Mar 30, 1990 - nephelometer in samples from 479 healthy French chil- ... Assessment ofthe protein status of children in France ...... 1988;34:1878-80. 26.
CLIN. CHEM.
38/3, 394-399
(1992)
Laser Immunonephelometry Reference Intervals for Eight Serum Proteins
in
Healthy
Children Denis J.-M. Malvy,’ Catherine Herbert,5
Jean-Dominuue Pov#{233}da,2 Monique Debruyne,2 Bernard Montagnon,3 Pierre Cac#{232}s, Odile Houot,7 and Olivier Am#{233}d#{233}e-Manesme’
Eight serum proteins were analyzed with the Behring nephelometer in samples from 479 healthy French children, ages three to 16 years. Girls had higher concentrations of 1gMand albumin than boys had. Age appeared to be a main factor of variation for the proteins tested.
Reference intervals are presented for lgG, IgA, 1gM, albumin, transthyretin (prealbumin), retinal-binding protein, a1-acid glycoprotein, and C-reactive protein. The significance of increased concentrations of C-reactive protein within a community is discussed. Keyphrases: immunoglobulins . albumin . transthyretin C-reactive protein . acute-phase proteins - sexand age-related effects . pediatric chemistry . population studies
AddItional
Assessment of the protein status of children in France has included little information about the usual values of such biochemical indicators as the concentrations of serum proteins. Knowledge of the distribution of serum proteins could contribute to a better understanding of the results obtained in field studies involving biochemical epidemiology. The purpose of the present study was to determine reference values for proteins in a freeliving population of healthy children in an industrial-
ized country, France. Subjects
and Methods
Study Population We surveyed
479 healthy
children-234
boys (49%)
and 245 girls (51%)-ages 3-16 years and living in France. They had spontaneous (free-living) food intake and were free of any acute or chronic disease at the time of sampling. They had no surgical or medical history likely to modiiS’ their nutritional status. At the time of blood collection, the studied children were ambulatory and none had been pretreated in any way. The study was conducted between December 1985 and
Bernard
Burtschy,4
March 1986 in the areas of Tours (central France) and Nancy (eastern France). Children were invited to participate in the survey through a health examination. A third group consisted of ambulatory children in the Bordeaux area (southwestern France) who attended for routine health checkups. those children being admitted for minor surgical procedures such as ear-tube insertions, tonsillectomies, and adenoidectomies were
Only
asked to participate. The subjects were distributed among five age groups: from three to five years (n = 31), from six to eight years (n = 129), from nine to llyears(n = 135), from 12th 13 years (n = 68), and from 14 to 16 years (n = 116). The subjects were sampled from a population of 378 children from the Tours area, 201 from the Nancy area, and 27 from the Bordeaux area. We used a two-stage cluster sampling procedure. At first, a sample of three primary sampling units was
chosen (three areas in France). From each chosen primary sampling unit, one sample (the set of children) was selected. Cluster sampling in more than one stage is efficient when the primary sampling units are large and heterogeneous,
such as cities or countries:
in this case,
cluster sampling is often referred to as area sampling. The total sample size was determined so as to approximately yield a desired precision at the 5% confidence level through the use of relative information about the order of magnitude of the standard deviation of quantitatively measured variables. Proportional allocation is frequently used for the determination of the total sample size in relation to the individual strata because this method is quite simple and often quite effective. Practi-
cal contingencies
had slightly modified these theoretical of sample
points of view, according to the availability size required in the two main geographical
locations.
Sample Processing
Blood was collected from fasting, seated individuals 0800 and 1000 h. Blood samples were taken from a forearm vein by use of steel needles. Serum was rapidly separated by centrifugation and transferred to plastic test tubes for assay. Without delay, the tubes were covered with aluminum foil and 05-mL serum aliquots were placed on ice for 5 h. Serum aliquots were removed and stored frozen at -20 #{176}C for as long as between
‘Unite de Recherches en Hepatologie p#{233}diatrique,INSERM U056, et Departement de P#{233}diatrie, Centre Hospitalier de Bicfitre, 78 rue du G#{233}n#{233}ral Leclerc, F-94275 Bic#{234}tre Cedex, France. 2Laboratoire de Virologie, Centre Hospitalier Pellegrin, place Am#{233}lie Raba-L#{233}on, F-33076 Bordeaux Cedex, France. ‘Unite de Recherches sur la Nutrition et l’Alimentation INSERM UO1, HOpital Bichat, 170 blvd. Ney, F-75877 Paris Cedex, France. 4Stat. Unit. T#{233}l#{233}com Paris, 46 rue Barrault, F-75634 Paris
Cedex 13, France. Diagnostic Sud-Ouest, F-33000 Bordeaux, France. 6IRSA, rue de la Parmenti#{232}re, F-37000 Tours, France. 7Centre de M#{233}decine Preventive, 2 avenue du Doyen Parisot, F-54500 Vandoeuvre-les-Nancy, France. Received March 30, 1990; accepted January 13, 1992. ‘Behring
394
CLINICAL CHEMISTRY, Vol. 38, No. 3, 1992
six months until shipped on solid CO2 for laboratory analysis. Just before assay, the samples were thawed slowly to room temperature with minimal exposure to light. All analyses were performed at the same time, so no serum was refrozen.
We used a likelthood-ratio F-test to assess the signifof the interaction terms as a group. if the result was significant, we used a simultaneous test procedure to eliminate subsets of the first-order interactions, to simplify the model (4). This procedure permits the screening of effects in large models without increasing the overall size of the test. The adjusted regression coefficients and their 95% confidence intervals (CIs) for log serum values were transformed back to the original scale. The lower and upper reference intervals were defined by the 2.5 and 97.5 percentiles (5). We used the SAS statistical package with the procedure GLM for the analysis (6).
Assays of Serum Proteins
icance
Eight serum proteins-IgG, IgA, 1gM, albumin, transthyretin (prealbumin), retinol-binding protein, a1-acid glycoprotein (orosomucoid), and C-reactive protein (CRP)-were quantified by nephelometry (1, 2) with a laser nephelometer and monospecific antisera, diluent, standards, and controls (all from Behring Diagnostic, Rueil-Malmaison, France). Instrument settings are summarized in Table 1. The cumulative quality-control data for each protein during the analysis period are shown in each table of results. Data Analysis Statistical analysis of the data was conducted as follows: We used multiple-regression analysis to study the relation between the concentrations of serum proteins and the sex, age, and geographic location of the subjects. Because blood concentrations were generally skewed toward higher values but were nearly symmetrical on a logarithmic scale, we used log-transformed variables, which provided good approximations to a gaussian distribution. In all analyses, we used transformed and reweighted values by post-stratification for age and sex, according to 1986 French National Demographic Registration data, so as to give the study sample the sex and age distribution observed in the general population during the collection period (3). To explore possible differences among subgroups of
Resufts Table 2 gives the medians and reference intervals by sex and age for the serum data. Certain general trends by age are easy to interpret: in particular, albumin and transthyretin showed increases with age in both sexes. Sex differences and age trends were both accommodated in the regression model described above. Significant sex-related differences were found for 1gM and albumin. Age and sex interaction were a significant predictor of protein amount for two of the constituent proteins: transthyretin and orosomucoid (P
