1424.full

Serum Concentrations of Insulin-like Growth Factor (IGF)-1and IGF Binding Protein-3 (IGFBP-3), IGF-1/IGFBP-3

Ratio, and Markers of Bone Turnover:Reference Values for French Children and Adolescents and

z-Score Comparability with Other ReferencesCorinne Alberti,1,2,3 Didier Chevenne,4 Isabelle Mercat,5 Emilie Josserand,2 Priscilla Armoogum-Boizeau,2

Jean Tichet,5 and Juliane Leger3,6,7,8*

BACKGROUND: A reference model for converting serumgrowth factor and bone metabolism markers into anSD score (SDS) is required for clinical practice. Weaimed to establish reference values of serum insulin-like growth factor-1 (IGF-1) and IGF binding protein 3(IGFBP-3) concentrations and bone metabolismmarkers in French children, to generate a model forconverting values into SDS for age, sex, and pubertalstage.

METHODS: We carried out a cross-sectional study of1119 healthy white children ages 6 –20 years. We as-sessed concentrations of serum IGF-1, IGFBP-3, car-boxyterminal telopeptide �1 chain of type I collagen(CrossLaps), and bone alkaline phosphatase concen-trations and height, weight, and pubertal stage, andused semiparametric regression to develop a model.

RESULTS: A single regression model to calculate theSDSs with an online calculator was provided. A positiverelationship was found between SDS for serum IGF-1and IGFBP-3, IGF/IGFBP-3 mol/L ratio, and anthro-pometric parameters (P � 0.0001), with slightly greatereffects observed for height than for body mass index(BMI). There was a negative relationship between se-rum CrossLaps concentration and BMI, and a positiverelationship between serum CrossLaps concentrationand height. A comparison of serum IGF-1 referencedatabases for children showed marked variation as afunction of age and pubertal group; smooth changeswith age and puberty were observed only in our model.

CONCLUSIONS: This new model for the assessment ofSDS reference values specific for age, sex, and pubertalstage may help to increase the diagnostic power of theseparameters for the assessment of growth and bone me-tabolism disorders. This study also provides informa-tion about the physiological role of height and BMI forthe interpretation of these parameters.© 2011 American Association for Clinical Chemistry

Serum concentrations of insulin-like growth factor-I(IGF-1)9 and IGF binding protein-3 (IGFBP-3) arecommonly used to assess the growth hormone (GH)-IGF-1 axis in patients with growth disorders (1 ). It hasbeen suggested that the molar ratio of IGF-1 toIGFBP-3 could be used as a crude indicator of IGF-1bioavailability (2 ), and several recent epidemiologicalstudies in adults have shown an association betweenhighly bioavailable IGF-1 concentrations and cancerrisk, raising concerns about the long-term safety ofhigh-dose GH treatment (3 ). The IGF-1/IGFBP-3 ratiois also associated with metabolic syndrome (4 ). Thesebiological parameters have become essential tools formonitoring GH replacement therapy in various condi-tions (3 ). Good clinical practice dictates that patientsshould be monitored regularly during GH replacementand that serum IGF-1 concentrations should be main-tained below the upper limit of the reference intervalfor the age and sex of the child (3 ). Serum concentra-tions of bone alkaline phosphatase (BALP) and the car-

1 INSERM, CIC-EC CIE5, Paris, France; 2 Assistance Publique-Hopitaux de Paris,Hopital Robert Debre, Unite d’Epidemiologie Clinique, Paris, France; 3 UniversiteParis Diderot, Paris, France; 4 Assistance Publique-Hopitaux de Paris, HopitalRobert Debre, Service de Biochimie-Hormonologie, Paris, France; 5 RegionalInstitute for Health, Tours, France; 6 Assistance Publique-Hopitaux de Paris,Hopital Robert Debre, Service d’Endocrinologie Pediatrique, Paris, France; 7 As-sistance Publique-Hopitaux de Paris, Centre de Reference des Maladies Endo-criniennes Rares de la Croissance, Paris, France; 8 INSERM, UMR 676, Paris,France.

* Address correspondence to this author at: Pediatric Endocrinology Department,

Centre de Reference Maladies Endocriniennes de la Croissance, INSERM U676,Hopital Robert Debre, 48 Boulevard Serurier, 75019 Paris, France. Fax �33-1-40-03-24-29; e-mail [email protected].

Received May 27, 2011; accepted July 29, 2011.Previously published online at DOI: 10.1373/clinchem.2011.1694669 Nonstandard abbreviations: IGF-1, insulin-like growth factor I; IGFBP-3, insulin-

like growth factor binding protein 3; GH, growth hormone; BALP, bone alkalinephosphatase; CrossLaps, carboxyterminal telopeptide �1 chain of type I colla-gen; BMI, body mass index.

Clinical Chemistry 57:101424–1435 (2011)

Endocrinology and Metabolism

1424

boxyterminal telopeptide �1 chain of type I collagen(CrossLaps) are recognized markers of bone formationand resorption, respectively. These markers are used toassess bone-remodeling status and the response totreatment in patients with bone metabolism disorders(5, 6 ).

Known determinants of these biological parame-ters include age, sex, and puberty, together with geneticand ethnic factors (7–19 ). There is some concern thatauxological parameters may modify serum concentra-tions. Illness and nutritional status are also known toaffect concentrations of all these serum growth factorsand bone metabolism markers. Thus, determination ofpediatric reference intervals for healthy children is es-sential for the assessment of growth and bone metabo-lism disorders and for the monitoring of response totreatment (20 ).

