The first microarray-based study published by Huang et al. in 2001 (Huang et al. 2001), showed very distinct difference in gene expression profile between papillary thyroid carcinoma (PTC) and normal thyroid tissue. They used microarray approach with 12,000 genes to profile eight PTCs and matching normal thyroid tissues. Fifty genes with the most distinct gene expression changes were specified. Genes of a number of categories, including thyroid metabolism (e.g. TPO (thyroid peroxidase); SLC5A5 (NIS, sodium iodide symporter), tumour suppressor genes (e.g. apoptosis regulator Bcl–2), cell adhesion (e.g. FGL–2, fibrinogen-like 2) and signal transduction (e.g. STC–1, stanniocalcin 1) had decreased expression. Decreased expression of genes specifically related to thyroid function (e.g TPO or NIS), proved that PTC, while being well-differentiated thyroid cancer, was less differentiated than normal thyroid. The list of 24 overexpressed genes included some genes already known for differential expression in PTC from single gene studies, e.g.; FN1 (fibronectin 1), MET (met proto-oncogene), or LGALS3 (galectin 3). However, in addition to already known ones, many new genes not previously associated with PTC were identified, such as CITED1 (Cbp/p300-interacting transactivator, with Glu/Asp-rich carboxy-terminal domain 1, also known as melanocyte-specific gene 1), involved in pigmented melanoma cells, which was initially believed to be melanoma specific. This later on gave an input to validate CITED1 as a marker for differential diagnosis of malignant from benign thyroid tumours (Prasad et al. 2004; Arora et al. 2009). Differences in gene expression profile between papillary PTC and normal thyroid tissue appeared very distinct and stable and were confirmed and extended by other authors (Giordano et al. 2005; Jarzab et al. 2005; Fontaine et al. 2008). As shown by a meta-analyses of PTC studies [(Griffith et al. 2006), Oczko-Wojciechowska, manuscript in preparation], the similarity between analyses was closer for studies using similar microarray platforms, but general tendencies were strong enough to be seen on various platforms. Considering gene ontology classes, genes related to cell adhesion, immune response, response to wounding and development controlling contributed the most to observed differences in gene expression. As we observed in our approach to PTC gene expression profile (Jarzab et al. 2005), a very characteristic feature of PTC was the huge prevalence of over-expressed genes, when compared with normal thyroid, while in majority of other cancers under-expressed genes, lost in cancer, prevailed.

The distinct feature of papillary thyroid cancer, indicated already by Huang et al. (Huang et al. 2001), was an over-expression of cell adhesion genes, responsible for migration and invasion of cancer cells. A large set of cell adhesion genes, mainly over-expressed, differed all PTCs not only from normal thyroids but also from follicular tumours as depicted in Fig. 3 (Handkiewicz-Junak et al. 2010).

Some studies have combined morphologic or genetic subclassification with gene expression profiling data to define expression profiles that correlate with specific PTC subtypes or mutations. Using a cohort of 51 PTC tumours Giordano et al. (Giordano et al. 2005) found a good correlation between gene expression profile and morphology as there was a good separation between follicular and classical variant of PTC, however, the tall cell variant loosely grouped among classical type without clear separation. They also showed that mutational status was more highly correlated to gene expression than morphology, confirming the previous observation (Fagin 2004) of BRAF and RET/PTC mutations as dominant alternative molecular events in PTC. The distinct differences in gene profile between PTC without BRAF mutations or RET/PTC rearrangements when compared to PTC with this mutations, were confirmed by others (Durand et al. 2009). However, similarities were also obvious: it was shown that BRAF mutations and RET/PTC rearrangements resulted in similar gene expression profile and led to decreased expression of genes related to thyroid differentiation (e.g. TG, TPO, NIS, DUOX1, DUOX2), while cancers that do not harbour BRAF or RET/PTC mutations retained differentiation profile similar to normal thyroid.

Gene expression profiles of FTC were analyzed in several microarray studies, with the aim to improve the molecular differentiation between FTC and follicular adenomas and to elucidate the molecular etiology of FTC (Giordano et al. 2005; Cerutti et al. 2004; Lacroix et al. 2005). Different sets of genes characteristic for FTC had been established. Weber et al. (Weber et al. 2005) identified the combination of CCND2 (cyclin D2), PCSK2 (proprotein convertase subtilisin/kexin type 2), and PLAB (growth differentiation factor 15) while Cerutti et al. specified completely different set of genes which comprised DDIT3 (DNA-damage-inducible transcript 3), ARG2 (arginase, type II), ITM1 (integral membrane protein 1) and C1orf24 (family with sequence similarity 129, member A). The lack of overlap between those studies is not unexpected, as changes in expression of many genes may provide similar differential diagnosis, the lack of independent confirmation of either study is more worrying. This indicates on heterogeneity of follicular thyroid tumours.

The studies mentioned above did not consider mutation status of FTC, yet it is known from PTC studies that it can have profound effect on gene profile. Indeed, studies by Lacroix et al. (Lacroix et al. 2005) and Giordano et al. (Giordano et al. 2006) identified a highly overlapping set of genes that were preferentially expressed in FTC with PAX8/PPARgamma rearrangements. Since many of the differentially expressed genes were known PPARgamma targets, these studies gave also a new insight into FTC pathobiology indicating that PAX8/PARgamma fusion gene may function as an oncogene analogous to other fusion genes present in leukaemia and soft tissue tumours (Teixeira 2006).

Cases of well-differentiated thyroid cancers (WDTC) with areas of poor or lacking differentiation, accompanied by progression in number of chromosomal abnormalities, suggest a model of progression from well-differentiated thyroid cancers to poorly differentiated or anaplastic ones (Wiseman et al. 2007). Indeed, molecular profiling showed that the MAP-kinase signalling pathway, which includes the well-known molecular lesions in WDTC, was also strongly affected in undifferentiated tumours (Montero-Conde et al. 2008). Other pathway showing important alterations was TGFbeta signalling pathway, associated with the epithelial-to-mesenchymal transition characteristic for less differentiated, advanced cancers (Grande et al. 2002). Other observations from microarray analysis also supported the hypothesis of progression from well to poorly differentiated thyroid cancers (Pita et al. 2009), suggesting that poorly differentiated cancers had expression profile closer to PTC, especially its follicular variant, than to FTC. The group of Santoro, by analysis of gene expression profile of anaplastic cancer, indicated on new potential molecular targets useful for therapy, among them POLO kinase (Nappi et al. 2009).

Although molecular etiology of autonomously functioning/hot thyroid nodules (AFTNs) is quite well established with activating TSH receptor observed in more than 60 % of cases, the molecular cause of cold thyroid nodules (CTNs), their benign, functional inactive counterparts is so far largely unknown. Also the data about altered signal transduction or changed expression of oncogenes, tumour suppressors or signalling molecules in thyroid nodules are rather scarce.

Recently a 20 genes set was identified for AFTNs has been established (Fujarewicz et al. 2007). In the selected gene set, cell adhesion and extracellular matrix-related genes were found, genes belonging to the wide group of growth process/apoptosis/immune response and signalling. By comparison with PTC, the AFTNs signature was not found in any of the cancers analyzed, that confirmed the low risk of malignancy in AFTNs. In the CTN multigene classifier mainly genes related to proliferation and growth processes were significant. Contrary to AFTN, the CTN multigene signature could be observed in papillary thyroid carcinoma, which indicated that gene expression changes characteristic for benign hyperplasia have to be separated from true cancer genes by further studies.