In this study, FOXA1 expression was investigated in thyroid lesions and tumors using whole immunohistochemical sections. No expression was detected in all inflammatory diseases, all multi-nodular goiters, and all adenomas, including follicular, Hurthle cell, and hyalinizing trabecular adenomas. Previously, Nucera et al.  reported as well that there was no expression of FOXA1 in all 30 follicular adenomas, 58 nodular hyperplasias, 8 lymphocyte thyroiditis, and 6 Graves’ diseases. More recently, no FOXA1 immunostaining was observed by Nonaka  in all benign thyroid lesions, including 10 nodular hyperplasias, 10 Hashimoto thyroiditis, and 10 Graves’ diseases.
Moreover, no FOXA1 immunostaining was detected in all differentiated thyroid carcinomas, including the 19 papillary carcinomas and 16 follicular carcinomas. Similarly, Nucera et al.  showed that the 60 differentiated carcinomas (48 papillary carcinomas and 12 follicular carcinomas) exhibited no specific FOXA1 expression. Furthermore, Nonaka  neglected any FOXA1 staining in 21 papillary carcinomas, 27 follicular, and 13 low-grade carcinomas. In addition, no FOXA1 expression was found in 15 Hurthle cell carcinomas of the thyroid . Herein, the only case of Hurthle cell carcinoma did not exhibit any FOXA1 positivity.
The absence of FOXA1 expression in benign lesions as well as in well-differentiated cancers of the thyroid could be explained by the silencing of the FOXA1 gene, transcriptional repression, or hypermethylation of the FOXA1 gene. Interestingly, in a very recent study, Chen et al.  demonstrated that FOXA1 is a direct target of miR-132 in papillary carcinoma cells. The FOXA1 decrease in thyroid cancer cells significantly inhibits cell proliferation, migration, and invasion, mimicking the suppression effect induced by the overexpression of miR-132 . Restoration of FOXA1 expression partially reversed the suppression effect induced by the miR-132 overexpression. Therefore, the authors claimed that miR-132 acts as a tumor suppressor targeting FOXA1 in thyroid papillary carcinoma . Similarly, in other human diseases, FOXA1 expression has been found to be inhibited by miR-30a-5p and miR-212 in hepatocellular carcinomas [24, 25], by miR-212 in osteosarcomas , by miR-194 in non-small cell lung cancers , and by miR-431 in necrotizing enterocolitis .
The current study included only one case of undifferentiated sarcoma of the thyroid. To the best of our knowledge, there was no previous study analyzing the expression of FOXA1 in mesenchymal tumors of the thyroid. The absence of FOXA1 expression in all tumor cells neglected the involvement of FOXA1 in the development of this mesenchymal tumor. More multicenter studies, using a much larger series of these rare malignancies, will be necessary to further confirm these results.
In our study, only three anaplastic carcinomas exhibited focally FOXA1 immunostaining. By contrast, in the Nucera et al. study , FOXA1 expression was reported in 90% of anaplastic carcinomas. The expression was strong in 70% of cases, weak in 20% of cases, and absent in the remaining cases . In addition, the DNA copy number of FOXA1 in the 14q21.1 locus was analyzed by fluorescent in situ hybridization . Nuclear overexpression of FOXA1 was associated with the amplification of the number of FOXA1 DNA copies. Therefore, these researchers suggested the potential role of FOXA1 as an oncogenic transcription factor in thyroid cancer . More recently, Nonaka  analyzed also the expression of FOXA1 in a larger series of anaplastic carcinomas. The expression of FOXA1 was detected in 55% of the studied cases, and it was variable in intensity and extent . Furthermore, well-differentiated components were all FOXA1-negative. No correlation was found between the FOXA1 staining and morphological subtype of anaplastic carcinomas . The difference in FOXA1 expression pattern observed between our study and previous reports could be explained partly by technical particularities. The immunohistochemistry on tissue microarray or whole sections and the use of different anti-FOXA1 antibody clones could be at the origin of this discrepancy.
To our knowledge and until the submission of this paper, there were only few previous studies investigating the FOXA1 expression in medullary carcinomas [9, 29]. Nonaka  described a diffuse and homogeneously strong FOXA1 nuclear expression in the tumor cells of all 67 medullary carcinomas regardless of cell type, growth pattern, mitotic count, presence of necrosis, and primary or metastatic. As compared to the heterogeneous expression of calcitonin and CEA reported in the same cases, Nonaka  considered that FOXA1 expression could serve as a reliable auxiliary marker for the diagnosis of medullary carcinoma of the thyroid. Interestingly, herein, we observed FOXA1 staining in all medullary carcinomas. Although we used a different FOXA1 antibody clone, the immunostaining was scored either 4+ or 3+ in all medullary carcinomas. Furthermore, the pattern of FOXA1 staining was similar to that of calcitonin and chromogranin A. These findings altogether support the usefulness of FOXA1 expression as a potential biomarker refining the diagnosis of medullary thyroid carcinoma. Nevertheless, since there was no significant correlation with all clinicopathological parameters and tumor recurrence, the prognostication role of FOXA1 is limited in these thyroid carcinomas.
In our study, the expression of FOXA1 was detected mainly among medullary carcinomas, supporting its specificity to thyroid C cell tumors. Interestingly, Nonaka  showed that all foci of C cell hyperplasia associated with thyroid lesions were diffusely and strongly FOXA1-stained. In addition, recently, the embryonic origin of thyroid C cells in mice and humans has been revised by Johansson et al. . Their study demonstrated that mouse thyroid C cells developed from the pharyngeal endoderm and not from the neural crest. Moreover, propagation of the C cell lineage involved FOXA1 and FOXA2 both together play crucial roles in organogenesis from the foregut endoderm . Additionally, Johansson et al.  showed that FOXA1 promotes the growth of neoplastic thyroid C cells. Furthermore, strong FOXA1 and FOXA2 immunostaining was detected in human medullary thyroid carcinomas, including both primary tumor nodules and lymph node metastases .
Finally, the current study presented some limitations since it is a retrospective survey of a relatively small number of cancer cases and heterogeneous histological types of thyroid cancer. Thereby, more multicenter studies, using a much larger series of medullary and anaplastic carcinomas, should be conducted to further confirm our findings.