMI (subjects age 2 to 17 years) was triglycerides measured 17 years after the first collection point, which shows its potential importance in predicting growth. Another cohort study reported an inverse correlation between childhood BMI at age 9 years and triglycerides and total cholesterol at age 50 years.14 However, some studies showed no significant association between BMI or fat mass and triglyceride levels.15 Despite these few conflicting studies, triglyceride levels have a strong association with growth prediction in 38 studies reviewed.
Total cholesterol (TC), low-density lipoprotein (LDL), and high-density lipoprotein (HDL) were reported in 15% to 20% of the studies, with relatively consistent findings. Elevated TC, elevated LDL, and decreased HDL were associated with increased BMI, waist circumference, skinfold thickness, and fat mass. Very few studies showed conflicting data. In another compelling study, BMI increased more in hypercholesterolemic girls (high LDL) compared to non-hypercholesterolemic girls 5 to 6 years of age over a 6-year period.16
The homeostatic model assessment of insulin resistance (HOMA-IR) is a method to quantify insulin resistance and beta cell function; it is calculated using a fasting insulin and glucose serum level. This parameter was calculated in 13% of the pediatric studies and was one of the more common parameters associated with growth pattern predictors. All of the studies that measured HOMA-IR found a positive correlation with increased BMI, whole body fat mass, truncal distribution of fat mass (calculated by [subscapular + waist] / [hip+ thigh] fat mass), waist circumference, weight, subscapular skinfold thickness, and visceral adiposity.15
Of the glucose and insulin markers, HbA1c was the least studied in the pediatric studies, with only 2% of the studies collecting this data. Although one study showed no statistically significant correlation with HbA1c and body composition reported by DEXA scan, other studies found a significant correlation with higher BMI (> 99th percentile) associated with elevated HbA1c levels.17 None of the studies that collected HbA1c did so in the preschool population but instead focused on children ages 6 to 17 years.
Fasting insulin was reported in approximately 19% of all pediatric studies. With few exceptions, most studies reported a positive correlation with increasing fasting plasma insulin as a predictor of overweight or obese growth patterns.18 These growth prediction measurements included truncal distribution of fat, waist circumference, intra-abdominal fat area, skinfold thickness, body fat percentage, waist-to-height ratio, and ponderal index, with the most commonly used measurement being BMI z-score or BMI. Butte and colleagues19 found that fasting insulin, along with leptin, ghrelin, and total T3, were independent predictors of child weight gain over 1 year in children ages 4 to 19 years, whereas Freedman and colleagues13 showed that childhood BMI (children ages 2 to 17 years) did not correlate with adult insulin levels collected 17 years later. Preschoolers (2 to 6 years) demonstrated an association with elevated insulin levels and BMI, and ponderal index was associated with elevated insulin levels.16
In these studies, gender differences in growth patterns reflected in the different biomarker associations were noted. For example, for girls, BMI was significantly related to systolic and diastolic blood pressure, HDL, and triglyceride concentrations. All of these increased with age. For boys, BMI was associated with insulin concentration and only systolic blood pressure.
C-reactive protein (CRP) was the most commonly measured inflammatory marker in approximately 20% of pediatric studies in this literature review. Consistently, studies showed a positive correlation between elevated CRP and overweight/obesity. Most of these studies were looking at populations greater than age 5 years, but one study collected CRP in 2- and 3-year-old Hispanic children and was the one study that found no association between BMI and CRP levels.20
Other inflammatory markers, including serum albumin, fibrinogen, interleukin-6 (IL-6), and tumor necrosis factor-alpha, were evaluated less frequently but are important to note. Of these markers, serum albumin has the least evidence supported by the literature of a correlation with growth pattern measurements.21 IL-6 levels were studied more often and showed variability of correlations. The majority of the IL-6 studies showed a positive correlation between IL-6 levels and BMI and fat mass. Within those, there is variability in results by gender (for example, IL-6 is elevated in obese women but not obese men).22 Fibrinogen, although less frequently studied, consistently had a positive correlation with higher percent fat mass, upper body subcutaneous fatness, and BMI.23 Tumor necrosis factor-alpha studies, on the other hand, showed variable results.22
Heart rate variability (HRV) is determined by 24-hour electrocardiogram/ambulatory blood pressure monitoring and measures overall sympathovagal balance. Decreased HRV, which represents autonomic dysfunction, was seen consistently in obese children defined by BMI.24 There were no studies in children below the age of 11 years, with most studies evaluating children ages 11 to 13 years. Homocysteine correlated significantly with BMI, fat mass, and percent fat mass in some studies.25
Diastolic blood pressure and systolic blood pressure were each measured in approximately 25% of all pediatric studies in this review, making these markers two of the most commonly studied AL markers. In the vast majority of studies, both systolic and diastolic blood pressure significantly correlated with BMI.26 Despite the studies that showed positive correlation, the Bogalusa Heart Study, which initially followed participants ages 2 to 17 years and then again after 17 years, found that childhood BMI did not correlate with adult systolic and diastolic blood pressure.13
Twenty percent of studies evaluated leptin levels and 21% evaluated adiponectin. The majority of the leptin studies showed a positive association between leptin and higher BMI, weight, waist circumference, WHR, fat mass, and fat mass percentage.27 Leptin was found to be an independent predictor of weight gain after a 1-year follow up in Hispanic children ages 4 to 14 years.19 Fleish and colleagues28 also found leptin to be a positive predictor of increased BMI and total body fat mass in 6- to 12-year old children followed up on average 4.4 years later. Only two of 32 studies found that leptin poorly predicted relative body weight in a 5- and 6-year follow-up.29 Although adiponectin has been included in most recent AL scores for adults, they have not yet been used consistently in children. The majority of adiponectin studies demonstrated a consistent negative association between adiponectin and growth measures, including BMI, waist circumference, percent body fat, and visceral adiposity, including in young children.30