Discrepancies of RET gene and risk of differentiated thyroid carcinoma

Abstract

BACKGROUND:

Somatic variations in rearranged during transfection (RET) proto-oncogene acts to influence Thyroid cancer (TC) in a low penetrance manner, but their effects tend to vary between different populations.

OBJECTIVE:

This case-control study was aimed to evaluate effect of RET G691S, S904S and L769L single nucleotide polymorphisms (SNPs) on the risk for differentiated thyroid carcinoma (DTC).

METHODS:

A total of 180 patients and 220 controls were genotyped by Polymerase chain reaction – restriction fragment length polymorphism (PCR-RFLP). Di-Deoxy Sanger sequencing was performed on 100 samples with variations and 20 wild samples for each amplified exon. In addition, In Silico tools were used to evaluate structural and functional impact of individual SNPs in disease progression.

RESULTS:

In RET G691S/L769L/S904S SNPs, frequency of variant genotypes in DTC cases was 61.1%, 54.4% and 76.6% as compared to 45.9%, 43.6% and 89.09% in controls respectively ( $P⩽$ 0.05). In Silico analysis revealed that different protein formed due to G691S substitution decreases the stability of 3D structure of protein. The RET G691S and L769L SNP followed “Dominant” but RET S904S SNP confirmed an “Additive” mode of inheritance.

CONCLUSION:

RET G691S/L769L/S904S SNPs are significantly associated with DTC with G691S SNP declining the stability of final protein product.

1. Introduction

Thyroid cancer is the most common malignancy of endocrine system. Majority of endocrine carcinoma deaths are associated with this cancer type, representing 3–4% of all cancers [1]. Though TC is relatively curable, the incidence rate has increased during the past few decades with higher prevalence in females as compared to males [2, 3, 4, 5]. The rising trend is commonly attributed to DTC which includes Papillary thyroid carcinoma (PTC), Follicular thyroid carcinoma (FTC) and Hurthle cell carcinoma (HCC), accounting for 80%, 15% and 5% of DTC cases respectively [2, 3, 4]. PTC and FTC generally have good prognosis, especially, if diagnosed earlier. The probable environmental risk factors are physical influences such as lifestyle, uptake of abnormal iodine, chemical pollutants, unnecessary neck scans and exposure to ionizing radiations [5, 6, 7]. Various low to moderately penetrant genes have been characterized with the pathogenesis of TC.

The human rearranged during transfection (RET) proto-oncogene is positioned in chromosome 10q11.2. The gene encodes transmembrane receptor kinase (TRK) family of proteins that are associated with different diseases, especially cancer [12, 13, 14, 15]. Gain-of-function mutations in RET proto-oncogene results in aberrant activation of RET gene, involved in human carcinomas [8]. Loss-of-function mutation in RET is accountable for genetic disorder known as Hirschsprung disease [9], while as, germ line point mutations in RET gene are responsible for the occurrence of Medullary thyroid carcinoma (MTC) [9, 10]. Furthermore, somatic mutations in RET are present in sporadic MTC and rearrangements in RET/PTC are found in early childhood PTC [11, 12].

Several SNPs are present in RET gene but G691S, L767L and S904S present in exon 11, 13 and 15 respectively are extensively studied because these are located in exons where MTC associated germline mutations are concentrated [6]. These polymorphisms are recurrently associated with the increase risk of MTC and represented as low penetrant genetic modifiers [13]. Several studies have reported association of RET Polymorphic haplotypes with the risk of DTC [14, 15, 16].

Therefore, in current study we have comprehensively investigated the role of G691S, L767L and S904S SNPs in DTC risk and association of genotypes with socio-demographic and clinicopathological risk factors of DTC.

2. Materials and methods

2.1 Study design

This was a case-control study conducted by the Department of Biochemistry, Pakistan Institute of Medical Sciences (PIMS), Islamabad and Central Military Hospital (CMH) Muzaffarabad, Pakistan. A written informed consent was obtained from each study subject. The study was approved by Ethical Review Board (ERB) of PIMS. Written informed consent was taken from all patients involved in the study procedure and allowed the use of their blood samples.

2.2 Study participants

The study was conducted on 180 histologically confirmed DTC patients attending Department of General Surgery, PIMS and CMH, Muzaffarabad Pakistan. The patients having any other type of cancer, genetic disorder and having received chemotherapy were excluded from the study. Controls ( $n=$ 220) were randomly selected from a pool of healthy volunteers who visited the hospital for health check-up during the same period and enrolled in the study.