The establishment of pediatric reference intervalsfor serum growth factors and bone metabolism mark-ers and the assessment of the relationships betweenthese parameters and sex, age, and puberty require alarge population of healthy children and adolescents,preferably specific to a particular ethnic group. More-over, the concentrations of these markers dependpartly on nutritional state and display circadian varia-tion; their determination from morning samples col-lected from fasting patients greatly decreases individualvariability and should therefore be preferred (21 ). Thedistribution of these biological parameters is oftenskewed, so appropriate curve-fitting procedures arealso essential. Very few normative studies have fulfilledthese conditions, and many have suffered from the in-clusion of only small numbers of study participants.There is also considerable variation in the assay meth-ods used (20, 22 ).

On the basis of results of our previous study, inwhich we investigated the relationship between abso-lute serum concentrations for components of the GH/IGF-1 axis and serum markers of bone turnover andmetabolism in healthy children (23 ), we aimed to es-tablish references values for sex, age, and pubertal stagefor these parameters in healthy French children and topresent a model for converting serum concentrationsof these factors into standard deviation scores (SDS)that can be adapted for easy practical use in other eth-nic populations. We investigated the relationship be-tween each parameter and body mass index (BMI) andheight expressed as SDS. We then compared our modelfor serum IGF-1 concentrations with previously re-ported models.

Methods

The study population (n � 1119) has been describedelsewhere (23 ). Briefly, morning serum samples were

obtained from healthy fasting white children and ado-lescents (n � 579 boys and n � 540 girls) 6 to 20 yearsold. This cross-sectional study also included pubertal-stage assessment by clinical examination, according toTanner stage, height and weight measurements, andBMI calculations. These anthropometric data are re-ported in Supplemental Table 1 and Supplemental Fig.1 in the Data Supplement that accompanies the onlineversion of this article at http://www.clinchem.org/content/vol57/issue10.

The study protocol was reviewed and approved bythe faculty ethics committee. It was explained to allstudy participants and their parents, who signed a writ-ten consent form for participation.

SERUM ANALYSIS

Serum IGF-1 concentrations were originally deter-mined by fully automated 2-site chemiluminescenceimmunoassays (Nichols Advantage®, Nichols InstituteDiagnostics), with interassay CVs of 5.3% at a serumconcentration of 45 �g/L, 4.8% at 413 �g/L, and 7.4%at 810 �g/L, and a detection limit of 6 �g/L. Because theNichols Advantage system has since been discontinuedand is no longer available, we used 93 blood samplesfrom children of both sexes and different ages to carryout a comparison of this method with the IRMA IGF-1-RIACT assay (Cisbio Bioassays). The correlation be-tween the 2 methods was r � 0.986, which yielded sim-ilar results for IGF-1 concentration, as shown by theweighted Deming regression equation: IGF-1 Cisbio �(1.016 � IGF-1 Nichols Advantage) � 9.5 (see onlineSupplemental Fig. 2). The serum IGF-1 concentrationsobtained with the Nichols Advantage assay were there-fore converted into Cisbio IGF-1 values with this equa-tion, making it possible to establish reference curvesthat could be used with the Cisbio assay. SerumIGFBP-3 concentrations were measured with an IRMAkit (IGFBP-3 Active® IRMA, DSL, BeckmanCoulter),with interassay CVs of 6.5% at a serum concentrationof 400 �g/L, 5.8% at 4200 �g/L and 3.9% at 7200 �g/L,and a detection limit of 50 �g/L. We calculated theIGF-1/IGFBP-3 mol/L ratio on the basis of the molec-ular weights of IGF-1 and IGFBP-3. Serum BALP con-centrations were determined with an IRMA kit (Os-tase, BeckmanCoulter), with interassay CVs of 9.2% ata serum concentration of 17 �g/L, 8.5% at 40 �g/L, and7.5% at 82 �g/L, and a detection limit of 2 �g/L. Theserum CrossLaps® assay is an ELISA (serum CrossLapsELISA, IDS) specific for a �-aspartate form of theEKAHD-�-GGR epitope derived from the cross-linkeddegradation products of C-terminal telopeptides oftype I collagen. For this assay, the detection limit is 20ng/L and the interassay CVs were 9.7% at a serum con-centration of 444 ng/L, 3.9% at 1066 ng/L, and 2.5% at1967 ng/L.

GH/IGF-1 Axis and Bone Turnover Metabolism in Children

Clinical Chemistry 57:10 (2011) 1425

http://www.clinchem.org/content/vol57/issue10

http://www.clinchem.org/content/vol57/issue10

STATISTICAL ANALYSIS

Quantitative variables are reported as medians (rangeor percentiles) or mean (SD).

We established age-specific reference intervals forall biological parameters by using the parametricmethod described by Royston and Wright, in line withthe recommendations of the WHO (24, 25 ).

This method can be used (a) to estimate outer per-centiles precisely, (b) to estimate percentiles simultane-ously such that they are constrained not to cross, (c) toestimate z scores and percentiles from direct formulas,(d) to account for both skewness and kurtosis whennecessary, and (e) to apply continuous age smoothing.This last aspect is particularly important when choos-ing the statistical method to be used. Indeed, referenceintervals are sometimes calculated as the mean (2 SD)for each age group, these groups being arbitrarily de-fined, often by using the middle of the class. This pro-cedure results in the absence of a smooth change in themean with age, as would be expected on biologicalgrounds, and the same is true for the SD curve (24, 25 ).A summary of the method is given below. For moredetails see the online Supplemental Methods file.

