Mutation spectrum analysis of 29 causative genes in 43 Chinese patients with congenital hypothyroidism.

Congenital hypothyroidism (CH) is the most common neonatal endocrine disorder with a genetic origin. The purpose of the present study was to analyze the mutation spectrum of CH patients in China. A targeted next‑generation sequencing panel covering all exons of 29 CH‑related causative genes was used in 43 Han Chinese patients with CH [11 dysgenesis and 32 glands in situ (GIS)]. The functional impact and pathogenicity of detected variants were analyzed using a comprehensive bioinformatics approach and co‑segregation studies. A total of 47 rare non‑polymorphic variants in 9 target genes associated with thyroid hormone synthesis (DUOX2, DUOXA2, TPO, TG, SLC26A4 and SLC5A5), thyroid stimulating hormone resistance (TSHR) and central hypothyroidism (PROP1 and TRHR) were identified in 31 patients (31/43, 72%). Of these variants, 8 were novel, including 3 in DUOX2, 2 in TPO, 3 in TSHR and 1 in SLC5A5. Variants were mostly affected by DUOX2, TG, TPO and TSHR. Approximately 44% of the patients (19/43) carried DUOX2 variants. The mutation detection rates in patients with GIS were higher compared with patients with dysgenesis [25/32 (78%) vs. 6/11 (54%)]. Oligogenic mutations were detected in 25.6% of the total cases and 35% of the mutated cases. Genetic basis was ascertained in 13 patients, reaching a diagnosis detection rate of 30%. In conclusion, genetic defects in dyshormonogenesis, mainly in DUOX2, were the main genetic cause of CH in the Chinese population. Oligogenicity is highly involved in CH pathogenesis and may thus be an important factor in common phenotypic variability observed in patients with CH.

Introduction

Congenital hypothyroidism (CH) is the most common congenital endocrine disorder with an incidence of approximately 1/2,000-4,000 newborns (1). According to the locations of lesions, CH can be classified into primary, central and peripheral hypothyroidism (1,2). Primary hypothyroidism accounts for >95% of CH cases (3), the majority of which (80-85%) are caused by alterations occurring during gland organogenesis. These alterations result in thyroid dysgenesis (TD) (1,4). The remaining cases (15-20%) are attributed to inborn defects in thyroid hormone synthesis. These defects are collectively known as dyshormonogenesis (DH) and are generally characterized by either goiter or normal thyroid glands (1,5,6). By contrast, central and peripheral hypothyroidisms are rare disorders.

Considerable progress has been made in the understanding of CH pathophysiology. Although most cases of CH occur sporadically, approximately 20% are familial and caused by genetic abnormalities (1,2). For TD, approximately 2–5% of reported cases have a genetic origin. Genes associated with TD (PAX8, NKX2-1/TTF-1, FOXE1/TTF-2, NKX2-5 and TSHR) play important roles during thyroid morphogenesis (1,4). The molecular mechanism of DH has been well characterized and most of the cases have been linked to mutations in genes involved in thyroid hormone synthesis (TG, TPO, DUOX1/2, DUOXA2, SLC5A5, SLC26A4/PDS, IYD/DEHAL1 and SECISBP2). These mutations are usually transmitted in an autosomal recessive mode (1,5,7). The underlying molecular basis of central and peripheral hypothyroidism remains unclear, although genetic ascertainment is possible in some cases (1,6). Mutations have been reported in genes controlling the biosynthetic pathway of thyroid stimulating hormone (TSH; TSHB, TRHR and IGSF1), pituitary development (POU1F1, PROP1, HESX1, LHX3, LHX4 and SOX3) (1,6) and thyroid hormone transport or action (SLC16A2/MCT8, THRB and THRA) (1). Other genes (FOXI1, GLIS3, UBR1 and ZNF252P) have been reported in cases with syndromic hypothyroidism or transient CH and may be involved in CH (8–13). These causative genes and their functions are described in Table SI.

Although CH can be classified as a disease with a strong genetic component, many issues remain unresolved. One is the commonly observed variable phenotype-genotype correlations in patients (5,14). This phenotypic or genetic heterogeneity suggests that mono- and polygenic factors and environmental modulators have roles in the determination of disease severity (4,5). Some cases have oligogenic mutations apart from single-gene mutations and demonstrate heterogeneous phenotypes to those carrying monogenic mutations (15–17). These cases may not be inherited in a monogenic manner; that is, a digenic or oligogenic inheritance may be considered, or mutations may occur, acting as a genetic modifiers (18,19). However, no definite evidence is able to prove this phenomenon. Next-generation sequencing (NGS) can be used for the simultaneous sequencing of multiple genes in a single sample and is useful in determining mutations in multiple genes that are potentially associated with diseases (20,21). Thus, NGS is a powerful tool for unraveling the pathogeneses of complex diseases. Given the genetic complexity and heterogeneity of CH, all known causative genes should be comprehensively screened for mutations for the proper understanding of CH pathogenesis.

According to the largest national newborn screening program between 2013 and 2015, the total incidence rate of CH in China is 4.13 per 10,000 live births, which is higher than the worldwide level (22,23). To date, the comprehensive screening of the known pathogenic genes in Chinese patients is limited. The present study designed a targeted NGS panel including 29 CH-related genes to screen mutations in a Chinese patient cohort from Shaanxi Province, China.