2.3 Sample collection and DNA extraction

Around 3 ml of blood was collected in EDTA coated vials from DTC patients and were stored at $-$ 80 ${}^{\circ }$ C until further processing. DNA was isolated from the DTC patients and healthy controls using DNA extraction kit (Invitrogen; Thermo Fisher Scientific, Inc., USA). The quality, purity and concentration of DNA was measured using Qubit 4.0 fluorometer (ThermoFischer Scientific, USA). In particular, purity of DNA was assessed by measuring optical density at A260/A280. Quality threshold used was; A260/A280 $⩾$ 1.7. Further quality of DNA was checked through agarose gel electrophoresis which gives the information weather the isolated DNA is intact or sheared.

2.4 Polymerase chain reaction (PCR)

Exon 11, 13 and 15 of RET gene were amplified using optimized PCR conditions and PCR primer sequences with considerable modifications, as described previously [17]. 50 $\mu$ l reaction mixture containing 50–100 ng genomic DNA; 1X PCR buffer (2 mM MgCl ${}_2$ ) (Biotools, B&M Labs, S.A. Madrid, Spain); 200 $\mu$ M dNTPs (Biotools, B&M Labs, S.A. Madrid, Spain); 1.5 Units of Taq polymerase (Thermo Fischer Scientific, USA); 0.25 $\mu$ M of forward and reverse primers (Eurofins Genomics, Germany) was setup for the amplification of respective exons resulting in the formation of 454 bp (exon 11 fragment), 460 bp (exon 13 fragment) and 332 bp (exon 15 fragment) amplicons.

2.5 Restriction fragment length polymorphism

Amplification fragments of exon 11, 13 and 15 were subjected to restriction digestion with BanI, TaqI and RsaI (Thermo Fischer Scientific, USA) for elucidation of genotypes for respective SNPs. G691S SNP (G2071A) was analysed by digesting the PCR product with BanI, resulting in the formation of 267 and 187 bp fragments (wild homozygote; GG), uncut 454 bp fragment (variant homozygote; AA), 454 bp, 267 bp and 187 bp fragments (heterozygous genotype; GA). 10 $\mu$ l of Digested DNA fragments were subjected to electrophoresis on a 2.0% agarose gel for resolution and then analysed under the UV illuminator. In case of L769L SNP (T2307G), the wild homozygous type (TT) gets digested with TaqI and characterized by 270 bp and 190 bp fragments, while variant homozygote (GG) is represented by an uncut 460 bp fragment and heterozygote form (TG) presented 460 bp, 270 bp and 190 bp fragments. 10 $\mu$ l of digested products were loaded into 2.0% agarose gel and then analysed under the UV illuminator. In case of S904S SNP (C2712G), PCR product was digested with RsaI restriction enzyme, resulting in one uncut fragment of 332 bp representing wild type (CC), variant homozygote (GG) characterized by 224 bp and 108 bp fragments while as heterozygote (CG) displaying 332 bp, 224 bp and 108 bp fragments. 10 $\mu$ l of digested products were loaded into 3.0% agarose gel and then analysed under the UV illuminator. Ethidium bromide was used as a nucleic acid stain in agarose gels which intercalates between the bases of DNA giving fluorescence with an orange colour, when exposed to ultraviolet light.

2.6 DNA sequencing

In order to re-confirm the PCR-RFLP results for G691S, L767L and S904S SNPs, a total of hundred ( $n=$ 100) DNA samples showing variations and twenty ( $n=$ 20) wild samples pertaining to each amplified exon were subjected to DNA sequencing based on the chain termination reaction, using the automated DNA sequencer ABI prism 310 (ThermoFisher Scientific, USA).

2.7 Dataset collection and computational prediction tools for functional impact of SNPs

A dataset of RET SNPs (G691S/L769L/S904S) was collected from NCBI database of SNPs (dbSNP) [18] for structural and functional analysis. In addition, protein sequence of RET (P07949) was retrieved from UniProt database [19]. Insilico methods of high accuracy were used for investigating the biological hypothesis. SNAP2 [20] and SNPnexus [21] were used to predict functional context of SNPs using data from external independent databases such as SIFT (Sorting Intolerant from Tolerant) [22] and Polyphen (Polymorphism Phenotyping) [23]. Score $⩽$ 0.05 was considered as damaging whereas score greater than 0.05 has been characterized as tolerant. On the other hand, Polyphen predicts the potential effect of amino acid substitution upon the structure and function of protein based on evolutionary and physical comparative considerations [24].

MU protein (mutant protein) online vector machine based server was used to predict protein stability of protein on single site amino acid mutation [25]. Furthermore, ModPred software [26] was used to predict post transcriptional modifications (PTM) in target protein sequences. HOPE is an automatic mutant analysis server used to determine structural effects of a mutation by calculating 3D coordinate information from wide range of web services including YASARA, WHAT IF software, UniProt database and DAS server [27].