Assuming a gaussian distribution of the variable ateach age, we calculated a percentile according to awidely used formula: percentile � mean � (U � SD),where the mean and SD are defined for a given age andU is the corresponding percentile constant of the stan-dard gaussian distribution (for example, for the 2.5thand 97.5th percentiles curves, U � �1.96). The aimwas to identify functions that adequately representedthe changes in mean and SD with age. Initial mathe-matical transformation was applied, if required, to re-duce the positive skewness and heteroscedasticity ofthe measurements of interest. Suitable functions forthe mean and SD based on age were obtained fromregression splines. The fitted values based on the re-gression gave the estimated mean curve and the SD. Weassessed the fit of the model by calculating the SDSs (zscore) as: z � (measurement – mean)/SD. The orderedz scores were plotted to provide a graphical check ofnormality. If the hypothesis of normality was accepted,then no further modeling was required. If not, then anexponential transformation was applied to z, and theskewness curve was fitted by the maximum likelihoodmethod, using fractional polynomials as powers of themodel of regression against age.

We calculated the percentile estimates and refer-ence intervals by introducing the fitted curves of themean, SD, and skewness into the percentile equation,which becomes a little more complex when addingskewness (see the online Supplemental Methods file formore details). For variables that have previously beentransformed, we obtained percentile curves on theoriginal scale by applying a back transformation to the

calculated curves. At the end of the process, we checkedseveral internal validity measures (26 ). Sex-specificcurves were generated and the effect of puberty wasexplored by adding the variable to the reference inter-val model.

We used linear regression analysis to investigatethe relationships between individual biological param-eters (expressed in SDS) and BMI and height (also ex-pressed in SDS), as a function of age and sex. We ex-plored the differences between our model andreference curves for IGF-1 as a function of age andpuberty stage generated from published data, by plot-ting the sex-specific median curves of 6 databases(7, 8, 10, 11, 14, 15, 19 ) together with our data.

All statistical analyses were carried out with SAS9.1 (SAS Institute), STATA (version 9) and MethodValidator (for Deming regression) software.

Results

Reference interval curves (medians and 2.5th–97.5thpercentiles) for IGF-1, IGFBP-3, IGF-1/IGFBP-3 ratio,CrossLaps, and BALP for age and sex are displayed inFig. 1 and reference values per midclass age integer aregiven in Table 1. All mean and scale curves were mod-eled with natural cubic splines. An online calculatoris available (http://urc-paris-nord.aphp.fr/login/root/IRSA_web/index.html).

As expected, despite considerable interindividualvariability, marked variation with age and sex was ob-served for all parameters. Median serum concentra-tions of IGF-1 and IGFBP-3, and the IGF-1/IGFBP-3ratio increased between the ages of 6 and 15–16 years inboys and 6 and 13–14 years in girls. Consistent with thetiming of the pubertal period, peak concentrations oc-curred at the ages of 13–14 years in girls and at 15–16years in boys, with slightly higher values in girls than inboys. Beyond the ages of 16 years in boys and 14 yearsin girls, serum concentrations of IGF-1 decreased,whereas those of IGFBP-3 reached a plateau. In bothsexes, median serum concentrations of CrossLaps andBALP remained fairly constant during the prepubertalperiod; increased during puberty, with slightly highervalues obtained for boys than for girls; and decreasedrapidly thereafter. At midpuberty (approximately age12–13 years in girls and 14 –15 years in boys), all refer-ence curves other than that for IGFBP-3 displayed alarger spread of values than before or after puberty. Fig.2 shows the effect of puberty stage, adjusted for age andsex. For serum IGF-1 concentration and IGF-1/IGFBP-3 ratio in girls, distinct curves were obtained forstages I, II, and III, whereas the curves for stages IV andV were confounded. Puberty stages could be assignedto 2 groups (I-II vs III-IV-V) for serum IGF-1 andIGFBP-3 concentrations and IGF-1/IGFBP-3 ratios in

1426 Clinical Chemistry 57:10 (2011)

http://urc-paris-nord.aphp.fr/login/root/IRSA web/index.html

http://urc-paris-nord.aphp.fr/login/root/IRSA web/index.html

Fig. 1. Reference curves for IGF-1 concentration, IGFBP-3 concentration, IGF-1/IGFBP-3 mol/L ratio, and BALP andCTX concentrations in boys (left column) and girls (right column), as a function of chronological age.

Curves represent the 2.5th percentile (- - -), 50th percentile (. . .), and 97.5th percentile (— — —).



Table 1. Age- and sex-specific reference values for IGF-1 concentration, IGFBP-3 concentration, IGF-1/IGFBP-3molar ratio, and BALP and CrossLaps concentrations.a

Age, years

6.5 7.5 8.5 9.5 10.5 11.5 12.5 13.5 14.5 15.5 16.5 17.5 18.5 19.5

IGF-1 males, �g/L

2.5th Percentile 81 83 85 87 90 98 126 189 261 296 296 284 270 255

Median 160 168 178 194 219 257 317 399 465 476 454 431 415 403

97.5th Percentile 280 298 325 369 438 540 649 732 759 721 662 622 605 602

IGF-1 females, �g/L

2.5th Percentile 105 105 107 115 135 173 224 272 299 305 293 269 235 196

Median 159 180 206 244 298 376 464 526 531 500 463 434 413 395

97.5th Percentile 341 360 404 465 547 654 760 807 765 680 609 571 558 559

IGFBP-3 males, �g/L

2.5th Percentile 2277 2366 2475 2613 2792 3026 3318 3625 3890 4055 4099 4044 3918 3746