Materials and methods

Subjects

A total of 43 patients with CH from 42 families were recruited in Xi'an Children's Hospital and Chang'an Hospital, Xi'an, China, between October 2015 and August 2016. The age of the patients at the time of the study was 3 months-13 years. The inclusion criteria were: Positive neonatal screening with a diagnosis of CH confirmed by serum thyroid function tests at 2–4 weeks of age. Neonatal screening for CH was taken from 72 h to 7 days after birth. Blood samples were collected from the heel to determine TSH levels by using time-resolved fluorescence assay (PerkinElmer, Inc.). Newborns (2-4 weeks) with TSH levels of >10 µIU/ml were recalled for the re-examination of serum TSH and FT4 levels by electrochemiluminescence assay (Cobas 6000, Roche Diagnostics). CH diagnosis was based on elevated serum TSH (>7.5 µIU/ml; normal: 0.27–3.20 µIU/ml) and decreased FT4 levels (0.5–3.1 pmol/l; normal: 12–22 pmol/l). Levothyroxine (L-T4) treatment was initiated when elevated TSH level (>10 µIU/ml) was confirmed. All the patients had a phenotypical classification by thyroid ultrasonography performed during the neonatal period prior to treatment. Additional information on the possible existence of thyroid disease in family members was collected in all cases. The neonatal screening, diagnosis and follow-up of each patient were conducted in the same hospital: Xi'an Children's or Chang'an Hospital. In addition, 100 subjects with normal FT4 and TSH levels and undergoing neonate thyroid screening were included in the normal control group. They were all Han Chinese from Shaanxi, China and consisted of 45 males and 55 females with a mean age of 5 days (4–10 days) on the day of sample collection for neonatal screening. Blood samples from the fathers and mothers of 20 patients were collected for segregation analysis. At the time of the study, the mean age of these fathers and mothers was 30.7 years (22–42 years) and 29.3 years (20–45 years), respectively. The parents of all the participants gave written informed consent in accordance with the Declaration of Helsinki. The study was approved by the Medical Ethics Committees of Xi'an Children's Hospital and Chang'an Hospital.

DNA extraction and sequencing

Blood samples were collected from recruited patients, their family members and the control subjects and stored in EDTA tubes. Genomic DNA was extracted and analyzed as previously described (24).

According to the previous findings described in published literature (1,4–6) and the retrieval results of the Human Gene Mutation Database (HGMD Professional 2016, http://www.hgmd.cf.ac.uk/ac/index.php), 29 causative genes (PAX8, FOXE1, NKX2-5, TSHR, NKX2-1, DUOX2, DUOXA2, TPO, SLC26A4, FOXI1, TG, SLC5A5, IYD, SECISBP2, TSHB, IGSF1, TRHR, HESX1, LHX3, LHX4, POU1F1, PROP1, SOX3, THRB, THRA, SLC16A2, GLIS3, UBR1 and ZNF252P) associated with CH were selected. The recruited patients were genetically screened with a customized AmpliSeq panel (Thermo Fisher Scientific, Inc.) that included 29 CH-associated genes. The primers for the customized panel were designed with Ion AmpliSeq Designer (https://www.ampliseq.com/browse.action) for the inclusion of coding exons and the 20 flanking base pairs of the splice junctions surrounding the exons of the targeted genes. A total of 457 amplified amplicons were obtained at each sequencing run. Amplicon length was 125–374 bp (median 358 bp; Table SII). The amplicon library preparation and DNA template preparation and enrichment were conducted according to the manufacturers' protocols. DNA sequencing was performed with an Ion Torrent PGM instrument. An Ion PGM 200 sequencing kit and Ion 316™ Chip (Thermo Fisher Scientific, Inc.) were used, according to the manufacturers' protocols.

Variant detection and prioritization

Raw data were processed by the Torrent Suite software (version 5.0.4; Thermo Fisher Scientific, Inc.) for the generation of sequence reads. Each read was aligned to the hg19 human reference genome for the detection of variants. Called variants were functionally annotated with Ion Reporter (https://ionreporter.lifetechnologies.com/ir/secure/home.html) and ANNOVAR package (http://wannovar.wglab.org/). Identified variants were filtered as follows: i) Synonymous variants and nonsplice variants in the intronic region were excluded; ii) variants with minor allele frequencies (MAF) of ≤0.01 or no MAF values in the dbSNP database (http://www.ncbi.nlm.nih.gov/projects/SNP/), 1000 Genomes Project (http://ftp.ncbi.nih.gov/), Exome Sequencing Project (http://evs.gs.washington.edu/EVS/) and the Genome Aggregation Database (gnomAD, http://gnomad.broadinstitute.org/) were selected; iii) variants without rs numbers in the dbSNP database were considered novel rare variants; iv) CH-associated variants reported in the published literature or by the HGMD database (HGMD Professional 2019.3) were selected even if ii) or iv) was not met; v) all selected variants were validated through Sanger sequencing with ABI3500 ×L Dx (Applied Biosystems, Thermo Fisher Scientific, Inc.); and vi) all validated novel variants were determined in the normal control by Sanger sequencing and were filtered by MAF ≤0.01. The present study was conducted on the basis of these prioritization criteria. The prioritized variants were detected in the parental samples of patients for the verification of the inheritance of variants and segregation with phenotype.