2.8 Statistical analysis

Independent and paired $t$ -tests were performed for continuous variables while as fisher’s exact test or ${\chi }^2$ tests were applied on discrete variables [28]. Hardy-Weinberg equilibrium was performed using goodness of fit test, to compare statistical significant difference of allele frequency and genotype distribution between cases and controls [29]. Statistical difference in allele, genotype and haplotype frequencies between cases (DTC) and controls were compared by pearson- ${\chi }^2$ test and odds ratios (ORs) with 95% confidence intervals (CIs) using logistic regression analysis. Degree of association between genotypes and associated risk factors were obtained using ORs, CIs and $P$ -value. $P$ -value $<$ 0.05 was considered as statistically significant. All statistical tests were performed using SPSS 20.0 software [30, 31] (SPSS Inc., Chicago, Illinois, USA) and power of the study was calculated using G-Power (version 16.0) [32, 33]. Haplotype analyses was conducted by using HapMap software. Akaike information (AIC) was used to estimate the relevance of each genetic model and haplotype [34].

Table 1

Socio-demographic and clinicopathological parameters of DTC cases and controls

Character- istics	DTC cases $n=$ 180 (%)	Controls $n=$ 220 (%)	Chi square	$P$ value
Age group
$<$ 55 years	131 (72.8)	158 (71.8)	0.05	0.91
$⩾$ 55 years	49 (27.2)	62 (28.2)
Gender
Male	68 (37.7)	98 (44.5)	1.87	0.18
Female	112 (62.3)	122 (55.5)
Dwelling
Rural	120 (66.6)	138 (62.72)	0.67	0.46
Urban	60 (33.4)	82 (40.0)
Smoking
No	124 (68.8)	149 (67.7)	0.06	0.83
Yes	56 (31.2)	71 (32.3)
BTD
No	75 (41.7)
Yes	105 (58.3)
Histological type
PTC	151 (83.9)
FTC	29 (16.1)
TSH level
Normal	52 (28.9)
High	128 (71.1)
Grade
WDTC	176 (97.8)
PDTC	04 (2.2)
Focality
Unifocal	96 (53.3)
Multifocal	84 (46.7)
Vascular/capsular invasion
No	93 (51.7)
Yes	87 (48.3)
Lymph node metastasis
No	110 (61.1)
Yes	70 (38.9)
Stage $<$ 55 years
Stage I	65 (36.1)
Stage II	66 (36.7)
Stage $⩾$ 55 years
Stage I $+$ II	26 (14.4)
Stage III $+$	23 (12.8)
above

BTD: Benign thyroid disease; PTC: Papillary thyroid cancer; FTC: Follicular thyroid cancer, WDTC: Well differentiated thyroid carcinoma; PDTC: Poorly differentiated thyroid carcinoma.

3. Results

3.1 Patient characteristics

The socio-demographic and clinicopathological parameters of DTC cases ( $n=$ 180) and controls ( $n=$ 220) are given in Table 1. The cases and controls were matched with respect to age, gender, dwelling and smoking status ( $P>$ 0.05). The mean age (in years) of cases and controls was 47.4 $\pm$ 13 and 46.1 $\pm$ 15.8 respectively ( $P>$ 0.05). In this study, females were more affected than males (62.3% vs. 37.7%) and people of rural areas were more predisposed to DTC than those living in urban areas (66.6% vs. 33.4%). Benign thyroid disease (BTD) was found in 58.3% of DTC patients (35.7%; multinodular goitre, 12.6%; thyroiditis and 10%; thyroid adenomas). 83.9% of enrolled patients were having PTC and 16.1% were having FTC (Table 1).

Figure 1.

Restriction fragment length polymorphism pattern of RET G691S (G2071A) using BanI restriction enzyme. Lane M: 100 bp ladder. Lane 4 and 9 represent the homozygous wild (G/G) genotype (size 267 bp $\mathrm{+}$ 187 bp). Lane 2, 3 and 5 represent homozygous variant (A/A) genotype (size 454 bp). Lane 1, 6, 7 and 8 represent G/A heterozygous genotype (size 454,267 and 187 bp).

Figure 2.

Restriction fragment length polymorphism pattern of RET L769L (T2307G) using TaqI restriction enzyme. Lane M: 100 bp ladder. Lane 2, 3, 5 and 6 represent the homozygous wild (T/T) genotype (size270 bp $\mathrm{+}$ 190 bp). Lane 4 represent homozygous variant (G/G) genotype (size 460 bp). Lane 1 represent T/G heterozygous genotype (size 460, 270 and 190 bp).

Figure 3.

Restriction fragment length polymorphism pattern of RET S904S (C2712G) using RsaI restriction enzyme. Lane M: 100 bp ladder. Lane 3, 4, 5 and 6 represent the homozygous wild (C/C) genotype (size 332 bp). Lane 1 represent homozygous variant (G/G) genotype (size 224 bp $\mathrm{+}$ 108 bp). Lane 2 and 7 represent C/G heterozygous genotype (size 332, 224 and 108 bp).

