Familial Papillary Thyroid Cancer—Many Syndromes, Too Many Genes?*

Not so long ago, claiming that nonmedullary thyroid cancer could be anything but sporadic was heresy. Clinical and genetic investigations have come a long way, even in the last few years, and now it is almost dogma to state that approximately 5–10% of all thyroid carcinoma cases will be hereditary. A higher proportion (25%) of all medullary thyroid carcinoma (MTC) presentations are hereditary. Whereas the susceptibility gene for hereditary MTC, namely multiple endocrine neoplasia type 2 (MEN 2), is well defined, finding the genetic etiology for the familial nonmedullary thyroid carcinoma syndromes is proving arduous. Germline mutations of the RET proto-oncogene, on 10q11.2, are associated with more than 92% of all MEN 2 probands (1), and RET testing is the clinical standard of care for all MEN 2 and MTC patients (2). The search for the genes that cause familial nonmedullary thyroid cancer syndromes is proving to be less straightforward. Part of the reason is that familial nonmedullary thyroid cancer is not a homogeneous entity, unlike MEN 2. One gene for a familial nonmedullary thyroid cancer syndrome has been identified, PTEN, which encodes a tumor suppressor: germline mutations in PTEN have been found in 80% of individuals with classic Cowden syndrome, which is characterized by multiple hamartomas and a high risk of benign and malignant breast and follicular and papillary thyroid tumors (3) (Table 1). However, PTEN only accounts for 5% or less of families with breast and papillary thyroid carcinomas (PTCs) without other features of Cowden syndrome (4). Familial adenomatous polyposis, which predisposes to colorectal carcinoma, is caused by mutations in the APC gene, and PTC is acknowledged as a minor component tumor in this syndrome (Table 1). However, APC seems to have been excluded as a susceptibility locus in at least some familial PTC families (5). Thus, other susceptibility genes for PTC exist. The first “PTC gene” was localized to 19p13.2 in a single French family with papillary thyroid tumors with cell oxyphilia, which are an unusual type of oncocytic PTC (6) (Table 1). A putative locus for multinodular goiter was mapped to 14q31 in a Canadian kindred (7). In this family, two PTCs were noted. Indeed, the majority of other familial PTC families are not linked to either of these two loci (6–8). In this issue of the journal, Malchoff et al. (9) have found another locus for familial PTC, but with a twist. These investigators have mapped this gene to 1q21 using a multigenerational family segregating PTC and papillary renal cell carcinoma (RCC). Unlike the 19p-linked family, the PTCs found in this family are classic“ garden variety” PTC. This is rather significant because we can consider this 1q21 locus the first classic PTC susceptibility locus to be mapped. The association with papillary RCC is also worthy of note. Previously, investigators interested in families with RCC identified the MET proto-oncogene, encoding a receptor tyrosine kinase, as the susceptibility gene for familial papillary RCC (10). MET was excluded as a participant in the 1q-linked PTC-papillary RCC family (9). Another unusual family segregating clear cell RCC and PTC was found to harbor a constitutional translocation, t(3;8)(p14.2;q24.1) (11) (Eng, C. and R. S. Brown, unpublished observations). Thus, perhaps a gene on 3p14 or 8q24 might be considered yet another familial PTC susceptibility gene. Malchoff et al. (9) have shown that the region that should contain the putative familial PTC susceptibility gene is about 20 cm in length. This would have been a daunting distance “in the old days,” but given the technology and informatics fallout from the human genome project, these investigators should be able to identify their gene within a realistic time interval. In traditional positional cloning, the next step for this group would have been to acquire more PTC families in the hope that there would be no or not much genetic heterogeneity and that there would be a new critical recombinant that would help narrow the region further. Once the region was narrowed to, say, 5 cm these investigators would perform physical mapping: the assembly of a contig of cloned fragments that would span the interval of interest, after which each plausible candidate gene within the interval would be examined for germline mutations in the families. Unfortunately, we already know that familial nonmedullary thyroid cancer, indeed even familial PTC, is not a single syndrome but several syndromes (Table 1). Furthermore, assigning affected status might not be straightforward, especially in smaller families; benign thyroid nodules are relatively common, occurring