**Table 4.**

*Other significant or suggestive susceptibility loci highlighted by genome-wide association studies on DTC (ordered by significance of association).*

*Genetic Susceptibility to Differentiated Thyroid Cancer DOI: http://dx.doi.org/10.5772/intechopen.107831*

#### *3.1.1 Locus 9p22.33*

At 9p22.33, the most robust association reported in GWAS was for rs96551 or rs1588635, a highly correlated proxy (r2 = 0.99 in the European population from the 1000 Genomes Project). Rs965513 has been associated with radiation-related DTC as well as with sporadic DTC, and it was subsequently consistently replicated in several different populations of European ancestry [75–82], of Asian ancestry [81, 83–85] as well as in admixed populations from Oceania [79, 86], Cuba [87], Colombia [88], or Kazakhstan [89].

In 2009, a study focusing on the role of *FOXE1* in DTC risk [90] suggested rs1867277 as a causal variant at 9p22.33. This variant is located in the 5' UTR of *FOXE1* (also known as *TTF2,* for *Thyroid Transcription Factor 2*) and was shown to affect the transcription of *FOXE1* which is involved in the development and regulation of the thyroid gland, and in the proliferation and differentiation of thyroid follicular cells. However, rs1867277 is less consistently replicated in other populations and the linkage disequilibrium (LD) between rs1867277 and rs965513 is moderate in Europeans (r2 = 0.3 in Europeans from the 1000 Genomes project) and even weaker in populations of Asian or African ancestries (r2 <0.01 in the 1,000 Genomes Project). Rs965513 is located 60 kb upstream of rs1867277, in an intron of the long intergenic noncoding RNA (lincRNA) *PTCSC2* that was reported for the first time in 2015 by He et al. [91]. The risk allele [A] of rs965513 was shown to significantly decrease the expression of unspliced *PTCSC2*, FOXE1, and TSHR in normal thyroid tissue.

#### *3.1.2 Locus 14q13.3*

Rs944289 was the first variant identified by GWAS at 14q13.3 [68]. This association was replicated in subsequent studies conducted in diverse populations [75, 79, 80, 82–84, 87–89, 92–94]. However, the most recent GWAS, conducted in several European populations [72, 73] and in one Asian population from Korea [74], analyzed a higher number of SNPs (using genotyped and imputed SNPs) and reported the strongest signal for respectively rs368187 and rs34081947 (r2 = 0.98 in Europeans and r2 = 0.90 in East Asians from the 1000 Genomes Project), which are in moderate LD with rs944289 (r2 <0.70 in Europeans or East Asians from the 1000 Genomes Project). The recent European GWAS also reported an independent signal at rs116909374 and rs368187. Rs116909374 replicated only in studies on European ancestry populations [79, 82, 88, 93] as this SNP is monomorphic or very rare in Asian populations. Interestingly, rs368187 is in high LD with rs34081947, which was highlighted by the GWAS conducted in the Korean population (r2 = 0.98 in East Asians from the 1000 Genome Project) (**Table 3**).

In 2012, Jendrzejewski et al. [95] described a novel lincRNA named *PTCSC3* located 3.2 kb downstream of rs944289. They showed that the expression of *PTCSC3* was strongly down-regulated in thyroid tumor tissue and that the risk allele [T] of rs944289 was associated with up-regulation of *PTCSC3* in normal thyroid tissue, suggesting that *PTCSC3* could act as a tumor suppressor gene. Most recent fine-mapping analyses [79, 93] confirmed that multiple independent SNPs are involved in DTC risk at 14q13.3, but the clinical significance of all these SNPs is still unknown.