Figure 4.

Partial electrophoretograms of Exon 11 fragment of RET gene for elucidation of G691S (G2071A) genotypes (A) Arrow shows homozygous wild (G/G) genotype at nucleotide position 2071 (B) Arrow shows heterozygous (G/A) genotype at nucleotide position 2071 and (C) Arrow shows homozygous variant (A/A) genotype at nucleotide position 2071.

Figure 5.

Partial electrophoretograms of Exon 13 fragment of RET gene for elucidation of L769L (T2307G) genotypes (A) Arrow shows homozygous wild (T/T) genotype at nucleotide position 2307 (B) Arrow shows heterozygous (T/G) genotype at nucleotide position 2307 and (C) Arrow shows homozygous variant (G/G) genotype at nucleotide position 2307.

Table 2

Genotypic and allelic frequency of DTC cases and controls in respective SNPs

Genotype/ allele	Cases ( $n=$ 180)	Controls ( $n=$ 220)	OR (95%CI)	$P$ value
G691S (G2071A)
Genotype
GG	69 (38.3)	119 (54.1)	1.0 (Ref)
GA	81 (45.0)	77 (35.0)	1.8 (1.18–2.8)	0.006
AA	30 (16.7)	24 (10.9)	2.1 (1.17–3.98)	0.01
GA $+$ AA	111 (61.7)	101 (45.9)	1.9 (1.2–2.8)	0.001
Allele type
$G$	219 (60.8)	315 (71.5)	1.0 (Ref)
$A$	141 (39.1)	125 (28.5)	1.6 (1.20–2.2)	0.0008
L769L (T2307G)
Genotype
TT	52 (28.8)	24 (10.9)	1.0 (Ref)
TG	90 (50.0)	134 (60.9)	0.3 (0.2–0.5)	0.00001
GG	38 (21.2)	62 (28.2)	0.28 (0.15–0.53)	0.00005
TG $+$ GG	128 (71.1)	196 (89.0)	0.3 (0.1–0.5)	0.000004
Allele type
$T$	194 (53.8)	182 (41.4)	1.0 (Ref)
$G$	166 (46.1)	258 (58.6)	0.6 (0.4–0.8)	0.0002
S904S (C2712G)
Genotype
CC	79 (43.8)	124 (56.4)	1.0 (Ref)
CG	83 (46.1)	84 (38.2)	1.5 (1.02–2.3)	0.02
GG	18 (10.1)	12 (5.4)	2.3 (1.07–5.15)	0.03
CG $+$ GG	101 (56.2)	96 (43.6)	1.6 (1.1–2.4)	0.01
Allele type
$C$	241 (66.9)	332 (75.5)	1.0 (Ref)
$G$	119 (33.05)	108 (24.5)	1.5 (1.1–2.06)	0.005

Figure 6.

Partial electrophoretograms of Exon 13 fragment of RET gene for elucidation of S904S (C2712G) genotype (A) Arrow shows homozygous wild (C/C) genotype at nucleotide position 2712 (B) Arrow shows heterozygous (C/G) genotype at nucleotide position 2712 and (C) Arrow shows homozygous variant (G/G) genotype at nucleotide position 2712.

3.2 Elucidation of RET genotypes and alleles

The representative pictures of restriction fragment pattern for RET G691S, L767L and S904S SNPs are shown in Figs 1–3 respectively. The Figs 4–6 show partial electropherograms pertaining to homozygous wild, heterozygous and homozygous variant genotypes of RET G691S, L767L and S904S SNPs respectively. Genotype and allele distributions of RET polymorphisms in DTC cases and controls are summarized in Table 2. The three SNPs have significant differences in genotype and allele frequencies between cases and controls. In case of G691S SNP, frequency of variant genotype (GA $+$ AA) in cases was greater as compared to controls (61.6% vs. 45.9%; OR $=$ 1.9; $P=$ 0.001). In case of L769L SNP, overall variant genotype distribution (TG $+$ GG) was lower in cases as compared to control group (71.1% vs. 89.0%; OR $=$ 0.3; $P=$ 0.000004). In case of S904S SNP, distribution of variant genotype (CG $+$ GG) was higher in cases against controls (56.2% vs. 43.6%; OR $=$ 1.6, $P=$ 0.01) (Table 2).