Bioinformatics and statistical analysis

The possible functional effects of detected variants were assessed by in silico programs. For missense or indel variants, five in silico tools were used, including Polymorphism Phenotyping v2 (http://genetics.bwh.harvard.edu/pph2/ index.shtml), MutationTaster (http://www.mutationtaster.org/), Rare Exome Variant Ensemble Learner (https://sites.google.com/site/revelgenomics/) and Mendelian Clinically Applicable Pathogenicity (http://bejerano.stanford.edu/MCAP/). For the splicing variants, the deleterious effect on RNA splicing was predicted with MaxEntScan (http://genes.mit.edu/burgelab/maxent/Xmaxentscan_scoreseq.html), Berkeley Drosophila Genome Project (http://www.fruitfly.org/seq_tools/ splice.html) and NetGene2 (http://www.cbs.dtu.dk/services/ NetGene2/).

The evolutionary conservation analysis was performed using the CLC Sequence Viewer 6.5.2 software (CLC bio, Qiagen AB). Protein domains and structures were obtained from UniProt and InterPor Knowledgebase (http://www.uniprot. orssswq333 g/; http://www.ebi.ac.uk/interpro/).

The pathogenicity of each variant was assessed according to the standards described by the American College of Medical Genetics (ACMG) (25).

The clinical features of the two groups (such as patients with DUOX2 mutation and patients without DUOX2 mutation) were compared through a Mann-Whitney test for nonparametric values. P<0.05 was considered to indicate a statistically significant difference.

Results

Demographic and clinical characteristics of the patients

The present study included 43 Chinese Han patients diagnosed with primary CH. The recorded demographic data and clinical features of the patients are presented in Table I: The enrolled patients consisted of 25 females and 18 males, aged 3 months to 13 years. All of the cases were from unrelated families with no histories of thyroid diseases, apart from patients 32 and 33, who were monozygotic twins. Thyroid ultrasonography suggested that 11 patients had TD, of which 6 had athyreosis, 4 had hypoplasia and 1 had ectopy. A total of 32 cases had eutopically located glands in situ (GIS), 18 of which had normal thyroid sizes and 14 had goiter (Table I).

Table I.

Clinical Information, detected variants, and results of family segregation analysis of studied patients with CH.

