Multimodal distribution versus logarithmic transformation of thyroid volumes in adolescents: detection of subgroup with subclinical thyroid disorders and its impact on the assessment of the upper limit of normal thyroid volumes.

摘要

Our objective was to evaluate whether there is a multimodal distribution of thyroid volume (ThV) in iodine-replete adolescents and to examine the relation between excessive ThV and the presence of thyroid hypoechogenicity (HE), serum thyroperoxidase antibodies (anti-TPO) and TSH levels >4.5 mU/l. ThV was measured by ultrasound in adolescents aged 13 yr (N = 1083) and 17 yr (N = 1089) from 22 schools in 6 districts of eastern Slovakia and expressed as ml and ml/m2 body surface area. For each age group the multimodal distribution of ThV values was tested by computing their frequency at intervals of 0.5 m/m2 and plotting the cumulative frequency on a probability scale in which each segment with normal distribution should give a straight line. In all examined subjects the HE was evaluated by ultrasound; in 924 (42.5%) of those anti-TPO was estimated by radioimmunoassay and TSH by immuno-electrochemiluminiscent assay. The medians of urinary iodine found in 55-164 spot urine samples from each of 6 districts (total number = 1003) were 126-142 microg/l, indicating an iodine-replete status. There was a trimodal distribution of ThV in each group, 80-85% in the lowest, 10-15% in the middle, and 5-7% in the upper segments. In the 10th ThV decile of the 17-yr group the frequency of HE (33/109 = 30.3%), anti-TPO (13/62 = 21.0%) and TSH (6/62 = 9.7%) was significantly higher than that in the 1st-9th decile (71/980 = 7.2%, P<0.001; 23/482 = 4.8%, P<0.001 and 5/482 = 1.0%, P<0.001, resp.). Similar differences were found in the 13-yr group (21/109 = 19.2% vs. 58/974 = 5.9%, P<0.001 for HE, 5/60 = 8.3% vs. 3/320 = 0.9%, P<0.001 for anti-TPO and 2/64 = 3.1% vs. 4/317 = 1.3% (not significant) for TSH >4.5 mU/l. Thus in the 10% of subjects with the highest ThV, the frequency of HE and anti-TPO was 4-5 times higher than in the remaining 90%. Our data indicate that an epidemiological evaluation of a large population of adolescents can detect a group with early signs of thyroid dysfunction (e.g. excessive ThV, increased frequency of HE, anti-TPO and TSH >4.5% mU/l), although such dysfunction may not be clinically apparent. This contrasts with numerous earlier reports which used a logarithmic transformation of the data in similar ThV sets, thus making the data appear homogeneous (unimodal) and with a normal distribution and obscuring the true multimodal distribution. This further prevents recognition of subjects with evidence of disordered thyroid status which thus become falsely included into a normal range.