Table 3

Analysis of genotype frequencies of SNPs associated with clinicopathological factors

	Cases	GG	GA $+$ AA	Controls	GG	GA $+$ AA	OR (95%CI)	$P$ value
G691S (G2071A)	$n$ (%)	$n$ (%)	$n$ (%)	$n$ (%)	$n$ (%)	$n$ (%)
Overall genotype	180	69 (38.3)	111 (61.7)	220	119 (54.1)	101 (45.9)	1.9 (1.3–2.8)	0.001
Age group
$<$ 55	131 (72.8)	47 (35.9)	84 (64.1)	158 (71.8)	85 (53.8)	73 (46.2)	2.1 (1.2–3.3)	0.002
$⩾$ 55	49 (27.2)	23 (46.9)	26 (53.1)	62 (28.2)	34 (54.8)	28 (45.2)	1.3 (0.64–2.9)	0.40
Gender
Male	68 (37.7)	32 (47.1)	36 (52.9)	98 (44.5)	52 (53.1)	46 (46.9)	1.3 (0.7–2.4)	0.53
Female	112 (62.3)	44 (39.3)	68 (60.7)	122 (55.5)	67 (54.9)	55 (45.1)	1.9 (1.1–3.2)	0.01
Residence
Rural	120 (66.6)	45 (37.5)	75 (62.5)	138 (62.7)	80 (57.9)	58 (42.08)	2.3 (1.4–3.8)	0.001
Urban	60 (33.4)	24 (40.0)	36 (60.0)	82 (37.3)	39 (47.5)	43 (52.4)	1.4 (0.7–2.67)	0.40
Smoking status
No	124 (68.8)	48 (38.7)	76 (61.3)	149 (67.7)	81 (54.4)	68 (45.6)	1.8 (1.1–3.0)	0.01
Yes	56 (31.2)	21 (38.0)	35 (62.0)	71 (32.3)	38 (53.5)	33 (46.5)	1.9 (0.9–3.9)	0.07
L769L (T2307G)
Overall genotype	180	52 (28.8)	128 (71.2)	220	24 (10.9)	196 (89.1)	0.3 (0.2–0.5)	0.00
Age group
$<$ 55	131 (72.8)	44 (33.5)	87 (66.4)	158 (71.8)	19 (12.0)	139 (88.0)	0.3 (0.15–0.49)	0.001
$⩾$ 55	49 (27.2)	08 (16.3)	41 (83.7)	62 (28.2)	05 (8.1)	57 (91.9)	0.4 (0.13–1.4)	0.20
Gender
Male	68 (37.7)	14 (20.5)	54 (79.4)	98 (44.5)	11 (11.2)	87 (88.8)	0.5 (0.2–1.1)	0.10
Female	112 (62.3)	38 (33.9)	74 (66.07)	122 (55.5)	13 (10.7)	109 (89.3)	0.2 (0.1–0.5)	0.0001
Residence
Rural	120 (66.6)	42 (35.0)	78 (65.0)	138 (62.7)	15 (10.86)	123 (89.1)	0.2 (0.1–0.43)	0.001
Urban	60 (33.4)	10 (16.6)	50 (83.3)	82 (37.3)	09 (10.9)	73 (89.1)	0.6 (0.2–1.6)	0.45
Smoking status
No	124 (68.8)	45 (36.2)	79 (63.7)	149 (67.7)	17 (11.4)	132 (8.6)	0.2 (0.1–0.4)	0.00001
Yes	56 (31.2)	07 (12.5)	49 (87.5)	71 (32.3)	07 (9.8)	64 (90.2)	0.8 (0.25–2.3)	0.70
S904S (C2712G)
Overall genotype	180	79 (43.8)	101 (56.1)	220	124 (56.3)	96 (43.6)	1.65 (1.1–2.45)	0.02
Age group
$<$ 55	131 (72.8)	57 (43.5)	74 (56.4)	158 (71.8)	88 (55.7)	70 (44.3)	1.6 (1.02–2.6)	0.04
$⩾$ 55	49 (27.2)	22 (44.9)	27 (55.1)	62 (28.2)	36 (58.1)	26 (41.9)	1.7 (0.8–3.6)	0.20
Gender
Male	68 (37.7)	36 (52.9)	32 (47.1)	98 (44.5)	58 (59.1)	40 (40.8)	1.2 (0.69–2.4)	0.40
Female	112 (62.3)	43 (38.4)	69 (61.6)	122 (55.5)	66 (54.1)	56 (45.9)	1.7 (1.01–2.83)	0.045
Residence
Rural	120 (66.6)	52 (43.3)	68 (56.6)	138 (62.7)	78 (56.5)	60 (43.5)	1.7 (1.03–2.78)	0.04
Urban	60 (33.4)	27 (45.0)	33 (55.0)	82 (37.3)	46 (56.1)	36 (43.9)	1.5 (0.8–3.05)	0.20
Smoking status
No	124 (68.8)	51 (41.1)	73 (58.8)	149 (67.7)	84 (56.3)	65 (43.7)	1.8 (1.1–3.0)	0.01
Yes	56 (31.2)	28 (50.0)	28 (50.0)	71 (32.3)	40 (56.3)	31 (43.7)	1.3 (0.6–2.6)	0.60

BTD: Benign thyroid disease; PTC: Papillary thyroid cancer; FTC: Follicular thyroid cancer, WDTC: Well differentiated thyroid carcinoma; PDTC: Poorly differentiated thyroid carcinoma.

