The Growth Trends And Association Between Systemic Low-grade Inlfammation And Body Fat Accumulation In Peripubertal Girls

BackgroundKnowledge about the interrelationship between adiposity and systemic low-gradeinflammation during pubertal growth is scarce, despite its importance in detecting the earlysigns of metabolic disorders. This longitudinal study evaluated the change in fat mass (FM)and its distribution, as well as high sensitive C-reactive protein (hsCRP) levels (a marker oflow-grade inflammation), in girls from pre-puberty to early adulthood, assessed whetherthe changes in FM and hsCRP levels had similar trajectories, and investigated whichfactors could explain the relationship between FM and hsCRP levels.MethodsThe study cohort consisted of396girls who participated in a longitudinal study for, onaverage,7.5years (Calex-study: determinants of body composition during growth).Detailed information regarding the participants has been reported previously. Briefly, forthe purpose of this report, we only included girls who had valid data for both bodycomposition and hsCRP concentration, and hence the total number of subjects was257girls at baseline,217girls at the2-year follow-up,119girls at the4-year follow-up and233girls at the7.5-year follow-up assessments, respectively. All information was collectedand laboratory tests were performed within a two-week period during the same monththroughout the7-year follow-up to avoid seasonal effects.Lifestyle and behavioral characteristics as well as medical history were collected via avalidated self-administered questionnaire. Girls under15years of age filled in thequestionnaire with their parents’ assistance, and all the questionnaires were checked by astudy nurse.Dietary information was obtained from three-day food-intake diary (two ordinaryschool days and one weekend day). Energy yielding nutrients including protein,carbohydrate, sucrose and fats (E%) were reported and computed into a dietary intakeindex which indicated the quality of diet.Leisure time physical activity (LTPA) score was calculated on the basis ofquestionnaire as metabolic equivalents (METs).There were13girls who reported use of oral contraceptives at the age of18years. Nodifference in FM between these13girls who were oral contraceptive users and thenon-users was found (p=0.71). Thus these13girls were included in the final analysis. The age at menarche was defined as the first onset of menstrual bleeding as reportedby questionnaire or phone call during the follow-up. Body weight and height weremeasured with subjects wearing light clothes and with bare feet. Weight was determinedwithin0.1kg for each subject using an electronic scale and was calibrated before eachmeasurement session. Height was determined using a fixed wall-scale measuring device tothe nearest0.1cm.Fat mass (FM in kg) of the whole body was assessed by dual-energy X-raydensitometry (DXA Lunar Prodigy Advance; GE Medical Systems Lunar, Madison, WI,USA). Regional fat mass was segmented automatically to define arms (FMarm), legs (FMleg),trunk (FMtrunk) as well as android (FMandroid) and gynoid (FMgynoid) areas using themanufacturer’s software (encore version14.10.022). Two repeated measurements of FMshowed a coefficient of variation (CV) of2.2%in this study.Blood samples were collected in the morning between7:00and9:00a.m. afterovernight fasting at each time point. From menarche onwards, the blood samples werecollected between2and5days after the menstrual bleeding started. Serum was extractedby centrifugation and stored immediately at-80°C until analyzed. The samples fromdifferent time points were analyzed by one technician using the same kits and instrument.Serum hCRP was assessed using enzyme-linked immonosorbent assay (ELISA, DuoSet,R&D Systems, Minneapolis, USA and Diagnostic Systems Laboratories, Inc. Webster,Texas, USA) and leptin using human leptin Kit (ELISA. Diagnostic SystemsLaboratories, Inc. Webster, Texas, USA). Total adiponectin was measured by anenzymeimmunoassay method using the Quantikine Human Total Adiponectin/Acrp30Immunoassay (R&D Systems). Estradiol (E2), testosterone (T) and sex hormone bindingglobulin (SHBG) were determined using ELISA (NovaTec Immundiagnostica GmbH,Dietzenbach, Germany). Inter-and intra-assay coefficients of variation (CVs) were4.6%and6.9%for hsCRP,2.7%and2.2%for leptin,3.3%and4.3%for adiponectin,3.2%and5.4%for E2,3.9%,6.2%for T, and1.1%and1.1%for SHBG, respectively.The study procedures followed were in accordance with the Helsinki Declaration of1975as revised in1983. The study protocol was approved by the ethical committee of theUniversity of Jyv skyl, the Central Hospital of