Median 3816 3996 4178 4363 4551 4745 4943 5127 5273 5361 5381 5349 5277 5181

97.5th Percentile 5354 5624 5880 6112 6310 6465 6568 6628 6657 6666 6664 6654 6637 6616

IGFBP-3 females,�g/L

2.5th Percentile 2636 2744 2866 3012 3192 3412 3638 3812 3896 3906 3887 3856 3819 3777

Median 4152 4318 4487 4660 4838 5022 5188 5299 5333 5316 5281 5247 5217 5187

97.5th Percentile 5669 5892 6108 6307 6485 6632 6738 6787 6774 6725 6675 6639 6614 6597

IGF-1/IGFBP-3 males

2.5th Percentile 0.08 0.08 0.08 0.08 0.08 0.10 0.12 0.15 0.18 0.22 0.24 0.24 0.23 0.21

Median 0.15 0.14 0.14 0.15 0.17 0.2 0.24 0.28 0.31 0.31 0.31 0.30 0.28 0.26

97.5th Percentile 0.24 0.24 0.24 0.26 0.30 0.36 0.44 0.49 0.48 0.44 0.39 0.36 0.35 0.33

IGF-1/IGFBP-3females

2.5th Percentile 0.10 0.10 0.1 0.1 0.11 0.14 0.18 0.21 0.22 0.21 0.2 0.18 0.16 0.14

Median 0.14 0.15 0.16 0.18 0.22 0.27 0.33 0.37 0.37 0.34 0.32 0.3 0.29 0.28

97.5th Percentile 0.25 0.26 0.28 0.32 0.37 0.45 0.53 0.57 0.54 0.48 0.43 0.4 0.4 0.41

BALP males, �g/L

2.5th Percentile 38 37.1 36.3 36.1 36.5 37.8 38.5 36.2 30.3 23.3 17.3 12.7 9.2 6.6

Median 73.0 68.9 66.3 66.2 69.5 77.2 84.7 84.1 71.9 55.4 41.0 30.2 22.0 15.8

97.5th Percentile 116.2 109.3 105.8 108 118.9 141.7 168.3 178.3 159.4 126.9 96.8 73.7 55.9 41.9

BALP females, �g/L

2.5th Percentile 32.4 36.3 39.2 40.2 38.4 33.4 26.7 20.2 14.8 11 8.9 8.1 8 8.4

Median 68.4 73.3 77.3 79.5 78.8 74.4 66.0 53.8 40.0 28.6 21.1 16.8 14.2 12.5

97.5th Percentile 110.1 117.8 125.4 132.3 138.1 142.1 139.8 124.9 98 70.7 50.3 36.9 27.6 20.8

CrossLaps males,ng/L

2.5th Percentile 1230 1285 1293 1289 1291 1323 1442 1629 1640 1335 983 778 697 679

Median 2227 2151 2102 2098 2157 2306 1799 2940 2997 2540 1958 1568 1355 1234

97.5th Percentile 2952 2902 2891 2959 3153 3517 4072 4697 4871 4272 3417 2772 2349 2050

CrossLaps females,ng/L

2.5th Percentile 1339 1413 1484 1525 1512 1423 1256 1026 778 581 469 429 436 472

Median 1981 2078 2173 2263 2342 2400 2368 2150 1751 1348 1065 908 826 785

97.5th Percentile 3692 3410 3346 3401 3564 3836 4046 3894 3280 2544 1965 1591 1342 1163

a Reference values were calculated from reference equations, according to sex and the age in the middle of the age class. For example, for the age class of 6 to7 years, we chose an age of 6.5 years to estimate the expected median IGF-1 concentration for a boy. This equation gives a value of 160 �g/L.


Fig. 2. Reference curves for IGF-1 concentration, IGFBP-3 concentration, IGF-1/IGFBP-3 mol/L ratio, and BALP and CTXconcentrations in boys (left column) and girls (right column), as a function of Tanner stage and chronological age.

Curves represent stage I (– – –), stage II (---), stage III ( — • —), stage IV (— — —), stage V ( - • - • - ).



boys and IGFBP-3 concentrations in girls, and were lesscontrasted for serum CrossLaps and BALP concentra-tions in boys. Thus, for certain markers, puberty pro-vides additional information that can be taken into ac-count when estimating SDS.

Table 2 shows the relationship of all biological pa-rameters to height and BMI, all expressed in SDS forage and sex. There were significant positive relation-ships between serum IGF-1 and IGFBP-3 concentra-tions, IGF-1/IGFBP-3 mol/L ratio, and anthropomet-ric parameters. The effect of height was slightly greaterthan that of BMI, as shown by the magnitude of thecoefficients. For bone markers, there was a negativerelationship between serum CrossLaps concentrationand BMI, and a positive relationship between serumCrossLaps concentration and height. Fig. 3 illustratesthe predicted 2.5th, 50th, and 97.5th percentile curves,based on the equations given in Table 2, for serumIGF-1 SDS values as a function of height SDS for astudy participant with a BMI fixed at �2 SDS (Fig. 3A)and as a function of BMI SDS for a study participantwith a height fixed at �2 SDS (Fig. 3B). These curvesdemonstrate the positive association of serum IGF-1SDS values with BMI SDS and height SDS.

