Clinical Chemistry 43: 963-967, 1997;
(Clinical Chemistry. 1997;43:963-967.)
© 1997 American Association for Clinical Chemistry, Inc.
Free triiodothyronine concentration in serum of 1050 euthyroid children is inversely related to their age
Paul Verheecke
Centraal Laboratorium Hasselt, Elfde Liniestr. 27, B 3500 Hasselt, Belgium. Fax ++-3211–243291; e-mail paul.verheecke{at}club.innet.be
 |
Abstract
|
|---|
Free triiodothyronine was assayed in serum of 1050 euthyroid children
with the Amerlex-MAB method. We studied 50 samples for each year of
age, from <1 to 20, and no child had more than one sample taken. A
gradient of values was observed, increasing by 31% from 5.04 pmol/L as
the mean of 20-year-old patients to 6.59 pmol/L in the serum of
children younger than 1 year. For practical reasons the lower normal
limit was proposed to remain at 3.4 pmol/L for all patients; the higher
limit can be set at 8.3 (0 to 3 years), 7.8 (4 to 7 years), 7.1 (8 to
11 years), 6.8 (12 to 15 years), and 6.7 (16 to 19 years). Preferably,
the regression line of the mean + 2 SD can be used, resulting in a high
normal limit of 8.3 pmol/L at birth to 6.3 pmol/L at 20 years,
decreasing by 0.1 pmol/L per year.
Key Words: indexing terms: normal ranges • reference values
|
Introduction
|
|---|
Hyperthyroidism is unusual in children. It has been calculated to
occur as Graves disease in up to 0.2% and 0.4% of the pediatric and
adolescent population, respectively (1). The attention of
endocrinologists has recently been drawn to this subject
(2). The hallmark of hyperthyroidism is a low thyrotropin
(TSH) value.1
However, a low TSH
value can occur in a host of other situations. The next most important
assay in the differential diagnosis is a triiodothyronine
(T3) assay: T3 increases earlier than thyroxine
(3). Like all other patients, these children deserve to
have dependable reference ranges for the hormones they will have
assayed in their serum.
Although recommended in nearly all kit inserts, the setting of normal
ranges by individual laboratories is often very difficult, and more so
for children. Free T3 (fT3) presents a special
problem. The direct assay of free thyroid hormones is the subject of
multiple criticisms (4)(5). Nevertheless,
these assays have imposed themselves in multiple laboratories for
reasons of cost and convenience. Normal ranges of fT3 in
children are hard to come by, as very few references are available
(1)(6)(7). When introduced in our
laboratory as the Amerlex-MAB fT3 assay, I was asked to set
them.
|
Patients and Methods
|
|---|
methods
FT3 was assayed exactly as prescribed by the
manufacturers (Amerlex MAB; Ortho Clinical Diagnostics, Amersham, UK;
sold in Belgium by Orange, Brussels, Belgium). Temperatures were kept
constant within 1 °C, at 22 °C. Pipettes are checked every 8
weeks for precision and repeatability. Interassay variance was
controlled by the inclusion of aliquots of Lyphocheck levels II (lot
93002) and III (lot 93003). Level II had a mean of 8.06 pmol/L; the CV
was kept under 6.5% throughout the 6-month period. Level III had a
mean of 25.9 pmol/L with a CV under 8.0%. Control sample sets were run
at the beginning of each daily assay, after the 50th patient tube, and
at the end if >25 tubes were present after the last control set. All
assays were done singly. The tubes were counted in a Canberra Packard
(Zellik, Belgium) Riastar 5420 (Packard Instruments, Meriden, CT; sold
in Belgium by Canberra Packard), with a spline interpolation to plot
the results on the calibration curve. Descriptive statistics were
calculated with the package included in Quattro Pro (version 6.01 for
Windows; Novell, Orem, UT). With the same package, the significance of
differences between variances of the groups were calculated with the
F-test and the significance of differences between the means
of the groups with the t-test for groups with unequal
variances. Tests for normality were done according to Snedecor and
Cochran (8). The regression analysis was calculated with
the programs of Microsoft Excel for Windows95 (Microsoft, Redmond, WA).
patients
All samples used for this investigation were drawn with the
purpose of reaching a clinical diagnosis. After suffi-cient serum had
been set aside for all requested analyses, the remainder of the serum,
rather than being thrown directly to the waste, was used for this
investigation, which is kept totally anonymous. Therefore, our
procedures are in accordance with the Helsinki Declaration of 1975, as
revised in 1983.
