A nuclear receptor in thyroid malignancy: is PAX8/PPARgamma the Holy Grail of follicular thyroid cancer?

Several types of thyroid cancer ± classic papillary, anaplastic and medullary thyroid carcinoma ± can be confidently diagnosed and distinguished through fineneedle aspiration cytology (FNAC) of a thyroid nodule, thanks to their characteristic morphology and additional immunocytochemistry. In contrast, the finding of follicular thyroid neoplasia, a cytological group diagnosis encompassing the three entities follicular adenoma, follicular carcinoma or follicular variant of papillary carcinoma, still represents a major diagnostic dilemma. Capsule or vascular invasion, the two decisive criteria of follicular thyroid malignancy, can be demonstrated or excluded only on histopathological investigation of the surgically removed specimen. As only 15±25% of all follicular thyroid neoplasia prove to be malignant, the majority of patients who undergo follicular FNAC diagnosis are unnecessarily subjected to cost-intensive surgery and its potential complications (1, 2). For this reason, identification of appropriate markers that will allow highly specific and sensitive preoperative differential diagnosis of follicular neoplasia has a high priority. Encouraging results have been reported regarding altered thyroperoxidase immunoreactivity (TPOmAB47) or positive staining for glycoproteins CD44v6 and galectin-3 in the case of follicular thyroid cancer (3, 4). Another contributor to the diagnostic dilemma presented by follicular neoplasia is undoubtedly the prevailing uncertainty about its aetiopathogenesis which, despite the impressive progress in identifying specific genetic alterations in other nodular thyroid diseases, remains unsolved. In their recent study, Kroll et al. (5) suggested a novel pathomechanism in follicular thyroid malignancy, which follows up on previous reports of specific chromosomal translocations in these tumors (6). In a series of five follicular thyroid carcinomas with a karyotype showing a t(2;3)(q13;p25) chromosomal rearrangement, they were able to define specific translocation breakpoints on respective chromosomes using interphase fluorescence in situ hybridisation technology: the chromosome 2q13 breakpoint was located in the region coding for the thyroid transcription factor PAX-8 and the PAX-8 3q25 partner was identified as the peroxisome proliferator-activated receptor (PPAR) g1 by means of rapid amplification of cDNA ends (RACE) PCR. Sequencing of the RACE products generated from cDNAs of the follicular thyroid carcinomas showed an in-frame fusion of four PAX-8 variants (exons 1±7, 1±8, 1±9 or 1±7 plus 9), presumably generated by alternate splicing, to PPARg exons 1±6. This resulted in a PAX-8/PPARg1 fusion protein comprised of the paired and partial homeobox DNA binding domains of PAX-8, but devoid of its transactivation domain, and the DNA binding, ligand binding and retinoid receptor X (RXR) dimerization and transactivation domains of PPARg1 (Fig. 1). The expression of the fusion gene was demonstrated at the mRNA level by northern blot analysis with PPARg (1.8 kb) and PAX-8 (3.1 kb) cDNA probes, both indicating a 3 kb transcript. Further investigation at the protein level, by immunoprecipitation with an antibody to wild-type PPARg1, confirmed the presence of a 98 kDa protein, in agreement with the expected molecular mass of the fusion protein (87±98 kDa). Importantly, the PAX-8/PPARg rearrangement was detected only in follicular carcinoma (5/8), but not in follicular adenoma (0/20), papillary carcinoma (0/10) or multinodular goitre (0/10) using nested-RT-PCR with primers located in PAX-8 exons 6 or 7 and PPARg1 exon 1 respectively, for screening. In the remaining three follicular carcinomas, which were negative for the RT-PCR in this series, an alternate PAX-8/PPARg1 fusion gene was suspected in one case with a proven t(2;3)(q13;p25) karyotype, and a fusion of PPARg to a non-PAX-8 partner in another case with a 3p25 but absence of a corresponding 2q13 translocation. In the third follicular carcinoma, there was no indication of involvement of either PAX8 or PPARg1. Considering the established molecular components of thyroid disease, these novel data reported by Kroll et al. (5) are highly interesting for several reasons. Firstly, although limited by a small sample series and the presence of a particular karyotype, it is the first description of a specific molecular defect for follicular thyroid cancer. Secondly, it fits perfectly with an almost 10-year-old cytogenetics-based hypothesis by Hermann et al. (6) suggesting that disruption of a putative tumour-suppresor gene on chromosome 3p is peculiar to follicular thyroid malignancy (loss of heterozygosity for loci on 3p in six of six studied follicular cancers). ISSN 0804-4643 European Journal of Endocrinology (2001) 144 453±456