					Screening		Neonatal period
Patients ID	Age[a], sex	Birth weight(g)	Gestational age (week+day)	Thyroid gland	TSH (uIU/ml)	Age	TSH (uIU/ml)	FT4 (pmol/l)	Detected variant	Father	Mother	Solved/ambiguous/unsolved
1	7y10m, F	3050	32+5	Hypoplasia	14.1	39d	15	6.7	SLC5A5 p.Q639* (CT)	NA	NA	Ambiguous
2	3y9m, F	3000	Full term	Athyreosis (53d)	35.4	69d	>100	1.2	TPO p.R361L (GT)	GG	GT	Ambiguous
3	1y1m, M	4000	Full term	Hypoplasia	20.3	76d	14.8	6.7	SLC26A4 p.A434T (GA), TRHR p.I168M (TG)	NA	NA	Ambiguous
4	1y3m, F	3400	Full term	Normal	28	25d	35.2	4.9	DUOX2 p.A1206T (GA)	GA	GG	Ambiguous
5	3m, M	3600	41	Normal	21.2	20d	92.8	3.6	DUOX2 p.E879K (GA)	NA	NA	Ambiguous
6	4y6m, M	3300	Full term	Goiter	29	74d	>100	2.6	DUOX2 p.K530* (AT), DUOX2 p.R1110Q (GA)	AA;GA	AT;GG	Solved
7	5m, M	3800	Full term	Hypoplasia	9.61	43d	20	5.7	DUOX2 p.T803fs (c.2406_2407insCCTG)	NA	NA	Ambiguous
8	4y, F	4000	40+5	Athyreosis (1y)	20	58d	36.5	4.8		NA	NA	Unsolved
9	7m, M	2750	36+5	Goiter	20.5	30d	30.2	5.1	DUOX2 p.R434* (CT)	CC	CT	Ambiguous
10	8m, F	3000	33+2	Normal	7.5	56d	16.9	6.6		NA	NA	Unsolved
11	1y9m, M	3800	Full term	Normal	19	60d	25	6.2	TSHR p.R528S (CA), TSHR p.R450H (GA)	CC;GA	CA;GG	Solved
12	3y, F	3400	Full term (Suspected)	Ectopy	>100	33d	>100	0.8	DUOX2 p.V779M (GA), SLC26A4 p.Y578H (TC)	GG;TC	GA;TT	Ambiguous
13	1y3m, F	3200	Full term	Normal	15.3	63d	20	6.4		NA	NA	Unsolved
14	2y9m, M	3400	Full term	Goiter	24.2	33d	32.8	5		NA	NA	Unsolved
15	2y4m, M	3900	Full term	Normal	21	57d	28	5.3	DUOXA2 p.R94C (CT), DUOXA2 p.Y246* (CG), TG p.N212S (AG)	CC;CG; AG	CT;CC, AA	Solved
16	11m10d, M	3200	Full term	Goiter	100	62d	>100	1.1	DUOX2 p.R376W (CT), DUOX2 p.R434_S440del (del/wt), DUOX2 p.R1110Q (GA)	CT;del/wt;GG	CC;wt/wt;GA	Solved
17	11m, M	3100	Full term	Normal	14.6	65d	19.54	6.5	DUOX2 p.R885L (GT), DUOX2 p.Y1415C (AG)	GT;AA	GG; AG	Solved
18	2y2m, F	2600	Full term	Normal	77	67d	77	4.1	TG p.R2585W (CT), DUOX2 p.A1206T (GA)	NA	NA	Ambiguous
19	1y, M	3050	full term	Hypoplasia	18	32d	21	6.3		NA	NA	Unsolved
20	9y, F	3500	Full term	Normal	>100	20d	>100	0.7	TPO p.E757* (dupT/wt), TG p.I1931V (AG), DUOX2 p.K530* (AT)	wt/wt;AA; AA	dupT/wt; AG;AT	Ambiguous
21	9y, F	3000	Full term	Athyreosis	>100	30d	>100	0.5		NA	NA	Unsolved
22	4y7m, M	3500	Full term	Normal	31	49d	40	4.8		NA	NA	Unsolved
23	3y, F	3000	Full term	Goiter	40	74d	40	4.8	TG p.R896Q (GA)	NA	NA	Ambiguous
24	1y6m, M	3500	Full term	Normal	19	61d	58.21	4.5	TPO p.R846W (CT), TG p.E955fs (c.2864delA), TG p.I1931V (AG), TG p.L2282fs (c.6840_6843delTTGT)	CT;del/wt; AG, wt/wt	CC;wt/wt;AA, del/wt	Solved
25	8m, F	2400	Full term	Normal	58.21	32d	>100	6.7	TG p.I2394M (CG)	CC	CG	Ambiguous
26	13y, F	3900	Full term	Goiter	16.5	28d	>100	6.5		NA	NA	Unsolved
27	2y6m, F	3100	38+1	Goiter	25.6	35d	35	6.3		NA	NA	Unsolved
28	2y, F	3000	40	Goiter	26	72d	46.5	4.7	TG p.V1738I (GA), TG p.S1912N (GA), DUOX2 p.D137E (CA), DUOX2 p.R432H (GA), DUOXA2 p.Y246* (CG)	GA;GA; CA;GG; CG	GG;GG; CC;GA; CC	Solved
29	3y3m, F	3300	Full term	Goiter	>100	58d	>100	2	DUOX2 p.R625* (CT), DUOX2 p.R1110Q (GA)	CC;GA	CT;GG	Solved
30	6y, M	3000	37	Goiter	18	58d	20	6.4	TPO p.S571R (CG), PROP1 p.G51V (GT)	CG;GG	CC;GT	Ambiguous
31	5y6m, F	2900	40	Normal	20.6	53d	18.3	6.5		NA	NA	Unsolved
32	5y11m, M	2200	37+1	Normal	20	65d	18.2	6.5	DUOX2 p.F591S (TC), DUOX2 p.E879K (GA), DUOX2 p.G1521* (GT)	TC;GA; GG	TT;GG; GT	Solved
33	5y11m, M	2600	37+1	Normal	22.5	65d	20.3	6.3	DUOX2 p.F591S (TC), DUOX2 p.E879K (GA), DUOX2 p.G1521* (GT)	TC;GA; GG	TT;GG; GT	Solved
34	1y4m, F	3300	40+1	Goiter	16.5	71d	15.8	6.7	DUOX2 p.R885L (GT), DUOX2 p.R1110Q (GA)	GG;GA	GT;GG	Solved
35	2y10m, F	3650	Full term	Athyreosis	>100	52d	>100	0.8		NA	NA	Unsolved
36	1y11m, F	3400	39+4	Athyreosis	>100	47d	>100	0.9		NA	NA	Unsolved
37	1y, M	3750	Full term	Goiter	35.1	78d	35	4.9	TSHR p.C176R (TC), TSHR p.K618* (AT), TPO p.P883S (CT)	TT;AT; CC	TC;AA; CT	Solved
38	1y2m, F	3300	Full term	Athyreosis	46.1	64d	>100	1.9	TPO p.P883S (CT), GLIS3 p.A753V (CT)	NA	NA	Ambiguous
39	9m, F	3600	39+2	Normal	>100	47d	>100	2.1	DUOX2 p.E389K (GA)	NA	NA	Ambiguous
40	8m, F	1900	38	Normal	15	33d	15	7.1	TPO p.S309P (TC)	NA	NA	Ambiguous
41	4y9m, F	2800	Full term	Normal	21	90d	18	6.5	DUOX2 p.G624fs (c.1871delG)	NA	NA	Ambiguous
42	1y3m, F	2700	41+2	Goiter	>100	68d	>100	0.7	SLC26A4 p.Y78H (TC), DUOX2 p.R885L (GT), DUOX2 p.R1110Q (GA)	TT, GG, GA	TC, GT, GG	Solved
43	1y1m, M	3800	40	Goiter	>100	74d	>100	0.6	TPO p.R846W (CT), TG p.R896Q (GA), c.3693+1G>T (GT) DUOX2 p.V407F (GT), DUOX2	NA	NA	Ambiguous
Normal			<10		0.27–3.2	12-22

the age of each patient was calculated based on the date of birth and sample collection. CH, congenital hypothyroidism; m, month; d, day; y, year; F, female; M, male; TSH, thyroid-stimulating hormone; FT4, free tetraiodothyronine; *, a nonsense mutation which caused a premature termination in gene product; NA, data not available.

Sequencing data analysis

NGS analysis was performed on the 43 patients with CH and the result showed that the number of mapped reads for individual samples was 149,698-1,458,751 (median: 394571, n=43). The percentages of the on-target sequences in each sample all had a median of 99%, with an average base coverage depth ranging from 215.9–3× in individual samples. The average total coverages of all the targeted bases were 94.80% at 20×, 87.06% at 100× and 57.51% at 500×. Coverage was also uniform across all samples. On average, 94.8% of the called bases had a quality score of ≥Q20 (Table SIII).