We plotted graphical comparisons of referencecurves of serum IGF-1 concentrations as a function ofage, sex, and puberty from different reference popula-tions and our models (Figs. 4 and 5). Despite knowndifferences in absolute circulating total IGF-1 concen-trations between different commercially availableIGF-1 immunoassays, in analyses of data with respectto age continuity, the median values obtained duringthe pubertal peak were generally less pronounced forboth boys and girls in studies other than this study (Fig.4). Furthermore, differences between databases variedconsiderably between pubertal groups, with smoothchanges with age and puberty observed only in ourmodel (Fig. 5).

Table 2. Relationship of biological markers to BMI and height (all expressed in SDS as a function of age andsex), as studied by multivariate linear regression.a

Intercept BMI Height

IGF-1 �0.197 (�0.266 to �0.126)2 0.116 (0.074 to 0.159) 0.182 (0.131 to 0.233)

IGFBP-3 �0.116 (�0.188 to �0.045) 0.108 (0.064 to 0.151) 0.094 (0.043 to 0.146)

IGF-1/IGFBP-3 �0.157 (�0.228 to �0.085) 0.068 (0.025 to 0.112) 0.159 (0.107 to 0.210)

CrossLaps 0 �0.149 (�0.192 to �0.106) 0.117 (0.073 to 0.160)

BALP �0.027 (�0.100 to 0.046) �0.026 (�0.070 to 0.018) 0.037 (�0.016 to 0.090)

a Results are expressed as regression coefficients with 95% CIs, and are illustrated in Fig. 3.b All P values are less than 0.001, except for BALP concentration, for which P � 0.47 for the intercept; P � 0.25 for BMI; and P � 0.17 for height.

Fig. 3. Relationship of IGF-1 SDS to BMI (A) andheight (B), both expressed in SDS and based on theequations given in Table 2.

A linear model shows predicted curves for the relationshipbetween serum IGF-1 SDS and height SDS for a child witha BMI arbitrarily fixed at �2 SDS and for the relationshipbetween serum IGF-1 SDS and BMI SDS for a child with aheight arbitrarily fixed at �2 SDS. Curves represent the2.5th percentile (– – –), 50th percentile ( ) and 97.5thpercentile ( ). For example, 95% of serum IGF-1 concen-tration lies between �0.27 and �0.58 SDS, with the 50thpercentile at �0.43 SDS for a child with a BMI of �2 SDSand height of 0 SDS, and 95% of serum IGF-1 concentra-tion lies between �0.39 and �0.73 SDS, with the 50thpercentile at �0.56 SDS for a child with a height of �2SDS and a BMI of 0 SDS.


Discussion

In this cross-sectional study, we provide reference val-ues specific for age, sex, and pubertal stage for serumIGF-1 and IGFBP-3 concentrations, the IGF-1/IGFBP-3 ratio, and CrossLaps and BALP concentra-tions in healthy French white children between ages6 –20 years. We constructed a model formula to calcu-

late the SD scores for all of these biological parametersas a function of age, sex, and pubertal stage. This singleregression model makes possible the simultaneous es-timation of all these known explanatory variables andtakes into account the interactions between variables,rather than dividing the sample into subgroups corre-sponding to each pubertal stage and fitting separateregression models for each subgroup. Our model also

Fig. 4. Comparison of our reference curve models for serum IGF-1 concentration (fitted mean, solid line) with 3pediatric models (fitted mean, dotted line) as a function of age and sex.

Data were analyzed with age continuity preserved. (A), Nichols Advantage. (B), Immulite; Siemens. (C), Immulite; Siemens.References in which authors reported the serum IGF-1 measurements obtained with these assays are also shown.



Fig. 5. Comparison of our reference curve models for serum IGF-1 concentration (fitted mean, solid line) with 4pediatric models (fitted mean, dotted line) as a function of age, sex and puberty.

Data were analyzed pubertal groups preserved. (A), DSL-5600 ACTIVE IGF-1 IRMA; BeckmanCoulter). (B), In-house RIA. (C),Mediagnost. (D), Immulite; Siemens. References in which authors reported the serum IGF-1 measurements obtained with theseassays are also shown.


shows that all these biological variables are related toage at each stage of puberty. A similar model was pre-viously proposed for serum IGF-1 and IGFBP-3 con-centrations and their ratio (7, 15, 16, 19 ). Our findingsconfirm those of previous studies showing an age-related increase in serum IGF-1 and IGFBP-3 concen-trations and their ratio during the prepubertal andearly pubertal stages, followed by a decrease in late pu-berty for IGF-1 and the ratio, but not for IGFBP-3, withslightly higher peak values in girls than in boys. Ourstudy is also the first to develop a model suitable for usein children that relates serum BALP and CrossLapsconcentrations to age, sex, and puberty simultaneouslyand converts these biological parameters of bone turn-over into SDS with a high degree of accuracy. Consis-tent with other studies (17, 27–29 ), serum BALP andCrossLaps concentrations during puberty—markersdescribing only net skeletal formation and resorption,respectively, and unable to distinguish betweengrowth, modeling, and remodeling— were found to behigher in boys than in girls, probably reflecting thehigher peak height velocity and peak bone mass in boysthan in girls, resulting in a greater degree of bone re-modeling, but also of bone modeling, during this crit-ical period of growth and bone mass accretion.