On all serum samples entering the laboratory, fT3 was
assayed if the patient was <21 years old, all requested tests already
had the necessary volume taken off, the patient had not yet a sample in
the cohort, and no thyroid test was requested on any of this patient's
samples in the database of the laboratory. This precaution, taken
together with the low prevalence of the rarer thyroid diseases with a
normal TSH, makes it extremely unlikely that children with such a
disorder would be included in this study. A TSH test as routine
screening did not cause an exclusion of the serum, provided the result
proved normal. At the outset I decided to try to have 50 assays for
each year of age and to stop after 6 months. All assays were done
between August 16, 1995 and February 5, 1996. Four parameters were kept
for each serum: age, fT3 result, code number for the
presumptive diagnosis, and identification number of the sample. The
diagnosis was presumed on the basis of either, when available, the
reason for the request given by the practicing physician or the type of
assays requested and the results thereof. Diagnoses were then grouped
in classes. When the presumed diagnosis suggested a possible thyroid
involvement, TSH was assayed and the sample included only if this TSH
result was normal.
|
Results
|
|---|
The mean fT3 value of the 20-year-old patients is 5.04
pmol/L. The 50 children of <1 year have a mean fT3 of 6.59
pmol/L. A gradient decreasing from young to older can be observed in
Fig. 1
. Deviations from the gaussian distribution are present in
various amounts within the different groups; none reaches statistical
significance (data not shown). To have dependable reference
populations, results are grouped in five classes of 4 years of age each
(Table 1
), with the 20-year-old patients present as a check on the adult
reference range. Thus each class has 200 samples. The distribution
within the classes is normal for all, as shown in Table 1
, except for
the skewness of the 4–7-year-old class. This is due to the presence in
this class of two of three values from the 1050 being between -3 and
-4 SD. Deleting those two values gives a perfectly gaussian
distribution. To exclude the possibility of striking changes in the
first few weeks of life, the results of the children younger than 1
year are plotted on a separate graph (Fig. 2
). Among the five classes mentioned, the significance of the
differences between the variances was calculated with the
F-test. Because several of the differences in variances are
highly significant (P between 0.15 and 10-9), the
comparisons between the means of the classes were done with
t-tests for groups with unequal variances. Most of the
differences between the means of the classes are highly significant (P
between the nearest classes 0.035 to <10-7). The
regression analysis across the 21 groups shows a highly significant
correlation (P <10-50) between age and
fT3 value, although with a large variance within the
groups, which results in a correlation factor R of only
0.44, R2 = 0.19. The details of this calculation
are in Table 2
. To use all the results in the calculation of the high normal
limit, a second regression line is calculated, this one with the mean +
2 SD values of the classes. The resulting equation is x
= -0.098y + 8.29. Here again, x is
fT3 in pmol/L and y is the age in years. There
are 23 values higher than the mean +2 SD regression line, almost
exactly 2.5%,which confirms the gaussian distribution of the
population. Three of those values are between 3 and 4 SD above the
mean. One is a 14-year-old child who has Still disease (juvenile
arthritis), a second is a 17-year-old athlete after a training session,
and the third is an 18-year-old girl with an acute infection. Although
low fT3 values are of little diagnostic use, there are also
three values between 3 and 4 SD under the mean. They are from a
4-year-old girl with negative results for a radioallergosorbent test
(RAST) for pollen, from a 6-year-old boy with an acute mycoplasma
infection, and from a 17-year-old girl with a growth problem.
View larger version (18K):
[in this window]
[in a new window]
|
Figure 1. fT3 concentration in serum of 1050 euthyroid
children by year of age (50 each).
Normal range for adults: 3.4–6.2 pmol/L. The regression
line is drawn; the higher line is the linear fit
between the mean + 2 SD of the classes.  , individual results; 
(on the columns 1, 5, 9, 13, and 17), the corresponding four-year mean.