In 25 genes (86.2%), all of the targeted amplicons were covered by 20 reads or more (Fig. S1). Exons that were covered less than 20× or missed by the NGS were subjected to Sanger sequencing (Table SIV).

Variant detection and mutation spectrum

Following functional filtering, 47 rare nonpolymorphic variants were identified in 31 of 43 patients (71%). These variants were distributed in 9 genes, including 6 DH-associated genes (DUOX2, TG, TPO, SLC26A4, DUOXA2 and SLC5A5), 1 gene associated with TD (TSHR) and 2 genes associated with central hypothyroidism (PROP1 and TRHR, Fig. 1A). Notably, all variants were detected in heterozygous status in patients. Various types of mutations have been detected and most of them were missense variants (Table II; Fig. 1A). A total of 8 novel variants were identified (Table II; Fig. S2); these were absent in the local control samples. The 39 remaining variants had been reported in HGMD, dbSNP, gnomAD and/or 1000 human genome databases.

Figure 1.

(A) Number of variants identified in each candidate gene according to the mutation type. (B) Number of variants identified in each candidate gene according to the CH phenotype. CH, congenital hypothyroidism; GIS, CH with gland in situ; TD, CH with thyroid dysgenesis.

Table II.

Potential pathological variants detected in the present study.

				Minor allele frequency
Gene	Amino Acids change	cDNA change	Exon/Intron position	rs ID	Patients (n=43[a])	GnomAD east asian	1000 Genome CHB	Status[b]	ACMG classification
TSHR	p.C176R	c.526T>C	6		0.012	0	0	Novel	LP
TSHR	p.R450H	c.1349G>A	10	rs189261858	0.012	0.002597	0.0049	Known[d], DM	P
TSHR	p.R528S	c.1582C>A	10		0.012	0	0	Known[d], DM	LP
TSHR	p.K618*	c.1852A>T	10		0.012	0	0	Novel	P
DUOX2	p.D137E	c.411C>A	5		0.012	0	0	Novel	LP
DUOX2	p.R376W	c.1126C>T	10	rs119472029	0.012	0	0	Known[d], DM	LP
DUOX2	p.E389K	c.1165G>A	11		0.012	0	0	Novel	VUS
DUOX2	p.V407F	c.1219G>T	11		0.012	0	0	Known[d], DM	VUS
DUOX2	p.R432H	c.1295G>A	12	rs530736554	0.012	0.0004769	0	Known[d], DM	LP
DUOX2	p.R434*	c.1300C>T	12	rs119472026	0.012	0.000116	0	Known[d], DM	P
DUOX2	p.R434_	c.1300_1320
	S440del	delCG	12		0.012	0	0	Known[d], DM	P
		AGATATGGGG
		CTGCCCAGC
DUOX2	p.K530*	c.1588A>T	14	rs180671269	0.024	0.009274	0.0095	Known[d], DM	P
DUOX2	p.F591S	c.1772T>C	15		0.024	0.00007081	0	Known[c]	VUS
DUOX2	p.G624fs	c.1871delG	16	rs769258094	0.012	0.0004638	0	Known[d], DM	P
DUOX2	p.R625*	c.1873C>T	16	rs770083296	0.012	0	0	Known[d], DM	P
DUOX2	p.V779M	c.2335G>A	19	rs145061993	0.012	0.004094	0.0049	Known[d], DM?	VUS
DUOX2	p.T803fs	c.2406_2407	19		0.012	0	0	Novel	P
		insCCTG
DUOX2	p.E879K	c.2635G>A	20	rs774556391	0.036	0.000954	0	Known[d], DM	P
DUOX2	p.R885L	c.2654G>T	20	rs181461079	0.036	0.005777	0.0049	Known[d], DM?	LP
DUOX2	p.R1110Q	c.3329G>A	25	rs368488511	0.06	0.002597	0.0049	Known[d], DM	P
DUOX2	p.A1206T	c.3616G>A	28	rs762588205	0.024	0.0001739	0	Known[d], DM?	LP
DUOX2	IVS28+1G>T	c.3693+1G>T	intron 28	rs200717240	0.012	0.001537	0	Known[d], DM	P
DUOX2	p.Y1415C	c.4244A>G	32	rs757012152	0.012	0.00008334	0	Known[c]	LP
DUOX2	p.G1521*	c.4561G>T	34	rs765781255	0.024	0.001044	0	Known[d]	LP
DUOXA2	p.R94C	c.280C>T	3		0.012	0	0	Known[d], DM	VUS
DUOXA2	p.Y246*	c.738C>G	5	rs4774518	0.024	0.00188	0.0291	Known[d], DM	P
SLC5A5	p.Q639*	c.1915C>T	15		0.012	0	0	Novel	VUS
TPO	p.S309P	c.925T>C	8		0.012	0	0	Novel	VUS
TPO	p.R361L	c.1082G>T	8	rs201781919	0.012	0.009273	0.0194	Known[d], DM	VUS
TPO	p.S571R	c.1713C>G	10		0.012	0	0	Novel	VUS
TPO	p.E757*	c.2268dupT	13	rs770781635	0.012	0.00159	0	Known[d], DM	P
TPO	p.R846W	c.2536C>T	15	rs28913014	0.024	0.00159	0.0049	Known[c]	VUS
TPO	p.P883S	c.2647C>T	16	rs190968346	0.024	0.005409	0.0146	Known[d], DM	LB
PROP1	p.G51V	c.152G>T	2	rs2233783	0.012	0.002151	0	Known[c]	VUS
SLC26A4	p.Y78H	c.232T>C	3	rs760794201	0.012	0	0	Known[d], DM	VUS
SLC26A4	p.A434T	c.1300G>A	11	rs757552791	0.012	0	0	Known[d], DM	VUS
SLC26A4	p.Y578H	c.1732T>C	16	rs781728302	0.012	0	0	Known[c]	VUS
TRHR	p.I168M	c.504T>G	1	rs13306060	0.012	0.00212	0.0049	Known[c]	VUS
TG	p.N212S	c.635A>G	5	rs187737243	0.012	0.002122	0	Known[c]	VUS
TG	p.R896Q	c.2687G>A	10	rs374707675	0.024	0.0007431	0.0049	Known[d], DM	VUS
TG	p.E955fs	c.2864delA	11	rs767858769	0.012	0.00005798	0	Known[c]	P
TG	p.V1738I	c.5212G>A	26	rs115053637	0.012	0.001325	0.0097	Known[c]	VUS
TG	p.S1912N	c.5735G>A	31	rs762807254	0.012	0	0	Known[c]	VUS
TG	p.I1931V	c.5791A>G	31	rs115877910	0.024	0.002391	0.0146	Known[d], DM?	VUS
TG	p.L2282fs	c.6840_6843	39	rs774153375	0.012	0	0	Known[c]	P
		delTTGT
TG	p.I2394M	c.7182C>G	41		0.012	0.00005798	0	Known[c]	VUS
TG	p.R2585W	c.7753C>T	44	rs114211101	0.012	0.005379	0.0049	Known[d], DM?	VUS