3.3 Stratification analysis of RET polymorphisms and risk of DTC

We stratified all three SNPs with respect to various socio-demographic and clinicopathological parameters of DTC cases and controls. Table 3 summarizes the parameters who were found to be significantly associated with variant genotype of studied RET SNPs. In case of G691S SNP, frequency of variant genotypes (GA $+$ AA; 64.1%) was significantly higher in young patients ( $<$ 55 years) than controls (46.2%) of same age group ( $P=$ 0.02). There was a significant high distribution of variant genotypes (GA $+$ AA; 60.7%) in female cases in contrast to female healthy controls (45.1%) ( $P=$ 0.01). Similarly non-smoker cases and cases living in rural areas had significantly higher frequency of combined GA and AA variant genotypes as compared to smoker controls ( $P=$ 0.00001) and controls living in rural areas ( $P=$ 0.001) respectively (Table 3).

Table 4

Insilico analysis of RET SNPs

Variation ID	Ch; location	NT change	AA position	SIFT		Polyphen		ClinVar
				Score	Prediction	Score	Prediction	Clinical significance	Phenotype
rs1799939 (G691S)	10; 436114671	G/A	691	0.940	Tolerated	0.015	Benign	Conflicting interpretation of pathogenecity	MTC, MEN2A, PTC, Hirschsprung disease
rs1800861 (L769L)	10; 43118395	T $>$ G	769	n/a	–	n/a	–	–	–
rs1800863 (S904S)	10: 43120185	C $>$ G	904	n/a	–	n/a	–	–	–

Table 5

Applicable inheritance models for RET SNPs

SNP	Genotypes	DTC cases	Controls	OR (95%CI)	$P$ value
	(Patients vs controls)	( $n=$ 140)	( $n=$ 180)
	(Dominant model)
G2071A	GG	69 (38.3)	119 (54.1)	1.9 (1.27–2.8)	0.001
(G691S)	GA $+$ AA	111 (61.6)	101 (45.9)
	(Dominant model)
T2307G	TT	52 (28.8)	24 (10.9)	0.3 (0.2–0.5)	0.0001
(L769L)	TG $+$ GG	128 (71.1)	196 (89.1)
	(Additive model)
C2712G	CC	79 (43.8)	124 (68.8)	2.3 (1.1–5.1)	0.03
(S904S)	GG	18 (10.1)	12 (5.4)

In case of L769L SNP, significant low frequency of variant genotype (TG $+$ GG) was found in cases with $<$ 55 years of age ( $P=$ 0.001), female gender ( $P=$ 0.0001), rural dwelling ( $P=$ 0.001) and non-smokers ( $P=$ 0.00001) as compared to their respective controls (Table 3).

In case of S904S SNP, frequency of combined CG and GG variant genotype was significantly high in cases having $<$ 55 years of age as compared to respective controls (56.4% vs. 44.3%; $P=$ 0.04). Similarly, variant genotype (CG $+$ GG) distribution was high in patients having female gender than controls of same gender (61.6% vs. 45.9%; $P=$ 0.045). Also, DTC patients of rural areas had higher frequency variant allele (CG $+$ GG) frequency than control group (56.6% vs. 43.5%; $P=$ 0.04). Additionally, patients who never smoked were found to have high frequency of variant genotype as compared to controls (58.8% vs. 53.7%; $P=$ 0.01). No statistical significant association was found between other clinical risk factors and genotype of all three SNPs.

3.4 In silico analysis of RET SNPs

Table 4 depicts the In Silico results for RET SNPs. Through In Silico analysis functional impact of non-synonymous variant G691S was analysed using different tools. In SNPnexus software, REACTOME database predicted G691S polymorphism as significantly associated with RET signalling, RAF/MAP kinase cascade and MAPK1/MAPK3 signalling pathway. While cancer genome interpreter predict that RET (RET_TPCN1), a biomarker of G691S act as a target against cabozantinib, vandetanib, Sunitinib and RET inhibitors (Pan-TK inhibitor/Pan-kinase inhibitor). MuPro determined that G691S polymorphism decreases the stability of protein by $-$ 0.825 and $-$ 1.14 kcal/mol respectively [35]. Additionally, ModPred showed G691S polymorphism influence low level of proteolytic cleavage in protein sequence. According to HOPE project no protein structure with G691S polymorphism was found in database, because SNP is located in loop region [27]. In the G691S polymorphism, the size of polymorphic amino acid increases whereas charge, polarity and hydrophobicity of 3D structure remain unchanged. Glycine, the only residue has the ability to maintain the torsion angles of 3D structure. When G was substituted with S, disruption of torsion angles could result in the conformational change and hence disturbed 3D structure of protein. Polymorphic residue has been found among homologous sequences as well, but probably not damaging to proteins. We found that synonymous L769L and S904S variations have no direct effect on protein structure and function.