[triaf] on the same columns, mean + 2 SD.
|
|
View this table:
[in this window]
[in a new window]
|
Table 1. Statistical analysis of the results shown in Fig. 1
, grouped by four years, with the 20-year-old patients as a check on
adult values.
|
|
The most often quoted possible diagnosis is infection or serological
check after an infection (378 samples). The next most frequent
diagnosis is allergy (253 samples): In the younger children the serum
was then sent for RAST for milk allergens; in older children it was for
RAST for pollen, epithelia of domestic animals, or house dust mite.
Other frequent reasons for sending a sample were anemia (91),
serologic control before or after a vaccination (81), checks on
athletes (48 older children), and preoperative controls—generally ear,
nose, and throat (46). There were also samples for investigation of
acne (26), arthritis (24), pubertal anomalies (20), therapeutic drug
monitoring (13), and renal disease (13). This leaves 57 samples sent
for other reasons.
|
Discussion
|
|---|
The results presented here are most probably representative for a
normal pediatric population. There is some discrepancy with the results
of Butler et al. (6)(7) with the older version
of the Amerlex fT3. Their conclusion that fT3
is higher in children of 7 to 12 years than in younger children seems
to be due to too small a population. Furthermore, the Amerlex-MAB
method gives lower fT3 values than the older Amerlex-M. Our
reference values for fT3, as the 2.5 to 97.5 percentiles of
>3000 patients with a normal TSH, children included, are 3.4 to 7.2
pmol/L, identical to the normal reference interval stated on the kit
insert. When the children are excluded we obtain the 97.5 percentile at
6.14 pmol/L, very near to the value obtained by Piketty et al. with 98
healthy blood donors: 6.3 pmol/L (9).
Comparing different methods of fT3 determinations is a very
tricky business, especially in patients with nonthyroidal illness and
those treated with amiodarone (10). Fortunately, none of
the children in our series was treated with amiodarone. Nonthyroidal
illness to the point of interfering with thyroid metabolism is
certainly uncommon in children. However, I cannot exclude that some of
the few outliers to the lower values were due to that phenomenon,
especially the 6-year-old boy with an acute mycoplasma infection.
Deleting them ensures a gaussian distribution, so acceptance of the
regression line at +2 SD as the upper limit of normal seems warranted.
Nevertheless, it was a bit surprising to discover that the mean
fT3 of the 20-year-olds in our series (5.04 pmol/L) is
nearer to the equilibrium dialysis mean (5.15 pmol/L) than to the
Amerlex-MAB (5.57 pmol/L) mean of the controls in Sapin et al.'s paper
(10). But then, we are comparing a group of 50 with a
group of 30. With such low numbers the difference between the means is
not significant. Therefore, since the control group with the
equilibrium dialysis assay has its values within the normal range of
the Amerlex-MAB, we conclude that there is no difference in
fT3 results between Amerlex-MAB and the equilibrium
dialysis in controls, i.e., when the structure of the serum is normal.
When discussing the normal ranges for fT3 it is important
to realize that, in the assessment of a patient, fT3 is
never a parameter on its own. It always must be interpreted in
conjunction with the TSH result and, of course, in conjunction with the
other information the clinician has on the patient.
I present here reference values for fT3 in a pediatric
population of 1050 euthyroid children determined by the Amerlex-MAB
kit, evenly distributed over 0 to 20 years of age.The group of 50
children <1 year has a mean fT3 of 6.59 pmol/L with a SD
of 1.02 and a SEM of 0.128 pmol/L. There is no steep increase towards
the very early weeks of life. The group of 50 20-year-olds has a mean
of 5.04 pmol/L with a SD of 0.69 and a SEM of 0.097 pmol/L. Thus the
mean fT3 of the children <1 year old is at 2 SD and 16 SE,
or 31% above the mean of the 20-year-olds (P
<10-15). There is a continuous rising gradient from older
to younger. It remains true that most children have fT3
values within the normal ranges for adults. Nevertheless, the
interpretation of slightly increased values should be made easier,
taking into account the results of this investigation. Thus the lower
limit for normal ranges would be kept at 3.4 pmol/L for all ages. For
the higher limit, preference can be given to 8.3 pmol/L (0 to 3 years),
7.8 (4 to 7 years), 7.1 (8 to 11 years), 6.8 (12 to 15 years), and 6.7
(12 to 19 years). This division in groups of 4 years of age is
arbitrary. The best way to assess a specific value is to look at Fig. 1
for comparison. As it is usually not feasible for the clinical
laboratory to print it on the results proto- cols, it is
possible to follow the lead of Oesterling et al., dividing their huge
group of prostate-specific antigen in decades of age of the men whose
serum was analyzed (11). However, with the regression line
of the mean + 2 SD, the higher limit for each year can be calculated as
6.3 pmol/L at 20 years, adding 0.1 pmol/L for each year less, to 8.3 at
<1 year. This is obviously a more precise method.