in 10% of females. In mapping the 19p familial PTC locus, Canzian et al. (6) assigned individuals with nodules and multinodular goiters as affected. It is unclear whether non-PTC nodules should be considered part of the phenotype or whether this would be more misleading than helpful. After all, there has never been solid evidence that demonstrates that such nodules represent premalignant lesions. Given the early mapping efforts (5, 6, 8, 9), we already know that there will be at least four susceptibility genes for familial PTC. There likely will be more. Given these inherent problems of syndromic heterogeneity, genetic heterogeneity, and a paucity of large families, how should these investigators proceed, barring serendipity? One way is to collect a series of PTC and paired normal thyroid sets and subject their messenger RNA transcripts to comparative expression array analysis. Transcripts that clearly show consistently decreased expression between normal thyroid and PTC (if looking for a tumor suppressor) or consistently increased between normal tissue and cancer (if an oncogene) would be of interest. Such classes of transcripts that map to the 1q21 20-cm interval would be the most promising candidate genes. In this manner, the investigators would be able to perform targeted mutation analysis. Another useful adjunctive maneuver would be for these investigators to use their contig of BACs (bacterial artificial chromosomes) to examine either for loss of heterozygosity or amplification in component tumors. What type of gene should these investigators be looking for? If we may take any clues from the sporadic setting, then they should be looking for proto-oncogenes, which encode kinases. RET and NTRK1, both encoding tyrosine kinases that are not normally expressed in thyroid follicular epithelium, are activated by being translocated and juxtaposed against the 5′ ends of genes that drive increased expression in the thyroid follicular cells, leading to PTC formation. The MET proto-oncogene belongs to the same family of receptor tyrosine kinase genes as RET, and it is the susceptibility gene for another papillary cancer. Furthermore, somatic MET mutations have been found in both sporadic papillary RCC and PTC (10). In addition, the putative PRCC gene, scissioned at 1q21.2 in a sporadic papillary RCC (12), seems a promising PTC candidate gene, and, indeed, this sort of mechanism is not inconsistent with activation of a proto-oncogene. In considering the putative PTC susceptibility genes, we must also ponder the etiology of sporadic PTC. For example, it has been well established that radiation exposure causes PTC by promoting the formation of the RET translocation. In Japan, there seems to be an increased incidence of Hashimoto thyroiditis-associated PTC, a proportion of which occurs in small familial clusters. Thus, we may postulate that one of the familial PTC susceptibility genes might not be a traditional oncogene or tumor suppressor gene but one that increases susceptibility to radiation exposure. Alternatively, there might be a gene or genes that promote autoimmune thyroiditis. In summary, it would seem that familial PTC is a “catch-all” term that encompasses different syndromes with genetic susceptibility to PTC. Malchoff et al. (9) might be correct in suggesting that familial PTC-papillary RCC is one such syndromic entity and that a particular putative susceptibility gene will be on chromosome arm 1q. Because the histology of PTCs in this kindred is “standard,” it might not be unreasonable to suggest that the putative 1q susceptibility gene could account for more families with PTC than the 19p locus. However, given the multiple syndromes represented by “familial PTC” and the likelihood of multiple susceptibility genes, likely comprised of those with variable penetrances and expressivity, researchers dedicated to sorting out the genetic etiology of familial PTC will be kept busy for years to come.\. Acknowledgments I thank Oliver Gimm and Albert de la Chapelle for critical review of the manuscript and acknowledge all the investigators who have contributed to the study of familial PTC but who could not be cited here because of space limitations. Inherited cancer syndromes where PTC may be a component feature PTCs have been found in rare Cowden individuals who are germline PTEN mutation negative, although mixed papillary-follicular histologies have been found in families that are PTEN mutation positive. Inherited cancer syndromes where PTC may be a component feature PTCs have been found in rare Cowden individuals who are germline PTEN mutation negative, although mixed papillary-follicular histologies have been found in families that are PTEN mutation positive.