Abstract
The specificity of CDX2 immunohistochemical expression in some subtypes of thyroid carcinoma is currently the subject of debate (1 –3). In papillary thyroid carcinomas (PTC), CDX2 expression has almost always been found only in the columnar cell variant (PTC-CCV) (1,3).
The PTC-CCV displays some very specific cytomorphological features, most importantly its being made up of pseudostratified columnar cells, which may contain a cytoplasmic vacuole resembling intestinal or endometrial differentiation (4). Recently, PTC-CCV immunohistochemical and molecular profiles have been defined (5). This peculiar PTC variant is traditionally divided into two clinically different subtypes: an indolent subtype, encapsulated or limited to the thyroid, and a more aggressive, usually widely invasive variant (2,3).
CDX2 expression has always been considered a specific marker of the gastrointestinal tract, and its expression has been used to prove a colorectal origin in doubtful neoplastic cases. In colorectal cancer (CRC), CDX2 suppresses tumorigenesis. Thus, its expression is lost in patients with high tumor grade, advanced tumor stage (stage IV), and/or reduced survival and in tumors displaying a phenotype of high CpG island methylation or microsatellite instability. Moreover, mutated BRAF in CRC patients favors tumor cell proliferation and is associated with overall survival reduction (6). In this setting, a significant correlation has been reported between the presence of a BRAF mutation and the absence of CDX2 expression (6). A BRAF mutation is detected in approximately 40–50% of all PTCs as well; numerous studies have found this mutation to correlate with aggressive features and with tumor progression (7), whereas others have not (8).
The inverse correlation between the presence of a BRAF mutation and CDX2 in CRC led us to study this relationship in PTC-CCV as well and to analyze the relevant literature data (1 –3,5). We investigated it by means of a comparison with other PTC subtypes, as the comparison would allow us to determine whether CDX2 expression was specific only to PTC-CCV.
Six cases of pure PTC-CCV and 30 cases of PTC [18 cases of classic type, 7 cases of follicular variant, 4 cases of tall cell variant, and 1 case of diffuse sclerosing variant, according to the WHO criteria (9)] were analyzed for CDX2 by immunohistochemistry (clone EPR2764Y; Ventana, Tucson, Arizona) and for a BRAF mutation by direct sequencing of genomic DNA. All 36 cases—30 (81%) females and 6 (19%) males, mean age of 45.2 years (range 16–78 years)—were retrieved from the database of our pathology departments. We subclassified the six PTC-CCVs in the study into aggressive and nonaggressive subtypes, as suggested by Wenig et al. (4). Five of the six cases were well-defined indolent lesions with a thin fibrous capsule, whereas the sixth was widely invasive and extended to the perithyroid tissues. The main clinicomorphological data are summarized in Supplementary Table S1 (Supplementary Data are available online at
CDX2 expression was found in 4 out of the 6 (66%) PTC-CCV cases. Specifically, three cases, including the widely invasive one, showed a strong and diffuse nuclear positivity of CDX2 in all cells (Supplementary Fig. S1A, B), and the nuclear staining was present in about 20% of cells in the fourth case. The remaining 2 cases were negative (Supplementary Fig. S1C, D), as were all 30 PTC samples analyzed. The difference in CDX2 expression between PTC and PTC-CCV was statistically significant (p=0.0002, two-tailed Fisher's exact test), confirming that CDX2 expression was strictly associated with PTC-CCV.
BRAF mutations were detected in 21 out of the 30 (70%) PTC cases, and in only 1 out of the 6 (17%) PTC-CCV cases, the widely invasive one. This difference was not statistically significant, most likely because of the low number of PTC-CCV cases analyzed. Thus, we cannot confirm the correlation between the loss of CDX2 expression and the occurrence of a BRAF mutation, as has been demonstrated in CRC. In our study comparing PTC and PTC-CCV, the presence of a BRAF mutation was associated with cases of more advanced stages (76.2% of T2/T3 cases vs. 40% of T1 cases, p=0.04, two-tailed Fisher's exact test).
Our results corroborate those of Enriquez et al. (1) but are at variance both with those in a very recent article by Sujoy et al. (3), who found CDX2 immunoreactivity in only 1 out of 10 PTC-CCV cases, and with those of Cameselle-Teijeiro et al. (2), who found CDX2 only in a classic PTC. These conflicting results provide an incentive to take the analysis of PTC-CCV further and to verify whether its peculiar immunomorphological features are somehow related to the “intestinal-type” differentiation of some PTCs (10). However, in the presence of thyroid lesions with a morphology resembling a colonic adenocarcinoma, the expression of CDX2 should not be regarded as proof of metastatic disease. Instead, the immunohistochemical evaluation of markers of thyroid origin, such as thyroglobulin and/or thyroid transcription factor-1, should be performed.
In conclusion, CDX2 expression in our series was limited to PTC-CCV; its positivity can be used to confirm the diagnosis. The discrepancy of results from various studies thus far suggests that future research is needed to assess the role of CDX2 in a larger cohort of PTC-CCV and whether there is any association with other molecular markers such as BRAF mutations.