There were a pair of twins in the studied patients, therefore, when calculating variant frequency, the total number of patients were counted as 42

status evaluated based on whether variants are reported in public databases or published literature.

variants were reported in public population databases, such as dbSNP, ExAc, or 1000 Genomes Project, but without phenotypic data and pathological assessment

variants were reported in the published literature as well as HGMD database (HGMD Professional 2019.3); DM, disease-causing mutation; DM?, a possible disease-causing mutation; DFP, disease-associated polymorphism with supporting functional evidence; fs, frameshift; *, a nonsense mutation which caused a premature termination in gene product; NA, data not available; P, pathogenic; LP, likely pathogenic; VUS, variants of uncertain significance; LB, likely benign.

Of the 9 mutated genes, the gene with the highest number of variants was DUOX2, followed by TG, TPO and TSHR (Fig. 1A). Twenty variants in DUOX2 were identified in 19 cases (19/43, 44%), that is, 17 patients with GIS and 2 patients with TD (Fig. 1B). Of these patients, 10 (10/43, 23%) carried ≥2 different heterozygous variants in DUOX2. A total of 6 patients carried DUOX2 mutation(s) in association with mutation(s) in TG (n=4) or SLC26A4 (n=2). The most common mutation was p.R1110Q (DUOX2: c.3329G>A), which was found in 5 patients, accounting for 11% of all the cases. Of the 3 novel variants in DUOX2, p.T803fs was a frameshift mutation and had a potential deleterious effect on protein function and p.D137E and p.E389K were missense mutations located in the peroxidase-like domain (Fig. S3A).

A total of 9 variants in TG were identified in 8 CH patients (8/43, 18.6%), 2 of which had ≥2 TG variants. Apart from carrying TG mutation(s), 6 cases also had mutation(s) in genes associated with DH (SLC26A4, DUOX2, DUOXA2 and TPO).

A total of 6 TPO variants were separately found in 6 patients (6/43, 14%) in heterozygous status. All but 1 patient had a TPO mutation in association with mutation(s) in different genes. A total of 2 novel variants, p.S309P and p.S571R, were located in a myeloperoxidase-like domain, the catalytic site of the enzyme (Fig. S3B).

A total of 4 TSHR variants were found in 2 patients and were compound heterozygotes for 2 different TSHR mutations. The TSHR variant p.R450H was a recurrent inactivating mutation (26) and p.C176R and p.K618 were novel. p.C176R is located in the leucine-rich repeat region of the extracellular domain and responsible for high-affinity hormone binding and p.R528S and p.K618* are located in the cytoplasmic loops (Fig. S3C).

Patients with GIS had a higher tendency to be affected with mutations than patients with TD [25/32 (78%) vs. 6/11 (54%), Fig. 2]. Variants in TG, TSHR, DUOXA2, SLC5A5 and PROP1 genes were found exclusively in patients with GIS, and 1 variant in TRHR was found in patients with TD. Other genes, including DUOX2, TPO and SLC26A4, were associated with either dysgenesis or GIS phenotype (Table II and Fig. 1B). The variants detected in the 6 patients with TD (two athyreosis, two hypoplasia and one ectopy) were all located in genes associated with DH. A total of 12 patients (12/43, 28%) carried only 1 heterozygous variant and 19 cases (19/43, 44%) had ≥2 variants, 8 of which were monogenic (having mutations in the same gene) and 11 were oligogenic (having mutations in different genes, Table III).

Figure 2.

Percent distribution of detected variants according to the type of CH. CH, congenital hypothyroidism; GIS, CH with gland in situ; TD, CH with thyroid dysgenesis.