#### *3.1.3 Locus 2q35*

The association between DTC and variants at 2q35 was first highlighted in 2009 in a GWAS that investigated genetic factors associated with thyroid stimulating hormone (TSH) levels in blood in 27,758 Icelandic individuals. The authors further investigated the role of the top SNPs associated with circulating TSH levels in DTC susceptibility of which rs966423 at 2q35 [82]. Another GWAS on DTC risk, conducted in 2013 in an Italian population [70], reported rs6759952 as the lead SNP at 2q35, which is moderately correlated to rs966423 (r2 = 0.69 in Europeans from the 1000 Genomes project). The most recent GWAS from 2017 [72–74] reported three new SNPs (rs11693806, rs3821098, and rs12990503) with the strongest signal at 2q35, which are highly correlated to each other but only moderately with the two previous reported SNPs (**Table 2**).

Finally, the recent *in silico* fine-mapping analysis at 2q35 conducted in a multiethnic study pinpointed rs16857609 as a possible causal SNP [96]. This SNP was strongly correlated to the three SNPs reported by recent GWAS and was associated with the expression of the two nearby genes *DIRC3* and *IGFBP5* in thyroid tumor cells. Interestingly, this SNP had also been previously associated with breast cancer risk [97].

#### *3.1.4 Locus 8p12*

At 8p12, rs2439302 was first highlighted in the Icelandic GWAS on TSH levels, and it was found to be associated with DTC risk in subsequent analyses [82]. The association between DTC and rs2439302 was then replicated in different populations [84, 88, 89, 92, 94, 98]. The recent GWAS [72–74] replicated this SNP and also reported several other variants within the gene *NRG1* that were independently associated with DTC in European and Asian ancestry populations (rs142450470, rs6996585, and rs12542743) (**Table 2**).

In 2018, He et al. [99] reported that the risk allele [G] of rs2439302 was associated with the expression of multiple NGR1 isoforms in normal thyroid tissue. They also suggested that multiple enhancer variants exist at this locus that may have a combinatory effect on the expression of *NRG1* and possibly on the susceptibility to DTC.

#### *3.1.5 Other loci*

Among the other loci reported by GWAS, only SNPs at four loci (1q42.2, 5p15.33, 1p13.3, 6q14.1) were replicated in independent studies on DTC (**Table 4**). Interestingly, some of the variants reported in **Table 4** were also previously associated with other diseases or traits. For instance, the SNPs at 5p15.33 (*TERT*) were shown to be associated with telomere length in European and Asian populations [100–102] as well as with risk of breast or ovarian cancers [100, 103]. Rs7902587 at 10q24.33 (*OBFC1*) was significantly associated with ovarian cancer [104], rs56062135 (*SMAD*) at 15q22.33 was also highlighted by a GWAS on coronary artery disease [105], the missense variant rs6793295 (*LRRC34*) at 3q26.2 was associated to systemic sclerosis [106], and rs7248104 (*INSR*) at 19p13.2 was associated to triglyceride levels [107].

#### **3.2 Polygenic risk scores**

Based on findings from GWAS, polygenic risk scores (PRS), which are calculated by computing the sum of risk alleles of identified susceptibility SNPs weighted by the effect size estimate from the GWAS, were proposed to predict DTC risk. Several studies evaluated a DTC PRS in different populations [114–118]. The most recent studies used PRS including 10 to 12 SNPs reported by the meta-analysis of GWAS [72] and estimated odds ratios per standard deviation of PRS from 1.55 to 2.31. Liyanarachchi et al. [118] estimated that about 8% of the genetic predisposition to PTC could be

accounted for by 10 SNPs (rs12129938, rs11693806, rs6793295, rs73227498, rs2466076, rs1588635, rs7902587, rs368187, rs116909374, rs2289261). They also estimated that individuals of European ancestry in the highest decile of PRS had a 6.9 higher risk to develop PTC than individuals in the lowest decile group. A recent study reported that the PRS improved significantly predictive scores based on clinical factors in the prediction of subsequent thyroid cancers in childhood cancer survivors of European ancestry [119]. Future studies should investigate the combined effect of PRS and exposure to lifestyle and environmental factors in order to enhance individualized DTC risk prediction. There is also a need to extend the DTC PRS to other ethnic groups.