We also investigated the determinants of these bi-ological parameters over the whole spectrum of heightand BMI in this large cohort of carefully selectedhealthy white children. Previous studies on the rela-tionships between growth factors and anthropometricmeasurements have yielded conflicting results, withsome, but not all, studies indicating positive associa-tions between these parameters and the serum concen-trations of IGF-1 and IGFBP-3 and their ratio in chil-dren (7, 11–13, 15, 16, 19, 30 –33 ) or adults (34 –37 ). Ithas been suggested that such associations may be at-tributable to the effects of IGF-1 on growth velocityrather than adult height, with no persistent effect onadult height. Nevertheless, the molar ratio of IGF-1/IGFBP-3, which provides an indication of the amountof bioavailable IGF-1, has been shown to be weaklypositively associated with the leg-to-trunk ratio, an in-dicator of relatively long leg length in relation to totalstature in adults (38 ). A positive association was foundbetween serum IGF-1 and IGFBP-3 concentrations,and also between each of these concentrations andheight growth over the next 2 years, in prepubertal chil-dren of ages from 5 years to 9 –10 years (39 ). We haveshown in this study that there are potentially usefulpositive relationships between serum IGF-1 concentra-tion, IGFBP-3 concentration, and their ratio, andheight and BMI expressed in SDS for age and sex. Ourfindings provide new insight for the interpretation ofthese biological parameters in clinical practice and, forexample, in short children, with respect to normative

data appropriate for sex, age, pubertal age, and BMI,which indicated that height SDS was a significant de-terminant of serum IGF-1 and IGFBP-3 concentra-tions and of their ratio, expressed as SDS values, overthe whole BMI spectrum, including the lowest endof the range. However, although GH secretion andchronic and acute starvation are known to be impor-tant determinants of IGF-1 production, the determi-nants of serum IGF-1 concentrations across the entireweight or height spectrum are poorly understood. Wewere therefore unable to establish definitively that therelationship between IGF-1 and height, for example,was comparable within groups of children with an-orexia nervosa or obesity or children who were veryshort or tall. Further studies in large groups of patientswith growth disorders would be required for this com-parison. Our results for bone metabolism markers areconsistent with those previously reported for serumCrossLaps concentration, a marker of bone resorption,but not with those for serum BALP concentration, forwhich a weak positive correlation was found with theSDS of anthropometric markers, suggesting greaterbone formation in children that are tall or heavy fortheir age (17 ), which was not seen in our study.

The use of different assays with different antibod-ies and ethnic groups, and differences in nutrition andage at onset of puberty between the populations stud-ied may account for the differences in absolute serumIGF-1 concentration observed (the most commonlyused parameters in clinical practice and the most fre-quently studied). We show here that our model, whichpreserves age continuity as children pass from one pu-bertal stage to another, increases the precision of thesmooth physiological changes observed with age andpubertal stage. This model increases the diagnosticpower of SDS values in clinical practice, particularly inchildren with puberty onsets that deviate from themean.

This study included a substantial number of clin-ically well-characterized healthy homogeneous whitechildren and adolescents, for whom determinationswere carried out after an overnight fast, but it nonethe-less had several limitations. First, the withdrawal of theassay initially used for the determination of serumIGF-1 concentrations in this study made it necessary toconvert each serum IGF-1 value obtained to an equiv-alent value for the assay currently in use. The validity ofthis conversion was higher than the interassay CV ofeach assay. Second, circannual variations in biologicalmarkers of bone turnover have been reported (40 ) andmay contribute to the biological variability of boneturnover. These variations may be at least partly attrib-uted to subclinical vitamin D deficiency during thewinter period, which is common in countries of thenorthern hemisphere. In our study, our blood samples



were obtained over an 18-month period and did nottake into account the potential circannual variation.Finally, we observed no longitudinal changes in bio-chemical and auxological markers because the studydesign was cross-sectional.

In conclusion, we present a newly developedmodel for assessing age-, sex-, and pubertal stage–specific SDS reference values for serum IGF-1 andIGFBP-3 concentrations and their ratio and BALP andCrossLaps concentrations. Informative, widely ac-cepted biological norms for the assessment of growthand bone metabolism disorders are required to helpclinicians make optimal decisions about patient man-agement. These reference values may also help to in-crease the diagnostic power of these parameters inFrance. However, we believe that this model is poten-tially suitable for use with other assays and in othercountries, provided it is correctly validated. Based onthis extensive normative data set, our findings high-light the importance of interpreting serum IGF-1 andIGFBP-3 concentrations and their ratio in children as afunction of height and BMI SDS values. Further studiesare required to determine whether the relationship be-tween serum and auxological data are comparable inchildren with growth disorders or obesity and to ex-plore the associations between growth and growth fac-tors, with the use of longitudinal growth measure-ments from childhood to early adulthood.

Author Contributions: All authors confirmed they have contributed tothe intellectual content of this paper and have met the following 3 re-quirements: (a) significant contributions to the conception and design,acquisition of data, or analysis and interpretation of data; (b) draftingor revising the article for intellectual content; and (c) final approval ofthe published article.

Authors’ Disclosures or Potential Conflicts of Interest: Upon man-uscript submission, all authors completed the Disclosures of PotentialConflict of Interest form. Potential conflicts of interest:

Employment or Leadership: None declared.Consultant or Advisory Role: None declared.Stock Ownership: None declared.Honoraria: None declared.Research Funding: Grant from Pfizer-France to the Regional Insti-tute for Health (Tours, France).Expert Testimony: None declared.

Role of Sponsor: The funding organizations played no role in thedesign of study, choice of enrolled patients, review and interpretationof data, or preparation or approval of manuscript.