Table III.

Mutation spectrum of ‘solved’ and ‘ambiguous’ cases with CH.

Solved (n=13), all were CH with GIS
	Monogenic (n=8)			Oligogenic (n=5)
Gene	Number of variants	Number of patients	Gene1	Number of variants	Gene2	Number of variants	Gene3	Number of variants	Number of patients
TSHR	2	1	TSHR	2	TPO	1			1
DUOX2	2	4	DUOXA2	2	TG	1			1
DUOX2	3	3	DUOX2	2	SLC26A4	1			1
			DUOX2	2	TG	2	DUOXA2	1	1
			TG	3	TPO	1			1
Ambiguous (n=18)
Monogenic(n=12)			Oligogenic (n=6)
Gene	Number of variants	Number of patients	Gene1	Number of variants	Gene2	Number of variants	Gene3	Number of variants	Number of patients
DUOX2	1	6	SLC26A4	1	TRH1	1			1
TG	1	2	DUOX2	1	SLC26A4	1			1
TPO	1	3	DUOX2	1	TG	1			1
SLC5A5	1	1	DUOX2	1	TG	1	TPO	1	1
			DUOX2	2	TG	1	TPO	1	1
			TPO	1	PROP1	1			1

CH, congenital hypothyroidism.

Pathogenicity assessment

The pathogenicity of the detected variants was classified in accordance with ACMG standards and guidelines (Table II, Tables SV and SVI). Among the 47 variants, 25 were classified as pathogenic (P) or likely pathogenic (LP), namely, 16 in DUOX2, 4 in TSHR, 2 in TPO, 2 in TG and 1 in DUOXA2 gene. A total of 21 variants were classified as variants of uncertain significance (VUS) and 1 variant in TPO was classified as likely benign. Among the 8 novel variants, 4 were classified as P (p.C176R and p.K618* in TSHR, p.T803fs in DUOX2) or LP (p.D137E in DUOX2), the other were classified as VUS.

Genotype and phenotype relationship

Through family segregation and pathogenicity assessment, 13 cases (patients 6, 11, 15,17, 16, 28, 29, 32, 33, 34, 37 and 42) were considered ‘solved,’ reaching a diagnosis detection rate of 30% (Tables I and III, and Fig. S1). These ‘solved’ cases all carried at least 2 pathogenic variants in the same gene, which were of either paternal or maternal origin, but not from a single parent, following the identification of a decisive link between genotype and phenotype. A total of 18 cases (41.9%) were considered ‘ambiguous’ owing to the weak link between genotype and phenotype. In addition, 12 cases were considered ‘unsolved’ because they carried no mutations in any of the listed genes.

Among the solved cases, 8 were monogenic and 5 were oligogenic. The main pathogenic genes were DUOX2 (n=9), TSHR (n=2), DUOXA2 (n=1) and TG (n=1). Notably, all of the ‘solved’ cases were patients with GIS (Tables I and III). Therefore, the diagnosis rate in patients with GIS patients was 40.6% (13/32; Table III).

According to the number of variants carried, the studied cases were classified into different groups and the serum levels of TSH and FT4 were compared among these groups (Fig. SIV). The results showed that only the average serum TSH level of patients with TG mutation at diagnosis were significantly higher than those without TG mutation (49.54 µIU/ml vs. 68.71 µIU/ml, P=0.037, Fig. S4A-b). The average serum levels of FT4 of patients with monogenic mutation at diagnosis were higher compared with patients with oligogenic mutations, but the difference was not statistically significant (P=0.05, Fig. S4D-c).

Discussion

To the best of our knowledge, the present study is the first in which the currently largest targeted NGS panel containing 29 known causative genes was used for the comprehensive examination of the mutation spectrum of Han Chinese CH patients. The present study found a high mutation rate (44%) in primary CH patients and most of the mutations (91.5%) were identified in genes associated with DH. In addition, mutations in genes associated with thyroid development or transcription were rarely identified. The majority of CH were caused by TD and <20% of cases showed strong genetic predisposition (1,2). However, in the patient cohort, DH (n=32) was more common than TD (n=11). This result is in agreement with the data reported in China (27–31). Given that DH is largely caused by genetic defects and considered a hereditary disease, a majority of CH cases in Chinese are hereditary and have a strong genetic origin. In the present study, mutation detection rate in CH patients with DH was 78% and the diagnosis rate in DH was 40.6%. Notably, the 6 patients with TD harbored mutations that were all associated with DH, 2 of whom had athyreosis. Patients with TD, especially those with athyreosis, are unlikely to carry variants associated with DH. However, in the present study, patients were subjected to thyroid ultrasonography rather than to whole-body nuclear magnetic resonance scanning for the examination of thyroid morphology. Thus, those examined as athyreosis could not be excluded for the likelihood of ectopy. Similarly, other studies have found variants associated with DH in patients with athyreosis (30,32). In addition, 2 individual variants in genes associated with pituitary development or central CH (PROP1 and TRHR) respectively were found in 1 CH patient with GIS and 1 patient with TD, and both variants co-occurred with genetic variants associated with DH. Currently, it is difficult to be sure that the genetic defects associated with DH contribute to the development of TD or the potential effect of variants associated with pituitary development on primary CH. However, the findings of the present study validated the complicated pathological mechanism of CH. Thus, studies on the genetic origin of TD or DH diseases should not be limited to well-known causative genes (4,33).