Taking into account the criticisms leveled at the direct free hormone
assays by endocrinologists (4)(5), it is
unlikely that the physiological relevance of this observed trend in
fT3 concentration related to age will ever be unraveled.
|
Acknowledgments
|
|---|
I am indebted to A. Van Herck for explaining to me the
possibilities of the Quattro Pro and Microsoft Excel programs; to my
many colleagues, general practitioners, and pediatricians who sent us
the blood samples; and to the technicians of the radioassay department
of our laboratory for the care taken with the analyses. I also am
grateful to Jos Kint for the critical reading of an earlier draft of
this manuscript and for useful discussions.
|
Footnotes
|
|---|
1 Nonstandard abbreviations: TSH, thyrotropin; T3, triiodothyronine; fT3, free T3; and RAST, radioallergosorbent test. 
|
References
|
|---|
-
Foley T, Malvaux P, Blizzard R. Thyroid disease. In:
Kappy MS, Blizzard RM, Migeon CJ, eds. Wilkins the diagnosis and
treatment of endocrine disorders in childhood and adolescence, 4th ed.
Springfield, IL: Charles C Thomas, 1994:474, 494..
-
Perrild H, Grüters-Kieslich, Feldt-Rasmussen U, Grant D, Martino E, Kayser L, Delange F. Diagnosis and treatment of thyrotoxicosis in childhood. A European questionnaire study. Eur J Endocrinol 1994;131:467-473.
[Abstract/Free Full Text]
-
Ingbar S. The thyroid gland. Wilson JD Foster DW eds. Williams textbook of endocrinology 7th ed. 1985:719 Saunders Philadelphia. .
-
Stockigt JR. Serum thyrotropin and thyroid hormone measurements and assessment of thyroid hormone transport. Braverman LE Utiger RD eds. Werner and Ingbar's the thyroid 6th ed. 1991:470-474 J.B. Lippincott Co. Philadelphia. .
-
Nelson JC, Weiss RM, Wilcox RB. Underestimates of serum free thyroxin (T4) concentrations by free T4 immunoassays. J Clin Endocrinol Metab 1994;79:76-79.
[Abstract]
-
Butler J, Moore P, Mieli-Vergani G, Moniz C. Serum free thyroxin and free tri-iodothyronine in normal children. Ann Clin Biochem 1988;25:536-539.
-
Soldin SJ, Hicks JM. Pediatric reference ranges 1995:133 AACC Press Washington, DC. .
-
Snedecor GW, Cochran WG. Statistical methods, 6th
ed., 4th print. Ames, IA: The Iowa State University Press, 1971:552..
-
Piketty ML, D'Herbomez M, Le Guillouzic D, Lebtahi R, Cosson E, Dumont A, et al. Clinical comparison of three labeled-antibody immunoassays of free triiodothyronine. Clin Chem 1996;42:933-941.
[Abstract/Free Full Text]
-
Sapin R, Schlienger JL, Kaltenbach G, Gasser F, Christofides N, Roul G, et al. Determination of free triiodothyronine by six different methods in patients with non-thyroidal illness and in patients treated with amiodarone. Ann Clin Biochem 1995;32:314-324.
-
Oesterling JE, Jacobsen SJ, Chute CG, Guess HA, Girman CJ, Panser LA, Lieber MM. Serum prostate-specific antigen in a community-based population of healthy men. Establishment of age-specific reference ranges. JAMA 1993;270:860-864.
[Abstract/Free Full Text]
Top
Abstract
Introduction