Acknowledgments: We thank Professor Paul Czernichow for hishelp in initiating the study and all doctors, nurses and others fromthe Regional Institute for Health in Western France who gave up timeto help us with the fieldwork. We also thank the laboratory staff at theRegional Institute for Health in Tours and at the Robert Debre Hos-pital in Paris for practical assistance in the performance of this study.We also thank all the families, especially the children, who partici-pated in this study.

References

1. Cohen P, Rogol AD, Deal CL, Saenger P, ReiterEO, Ross JL, et al. Consensus statement on thediagnosis and treatment of children with idio-pathic short stature: a summary of the GrowthHormone Research Society, the Lawson WilkinsPediatric Endocrine Society, and the EuropeanSociety for Paediatric Endocrinology Workshop.J Clin Endocrinol Metab 2008;93:4210–7.

2. Chen JW, Hojlund K, Beck-Nielsen H, SandahlChristiansen J, Orskov H, Frystyk J. Free ratherthan total circulating insulin-like growth factor-Idetermines the feedback on growth hormone re-lease in normal subjects. J Clin Endocrinol Metab2005;90:366–71.

3. Collett-Solberg PF. Update in growth hormonetherapy of children. J Clin Endocrinol Metab2011;96:573–9.

4. Sierra-Johnson J, Romero-Corral A, Somers VK,Lopez-Jimenez F, Malarstig A, Brismar K, et al.IGF-1/IGFBP-3 ratio: a mechanistic insight intothe metabolic syndrome. Clin Sci (Lond) 2009;116:507–12.

5. Bonnick SL, Shulman L. Monitoring osteoporosistherapy: bone mineral density, bone turnovermarkers, or both? Am J Med 2006;119:S25–31.

6. Szulc P, Seeman E, Delmas PD. Biochemical mea-surements of bone turnover in children and ado-lescents. Osteoporos Int 2000;11:281–94.

7. Bereket A, Turan S, Omar A, Berber M, Ozen A,Akbenlioglu C, et al. Serum IGF-1 and IGFBP-3levels of Turkish children during childhood and

adolescence: establishment of reference rangeswith emphasis on puberty. Horm Res 2006;65:96–105.

8. Brabant G, von zur Muhlen A, Wuster C, RankeMB, Kratzsch J, Kiess W, et al. Serum insulin-likegrowth factor I reference values for an auto-mated chemiluminescence immunoassaysystem: results from a multicenter study. HormRes 2003;60:53– 60.

9. Crofton PM, Evans N, Taylor MR, Holland CV.Serum CrossLaps: pediatric reference intervalsfrom birth to 19 years of age. Clin Chem 2002;48:671–3.

10. Elmlinger MW, Kuhnel W, Weber MM, Ranke MB.Reference ranges for two automated chemilumi-nescent assays for serum insulin-like growth fac-tor I (IGF-1) and IGF-binding protein 3 (IGFBP-3).Clin Chem Lab Med 2004;42:654–64.

11. Juul A, Bang P, Hertel NT, Main K, Dalgaard P,Jorgensen K, et al. Serum insulin-like growthfactor-I in 1030 healthy children, adolescents,and adults: relation to age, sex, stage of puberty,testicular size, and body mass index. J Clin En-docrinol Metab 1994;78:744–52.

12. Juul A, Dalgaard P, Blum WF, Bang P, Hall K,Michaelsen KF, et al. Serum levels of insulin-likegrowth factor (IGF)-binding protein-3 (IGFBP-3)in healthy infants, children, and adolescents: therelation to IGF-1, IGF-1I, IGFBP-1, IGFBP-2, age,sex, body mass index, and pubertal maturation.J Clin Endocrinol Metab 1995;80:2534–42.

13. Juul A, Holm K, Kastrup KW, Pedersen SA, Mi-chaelsen KF, Scheike T, et al. Free insulin-likegrowth factor I serum levels in 1430 healthychildren and adults, and its diagnostic value inpatients suspected of growth hormone deficiency.J Clin Endocrinol Metab 1997;82:2497–502.

14. Kong AP, Wong GW, Choi KC, Ho CS, Chan MH,Lam CW, et al. Reference values for serum levelsof insulin-like growth factor (IGF-1) and IGF-binding protein 3 (IGFBP-3) and their ratio inChinese adolescents. Clin Biochem 2007;40:1093–9.

15. Lofqvist C, Andersson E, Gelander L, Rosberg S,Blum WF, Albertsson Wikland K. Reference valuesfor IGF-1 throughout childhood and adolescence:a model that accounts simultaneously for theeffect of gender, age, and puberty. J Clin Endo-crinol Metab 2001;86:5870–6.

16. Lofqvist C, Andersson E, Gelander L, Rosberg S,Hulthen L, Blum WF, et al. Reference values forinsulin-like growth factor-binding protein-3(IGFBP-3) and the ratio of insulin-like growthfactor-I to IGFBP-3 throughout childhood andadolescence. J Clin Endocrinol Metab 2005;90:1420–7.

17. Rauchenzauner M, Schmid A, Heinz-Erian P, Ka-pelari K, Falkensammer G, Griesmacher A, et al.Sex- and age-specific reference curves for serummarkers of bone turnover in healthy children from2 months to 18 years. J Clin Endocrinol Metab2007;92:443–9.


18. Veldhuis JD, Roemmich JN, Richmond EJ, RogolAD, Lovejoy JC, Sheffield-Moore M, et al. Endo-crine control of body composition in infancy,childhood, and puberty. Endocr Rev 2005;26:114–46.