In the present study, DUOX2 was the most common genetic alteration identified in CH patients. The detection rate of DUOX2 mutation in the studied CH cohort and the GIS patients were 44% (19/43) and 53.1% (17/32), respectively, most of which carried ≥2 DUOX2 mutations. This finding was in accordance with those previously reported showing that DUOX2 mutation is the leading genetic cause of CH in Asian populations, including other Han Chinese, Japanese and Koreans; the detection rate of DUOX2 in patients from Asian populations is 16.5–3% and ≤83% in patients with DH (27,29,31,34–40). TG and TPO mutations were also commonly found in the studied cohort. However, a majority of these mutations were either separately presented at a heterozygous status or detected with mutations in different genes. Thus, TG and TPO may act as contributing genetic factors apart from being the main genetic causes of CH in the studied cohort. In some Caucasian cohorts, TPO has been identified as the main genetic cause of CH (41–43). In addition, loss-of-function (LOF) mutations in the TSH receptor (TSHR) gene were identified as the most frequent cause of TSH resistance, leading to a wide spectrum of phenotypes ranging from severe CH to mild euthyroid hyperthyrotropinemia (26,44). In the present study, 2 patients with CH who were compound heterozygotes for 2 different TSHR mutations (1 for p.R528S and p.R450H and 1 for p.C176R and p.K618*) demonstrated mild clinical phenotypes (5 pmol/l≤FT4<10 pmol/l) (45). This finding is in agreement with previous studies reporting that compound heterozygotes of less severe LOF mutations are usually associated with mild/borderline forms of hypothyroidism, wherein an appropriate increase in TSH serum levels can compensate for the reduced sensitivity of the thyroid (partially or fully compensated TSH resistance) (26,44,46,47). Previously reported clinical cases with TSHR mutations are always characterized by normal-sized or hypoplastic thyroid gland (26,46,47). However, in the present study, patient 37 suffered from goiter. In addition to TSHR mutations, this patient also carried a heterozygous variant in TPO, which may be the reason leading to this phenotypic variability.

At present, patients with CH caused by genetic defects are considered to be inherited in a monogenic manner. However, phenotypic variability observed in patients with same mutations indicated the influence of other factors, such as genetic heterogeneity (15,34,42,43,48–50). The present study found a high percentage (25.6%) of involvement of oligogenic mutations in studied cases, which is similar to that of previous studies simultaneously assessing multiple genes (15,30–32). These studies reported frequent oligogenic involvement in CH, with oligogenic mutations in 20–43.5% of patients with CH and GIS and/or patients with TD in different ethnic populations. In addition, among the 13 ‘solved’ cases through family segregation and pathogenicity assessment, 5 cases carried oligogenic mutations and none of the mutations was inherited from a single parent. These findings suggested that, not only monogenic inheritance, but also digenic or oligogenic inheritance is involved in the pathogenesis of CH. However, available evidence is insufficient for oligogenic inheritance verification, and protein-protein interaction for the two proteins or genes, pedigree data, animal models or very specific functional experiments are key factors (18). At present, only 1 study performed in mice demonstrated a multigenic origin of CH with TD (51). Therefore, further studies are required for the validation of the role and mechanism of oligogenicity in CH pathogenesis.

The data of the present study were compared with those of several similar studies that analyzed the mutation spectrum of CH patients in China by using NGS (29–31). The investigated patients in the present study were from northwestern China (Shaanxi Province), whereas those in the previous studies were mainly from southern China (Jiangsu and Guangxi Provinces). The general mutation profiles of patients with CH demonstrated by these studies were similar. For example, the total mutation rate in CH patients reported in these studies was relatively high, i.e., 48.5% (29), 65.09% (31), 80.9% (30) and 72% (the present study). DUOX2 mutations were the prevalent genetic alterations in these studies, with a mutation rate of 31.8% (29), 31.3% (31), 60% (30) and 44% (the present study). Therefore, the region (or population)-specific characteristics in patients with CH from these studies could not be ascertained. However, some different findings in the present study were observed. For example, DUOX2 and TG mutations were the first and second most common mutations detected in all these studies. However, the third and fourth most common genetic mutations were different. In the present study, the third and fourth most common genetic mutation were TPO and TSHR mutations, respectively. in addition, the third and fourth most common genetic mutations reported by Long et al (31) were TSHR and GNAS mutations, respectively and those by Sun et al (30) were TPO and DUOXA1 mutations, respectively. These discrepancies may be caused by the relatively small sample size, sampling criteria and/or different targeted genes determined.

Several limitations were observed in the present study. The sample size was relatively small, and most patients were too young to exhibit clinical phenotypes. Thus, determining the clinical significance of the detected mutations was difficult. Pedigree analysis was not performed in all the cases carrying mutations and evidence to support the pathogenicity of detected variants was insufficient. The diagnosis detection rate in the present study would be >30% were these requirements met. Finally, in vitro functional study of novel variants identified in the current study should be carried out.

In conclusion, using the currently largest targeted NGS panel containing 29 known genes, the mutation spectrum of 43 Han Chinese patients with CH was comprehensively determined. The main findings showed that DH other than TD is the common cause of CH in Chinese populations and genetic alterations associated with thyroid hormone biosynthesis, especially DUOX2 mutations, are the main genetic causes of CH. In addition, a high percentage of involvement of oligogenic mutation in the studied cases confirmed the potential role of oligogenicity or non-Mendelian inheritance in CH pathogenesis.