Central Finland and the Finnish NationalAgency of Medicines (memo22/8/2008). The participants provided their written consent inaccordance with the guidelines laid down by the ethical committees. When the girls wereunder the age of18years old, both participants and their parents signed the informedconsent prior to the assessments.Statistical analysisAll data were checked for normality using the Shapiro-Wilk’s W-test in IBM SPSSStatistics20.0for Windows. If data were not normally distributed, their natural logarithmswere used for the correlation analysis. Descriptive statistics were presented as means and95%confidence interval for the mean at the four follow ups. The baseline,2-year and 4-year data were compared to the7-year data using the general linear model for repeatedmeasurements.A hierarchical nonlinear model with random effects was employed to depict the changepatterns of FM and hsCRP concentration from pre-puberty to early adulthood (MLwin2.20software, Multiple Project, Institute of Education, University of London, UK). Timerelative to menarche (TRM) was entered as the explanatory variable in the form ofpolynomial spline functions to explain the change of target variables over time, asdescribed in detail elsewhere. The best model was determined by three criteria: the largestreduction in deviance test (-2*loglikelihood by iterative generalized least squares [IGLS]),the lowest within-individual variance, and the necessary parsimony of the model.To determine if FM and the level of hsCRP could be tracked from baseline to youngadulthood, a group-based semi-parametric method was employed (PROC TRAJ macro inSAS version9.2) to identify distinct developmental trajectories among the cohort and toexamine the possibility of different temporal associations across different subgroupsbetween FM and hsCRP levels. PROC TRAJ allows for individual heterogeneity ingrowth at both individual and group levels and has been shown to be superior foridentifying underlying longitudinal trajectories. Model fitting involved a two-stageselection process of identifying both the optimal number of trajectory groups and theoptimal order of polynomial that best described the trajectory of each group. Finally, wecompared the longitudinal multidimensional relationships between FM and hsCRP toestimate linkage probabilities.To determine which factors were associated with longitudinal changes of FMtotalandhsCRP before and after menarche, we used hierarchical models where the outcome variableis FMtotalor hsCRP, respectively. The independent variables for FMtotalwere hsCRP, E2, T,SHBG, leptin, adponectin, diet index (principle component of energy yield nutrients ofprotein, carbohydrate, sucrose and fats) and LTPA. The independent variables for hsCRPwere FMtotal, E2, T, SHBG, leptin, adponectin, diet index and LTPA. The time of menarcheis selected as a shift knot for the model, which means that the coefficients of independentvariables could be different before and after menarche. Thus, the associations between FMor hsCRP and each predictor can be assessed by regression coefficients before and aftermenarche, respectively, adjusting for other predictor. A t-test was used to assess whetherthe βs were statistically different from0.ResultsBoth FM and hsCRP increased with age and had similar trajectories but differentinter-and intra-variance patterns from pre-puberty to early adulthood. A joint analysis offat distribution and hsCRP indicated that the linkage probabiliteds across differenttrajectory subgroups between regional FM and the corresponding hsCRP levels varied from16%to53%. In a longitudinal regression model, the common predictor for both FM andhsCRP was testosterone (β=0.065, p<0.01and β=-0.213, p<0.05, respectively) before menarche. Other factors predicting FM before menarche were SHBG (β=-0.196, p<0.01)and leptin (β=0.381, p<0.01); and after menarche hsCRP (β=0.048, p<0.01), testosterone(β=0.089, p<0.01), leptin (β=0.340, p<0.01), and adiponectin (β=-0.086, p<0.05). Of thefactors assessed, only FM was associated with hsCRP both before and after menarche(β=1.058, p<0.01and β=1.121, p<0.01, respectively).ConclusionsDifferences in regional body fat deposits and hsCRP levels in adulthood are largelyestablished early in childhood. High FM contributes to high hsCRP during pubertal growth,but the reverse was not true before menarche suggesting that FM may be one of theoriginal contributors of low-grade inflammation and not vice versa. In addition,testosterone (before menarche) and SHBG (after menarche) were the common factorspredict both FM and hsCRP, while dietary intakes and physical activity played no majorrole in the present study.