3.5 Genetic association study and inheritance models

Different models of inheritance were found appropriate to RET G691S, L769L and S904S polymorphism. Dominant model of inheritance was found suitable for the inheritance of RET G691S and L769L polymorphism while as Additive model of inheritance was appropriate for RET S904S polymorphism (Table 5).

Table 6

Haplotype frequencies estimation and haplotype association with disease (adjusted by gender $+$ age $+$ smoking status $+$ dwelling)

G2071A	T2307G	C2712G	Total frequency	Cases	Controls	OR (95% CI)	$P$ value
G	T	C	0.5986	0.5820	0.6152	Ref. (1.00)	–
G	G	C	0.3200	0.2880	0.3252	1.38 (0.22–0.67)	0.2
G	T	G	0.2016	0.1896	0.2136	1.1 (0.22–0.61)	0.002
A	T	C	0.0843	0.0555	0.1131	1.69 (0.84–3.38)	0.14
A	G	C	0.0572	0.0695	0.0449	1.23 (0.07–0.69)	0.0094
A	T	G	0.0459	0.0220	0.0698	9.03 (2.14–38.13)	0.003

Global haplotype association. $P$ value: $<$ 0.0001. Haplotypes having $<$ 1% frequency were considered rare and excluded. All haplotypes were included in the same model and the OR (95%CI) of DTC associated with each haplotype estimated by comparison with the common reference haplotype. ref: reference group.

3.6 Haplotype frequencies estimation and its association with disease

Haplotype analysis was conducted to assess the combined effect of the RET G691S, L769L and S904S polymorphisms on DTC risk. Both the cases and controls had haplotypes frequencies of $>$ 5%. G T C was the most common haplotype, having frequencies of 58.2% and 61.5% in cases and controls respectively. There was a marked difference between overall distribution of G T G, A G C and A T G haplotype in cases vs. controls ( $P<$ 0.0001). Haplotype frequencies were adjusted by gender, age, smoking status and dwelling. Table 6 reveals the haplotype pattern for three SNPs among patients and controls.

4. Discussion

In addition to environmental and lifestyle factors, certain SNPs in RET gene influence TC in complex, polygenic and low penetrance manner. RET polymorphisms have strong correlation with disease phenotypes and its incidence was detected upto 60% of sporadic PTC [36]. Up to this time, only limited number of studies analysed RET SNPs in DTC that too with conflicting results. We have firstly conducted this case-control study to find out the frequency of RET genotypes; isolated and synergistic effect of RET variants on socio-demographic and clinicopathological risk factor of DTC patients and controls of Pakistan.

In case of RET G691S SNP, the overall frequency of variant genotype (GA $+$ AA) was significantly more frequent in cases as compared to controls (61.7% vs. 45.9%; $P=$ 0.001). The confidence on results has been augmented when earlier studies from different population were in line with genotype and allele frequencies found in this study [15]. Similarly, the study conducted by Huang et al. indicated the influence of RET G691S variant genotype (AA) in the predisposition of DTC [37]. Our study was contradictory to previous reports in which RET G691S genotype frequency was same in both cases and controls [38]. Furthermore, in our study, RET G691S variant genotypes (GA $+$ AA) were found to be significantly associated with DTC patients of $<$ 55 years of age, females, rural dwellers and patients with no history of smoking. Khan and his co-workers observed that variant genotypes (GA $+$ AA) were significantly associated with females and patients without smoking history [15]. Interestingly some previous studies have proposed that cigarette smoking might lower TSH (thyroid stimulating hormone) levels, thus preventing thyroid carcinogenesis [39]. In contrary, there are some studies which have not supported the influence of smoking on TSH levels [40]. For further classification, high distribution of variant alleles (GA $+$ AA) was found in patients of age $<$ 55 years as equated to healthy controls which is in contrast to the study of Mudassar et al. [41], in which RET G691S polymorphism was not associated with patients of lower age group ( $<$ 55 years). Furthermore, RET G691S polymorphism was associated with patients dwelling in rural areas as compared to respective healthy controls, which is in discordance with earlier studies [15].

