Abstract： Objective:To explore the differences in genetic variations and clinical features of children with congenital hypothyroidism (CH) with different thyroid morphologies.Methods:A retrospective study was conducted on 98 children with CH diagnosed at Changzhou Maternal and Child Health Care Hospital, Changzhou Medical Center, Nanjing Medical University, and Lianyungang Maternal and Child Health Care Hospital from August 19, 2011, to November 13, 2019. According to thyroid morphology, they were divided into the thyroid dysplasia (TD) group ( n=24), the gland-in-situ (GIS) group ( n=67), and the goiter group ( n=7). Whole exome sequencing was used to detect genetic variants, and pathogenic, likely pathogenic variants, and variants of uncertain significance were defined as potential functional variants. General condition, genetic variants, and treatment were compared between the three groups. Statistical analysis was performed using Chi-square and Bonferroni correction tests, Kruskal-Wallis test, or Mann-Whitney U test. Results:(1) The proportion of female infants in the TD group was higher than that in the GIS and goiter groups [87.5% (21/24) vs.47.8% (32/67) and 3/7, Bonferroni correction, respectively, both P<0.017]. The difference in serum thyroid stimulating hormone levels when diagnosed with CH was statistically significant in the TD, GIS, and goiter groups [128.00 mU/L (33.30-208.00 mU/L), 55.40 mU/L (17.73-116.00 mU/L), and 32.00 mU/L (21.55-57.65 mU/L), H=7.02, P=0.030], but was not statistically significant in pairwise comparisons (all P>0.017). (2) The detection rates of potential functional variants in the TD, GIS, and goiter groups were 45.8% (11/24), 88.1% (59/67), and 6/7, respectively, and the detection rate in the GIS group was higher than that in the TD group (Bonferroni correction, P<0.001). The detection rate of potential functional variants in DUOX2 gene was the highest [59.2% (58/98)], which was 20.8% (5/24), 73.1% (49/67), and 4/7 in the TD, GIS, and goiter groups, respectively, with statistically significant differences between groups ( χ2=20.02, P<0.001). Single allele variants were more common in the TD group (7/11), while double allele variants were more common in the GIS and goiter groups [71.2% (42/59) and 4/6, respectively], in addition, six cases of oligogenic variants were detected in the GIS group (10.2%, 6/59). (3) The difference in the dose of levothyroxine (μg/d) administered to children in the TD, GIS, and goiter groups was statistically significant at 2 years of age [37.50 (25.00-45.00), 25.00 (16.60-25.00), and 25.00 (16.50-40.00), H=16.53] and 3 years of age [37.50 (27.12-47.50), 20.00 (6.25-29.25), and 31.25 (9.38-52.50), H=14.16] (all P<0.001), and the dose of levothyroxine administered in the TD group was higher than that of the GIS group at the age of 2 and 3 years ( Z were -4.06 and -3.75, both P<0.017). Conclusions:The detection rate of underlying functional variants varies among children with CH with different thyroid morphologies. DUOX2 gene variants are the most prevalent and double allele variants are common. Infants and toddlers with TD may require higher doses of levothyroxine as they grow.