19. Xu S, Gu X, Pan H, Zhu H, Gong F, Li Y, et al.Reference Ranges for Serum IGF-1 and IGFBP-3Levels in Chinese Children During Childhood andAdolescence. Endocr J 2010.

20. Clemmons DR. Consensus statement on the stan-dardization and evaluation of growth hormoneand insulin-like growth factor assays. Clin Chem2011;57:555–9.

21. Wichers M, Schmidt E, Bidlingmaier F, Kling-muller D. Diurnal rhythm of CrossLaps in humanserum. Clin Chem 1999;45:1858–60.

22. Pokrajac A, Wark G, Ellis AR, Wear J, WieringaGE, Trainer PJ. Variation in GH and IGF-1 assayslimits the applicability of international consensuscriteria to local practice. Clin Endocrinol (Oxf)2007;67:65–70.

23. Leger J, Mercat I, Alberti C, Chevenne D, Ar-moogum P, Tichet J, et al. The relationship be-tween the GH/IGF-1 axis and serum markers ofbone turnover metabolism in healthy children.Eur J Endocrinol 2007;157:685–92.

24. Borghi E, de Onis M, Garza C, Van den Broeck J,Frongillo EA, Grummer-Strawn L, et al. Construc-tion of the World Health Organization childgrowth standards: selection of methods for at-tained growth curves. Stat Med 2006;25:247–65.

25. Royston P, Wright E. A method for estimatingage-specific reference intervals based on frac-tional polynomials and exponential transforma-tion. J R Stat Soc A Stat 1998;161:79–101.

26. Royston P, Wright EM. Goodness-of-fit statistics

for age-specific reference intervals. Stat Med2000;19:2943–62.

27. Chailurkit LO, Suthutvoravut U, Mahachoklert-wattana P, Charoenkiatkul S, Rajatanavin R. Bio-chemical markers of bone formation in Thai chil-dren and adolescents. Endocr Res 2005;31:159–69.

28. Fares JE, Choucair M, Nabulsi M, Salamoun M,Shahine CH, Fuleihan Gel H. Effect of gender,puberty, and vitamin D status on biochemicalmarkers of bone remodedeling. Bone 2003;33:242–7.

29. van Coeverden SC, Netelenbos JC, de Ridder CM,Roos JC, Popp-Snijders C, Delemarre-van de WaalHA. Bone metabolism markers and bone mass inhealthy pubertal boys and girls. Clin Endocrinol(Oxf) 2002;57:107–16.

30. Fall CH, Pandit AN, Law CM, Yajnik CS, Clark PM,Breier B, et al. Size at birth and plasma insulin-like growth factor-1 concentrations. Arch DisChild 1995;73:287–93.

31. Garnett S, Cowell CT, Bradford D, Lee J, Tao C,Petrauskas V, et al. Effects of gender, body com-position and birth size on IGF-1 in 7- and 8-year-old children. Horm Res 1999;52:221–9.

32. Hoppe C, Udam TR, Lauritzen L, Molgaard C, JuulA, Michaelsen KF. Animal protein intake, seruminsulin-like growth factor I, and growth in healthy2.5-y-old Danish children. Am J Clin Nutr 2004;80:447–52.

33. Ong K, Kratzsch J, Kiess W, Dunger D. CirculatingIGF-1 levels in childhood are related to bothcurrent body composition and early postnatalgrowth rate. J Clin Endocrinol Metab 2002;87:1041–4.

34. Ben-Shlomo Y, Holly J, McCarthy A, Savage P,

Davies D, Gunnell D, et al. An investigation offetal, postnatal and childhood growth withinsulin-like growth factor I and binding protein 3in adulthood. Clin Endocrinol (Oxf) 2003;59:366–73.

35. Bray I, Gunnell D, Holly JM, Middleton N, DaveySmith G, Martin RM. Associations of childhoodand adulthood height and the components ofheight with insulin-like growth factor levels inadulthood: a 65-year follow-up of the Boyd Orrcohort. J Clin Endocrinol Metab 2006;91:1382–9.

36. Brugts MP, Ranke MB, Hofland LJ, van der Wan-sem K, Weber K, Frystyk J, et al. Normal values ofcirculating insulin-like growth factor-I bioactivityin the healthy population: comparison with fivewidely used IGF-1 immunoassays. J Clin Endocri-nol Metab 2008;93:2539–45.

37. Sandhu J, Davey Smith G, Holly J, Cole TJ, Ben-Shlomo Y. Timing of puberty determines seruminsulin-like growth factor-I in late adulthood.J Clin Endocrinol Metab 2006;91:3150–7.

38. Gunnell D, Oliver SE, Donovan JL, Peters TJ, Gil-latt D, Persad R, et al. Do height-related varia-tions in insulin-like growth factors underlie theassociations of stature with adult chronic dis-ease? J Clin Endocrinol Metab 2004;89:213–8.

39. Rogers I, Metcalfe C, Gunnell D, Emmett P,Dunger D, Holly J. Insulin-like growth factor-I andgrowth in height, leg length, and trunk lengthbetween ages 5 and 10 years. J Clin EndocrinolMetab 2006;91:2514–9.

40. Woitge HW, Scheidt-Nave C, Kissling C, Leidig-Bruckner G, Meyer K, Grauer A, et al. Seasonalvariation of biochemical indexes of boneturnover: results of a population-based study.J Clin Endocrinol Metab 1998;83:68–75.