In case of RET L769L SNP, we observed a significantly higher frequency of variant genotypes (TG $+$ GG) in DTC cases compared to controls (71.0% vs. 89.0%; $P=$ 0.000004) indicating a protective role of RET L769L SNP in predisposition to DTC. Our study corresponds to the previous reported studies of different ethnic groups such as from European/Australian and North American population which reported association of RET L769L polymorphism with TC [42, 43]. Our results indicate that variant genotypes in L769L SNP were associated with non-smokers which was in agreement with a previous study on DTC [41]. For further stratification in groups, our study found low frequency of variant alleles (TG $+$ GG) in patients of age $<$ 55 years, females and dwelling in rural areas as compared to healthy controls which is in line with the study performed on a different ethnic group (North Indian population) [15].

In the case of RET S904S SNP, we observed over representation of variant genotype (CG $+$ GG) in DTC with significant difference between DTC cases and controls (56.2% vs. 43.6%; $P=$ 0.01). This study was supported by a recent study in which polymorphic allele (G) was more frequent in cases as compared to controls [41], but in contrast to a previous study which reported low frequency of polymorphic allele (G) in cases as compared to controls [16]. Our study was in contradiction to the study by Santos et al. wherein they have found no difference in the distribution of genotype and allele frequency between cases and controls [17]. Furthermore, variant genotype was significantly more frequent in DTC patients having female gender and lower age group ( $<$ 55 years). Santos et al. reported no significant difference of variant genotypes (CG $+$ GG) between cases and controls with respect to gender and age [17]. Khan et al. reported that DTC patients with male gender and higher age group ( $⩾$ 45 years) were more susceptible to DTC risk [35, 15]. Furthermore, overrepresentation of variant genotype (CG $+$ GG) was observed in patients with no smoking status and living in rural areas as compared to matched healthy controls. Khan et al. also observed high frequency of S904S SNP in patients with no smoking history but did not find any association of this SNP with patient’s residence [15].

RET G691S SNP involves a change of glycine to serine and has understated effects on its function. RET G691S alter the 3D structure by disrupting torsion angles and flexibility of protein. The new polymorphic protein decreases the stability of 3D structure, and drug resistance occurs due to sub-localization of tyrosine kinase receptor. This is in coherence with previously reported studies [44]. RET L769L and S904S SNPS are synonymous/silent. Although not clearly understood, the silent polymorphisms in the RET gene may affect the development and progression of DTC by effecting the RNA stability thereby indirectly effecting the final protein synthesis [15]. In addition, the fact that RET L769L and S904S SNPs are present in linkage disequilibrium with other unknown functional SNPs, cannot be ignored [45] suggested that these silent polymorphisms have a non-neutral effect on RET function thus playing a key role in thyroid carcinogenesis.

This study evaluated association of multiple RET risk alleles in DTC development and aggressiveness, indicating dominant and additive effect of different RET polymorphisms in the pathogenesis of DTC. We considered Genomic association studies (GAS) with significant association only. In RET G691S SNP, heterozygous genotype (GA) lies towards homozygous mutant genotype (AA) conferring the deleterious role to AA genotype. In case of RET L769L SNP, heterozygous genotype (TG) lies towards homozygous mutant genotype (GG) conferring the deleterious role to GG genotype [46, 47]. In both G691S and L769L SNPs, single allele seems to modulate the disease susceptibility thus following “Dominant mode of inheritance” [46]. The findings were in line with previous studies [44]. In case of RET S904S SNP, the heterozygote (CG) lies in the middle of the two homozygotes, with homozygous variant (GG) genotype having the maximum susceptibility of being diseased and homozygous wild (CC) having the least thus supporting non-dominance (i.e., perfect additivity) [47]. Furthermore additive model selected for RET S904S was supported by a study of Khan et al. [15]. Our results showed that G T G, A G C and A T G haplotypes are associated with an increased risk of DTC. These haplotype includes variant alleles of G691S and S904S SNP that are associated with the increased risk of DTC in the single locus analysis. The variant alleles of RET L769L seem to have protective role towards thyroid carcinogenesis. The two minor alleles in combination may amplify the influence of single SNPs and may produce a synergistic effect, the phenomenon shown in other tumors [48]. Therefore, S904S/G691S in combination may amplify the risk of aggressiveness of the disease which is in line with the study of Santos et al. [17] and Khan et al. [15].

5. Conclusion

In conclusion, current data suggests that the RET G691S and S904S SNPs significantly increase the risk of DTC while as RET L769L SNP has a protective role towards predisposition to DTC. Specific haplotypes of RET such as G T G, A G C and A T G are risk factors for DTC. Furthermore, these haplotypes are over represented, and may act as low penetrant alleles in the predisposition of disease. In future, more studies with large sample size are required to authenticate the role of RET polymorphisms in DTC.