Cytoplasmic Staining of TTF-1 Antibody in the Diagnosis of Hepatocellular Carcinoma : Study on 299 Cases Using TissueMicroarray

The diagnosis of hepatocellular carcinoma (HCC) remains obscure in some complicated cases. Thyroid transcription factor-1 (TTF-1) was presumed to be helpful in the diagnosis of HCC. This paper aims to study the utility of TTF-1 in the diagnosis and differential diagnosis of HCC and to assign clinicopathological correlations. Immunohistochemical detection of TTF-1 is performed with tissue microarray which contains a total of 781 cases including 299 hepatocellular carcinomas (HCCs), 31 intrahepaticcholangiocarcinomas (ICCs), 86 metastatic adenocarcinomas (MACs), 5 metastatic hepatocellular carcinomas (mHCCs), 9 hepatoblastomas (HBs), 40 renal clear cell carcinomas (RCCs), 215 adjacent nonneoplastic hepatic tissue, and 96 normal hepatic tissue. The result shows that TTF-1 is positive in 55.2% (165/299) HCCs, 6.4% (2/31) ICCs, 3.5% (3/86) MACs, 60% (3/5) mHCCs, 33.3% (3/9) HBs, 0% (0/40) RCCs, 99.0% (213/215) adjacent nonneoplastic hepatic tissue, and 95.8% (93/96) normal hepatic tissue. Expression of TTF-1 is correlated with differentiation of HCC, but not with patients’ age, gender, serum AFP, tumor size, number of tumor nodules, and tumor thrombus in vein or bile duct. These data suggest that TTF-1 antibody is sensitive and specific in HCC and might serve as a candidate for the diagnosis and differential diagnosis of HCC.


Introduction
Hepatocellular carcinoma (HCC) is the most common malignancies in liver.Generally accurate pathological diagnoses are not difficult to be reached.However, confirmatory methods are still required in certain complicated cases.Although some immunohistochemical markers, such as αfetoprotein (AFP), hepatocyte paraffin 1 (Hep Par1), and CK8, have played critical parts in practice, each of them has showed some limitations [1,2].Exploring for more valuable immunohistochemical markers is of great help to establish the diagnosis of HCC in those uncertain cases.Thyroid transcription factor-1 (TTF-1) has been commonly used in the diagnosis of thyroid and pulmonary tumors, prediction for prognosis of which has been proved in the literature [3][4][5].In the past years, TTF-1 staining was well known to be nuclear, with only occasional reports of cytoplasmic staining.However, increasing attention has been drawn to the significance of cytoplasmic staining of TTF-1 in both HCC and nonneoplastic hepatic tissue recently [1,6,7], which suggested that TTF-1 may serve as the potential valuable marker to confirm the diagnosis of HCC in uncertain cases.Nevertheless, in previous studies, there are not enough number of cases to verify the significance of cytoplasmic staining of TTF-1 nor did they correlate the expression of TTF-1 with clinical data.These limited their power of persuasion.In the present study, immunohistochemical detection of TTF-1 is performed with tissue microarray containing 776 specimens, and the clinicopathological correlations were analyzed to verify the diagnostic significance of TTF-1 in HCC.
The tissue samples consist of 299 HCCs, 31 CCs, 86 MACs, 40 RCCs, 215 adjacent non-tumor livers (liver tissue surrounding HCC), and 96 normal livers.Two cores were punched in each case; each of tissue cores is 1.0 mm in diameter; consecutive 4 μm thick sections were cut and airdried overnight.One section from the tissue microarray block was stained with hematoxylin and eosin to confirm the homogeneity of the cell populations in the tissue samples.4 mHCCs and 9 HBs were performed as conventional sections.

Immunohistochemistry.
The sections were deparaffinized in xylene and rehydrated in graded series of ethanol.For antigen retrieval, the sections were immersed in EDTA solution (pH 9.0) and processed in a heated steamer at 100 • C for 5 minutes.After cooling at room temperature for 30 minutes, slides were washed in phosphate-buffered saline solution (pH 7.4) and treated with 3% hydrogen peroxide for 15 minutes.The sections were incubated for 1 hour with the monoclonal mouse antibody specific for TTF-1 (clone 8G7G3/1, Santa Cruz,USA) at a dilution of 1 : 100, then washed with PBS solution (pH 7.4), and incubated for 30 minutes at 37 • C with the secondary antibody (catalog: PV-8000, Zhongshan Co., itd, China), which is a biotin-free detection system that consists of a secondary antibody covalently linked to peroxidase-coated dextrose polymers.Finally, the sections were visualized with 3,3diaminobenzidine (DAB, DAKO) and counterstained with hematoxylin.Normal thyroid tissue sections were used as positive controls.Negative controls substituted PBS buffer for primary antibody.
Intensity was designated as weak when immunostaining was present but only barely detectable.To correlate extent and intensity of immunostaining, these values in positive cases were converted into composite immunohistochemical scores by multiplying the individual scores of extent by intensity (possible range of values from 0 to 6).For example, a moderate 3 + extent (3 points) and moderate intensity of immunostaining (2 points) would have an immunohistochemical composite score of 3 × 2 = 6 [9]; according to this, the results of immunostaining were classified as negative (0, 1), moderate positive (2, 3), and strong positive (4, 6).

Statistical Analysis.
Chi-square analysis was used for statistical analysis.A P value of less than .05was considered statistically significant.For all statistics, SPSS 13.0 for Windows software (or other) was used.
In MACs, 3 (3.5%) of 86 cases showing cytoplasm staining with diffuse and weak reactivity, all of these were welldifferentiated adenocarcinoma.In ICCs, 2 (6.4%) of 31 caes, showed weak cytoplasmic staining for TTF-1.No one RCC was positive for TTF-1.3 mHCCs (60%) revealed cytoplasmic staining for TTF-1 (all of them were vertebra metastasis) (Figures 2(g) and 2(h)).Additionally, in 9 HBs, 3 cases demonstrated cytoplasmic staining for TTF-1 (the staining type was similar to HCC), with only fetal cells showed positive, while embryonal cells-relatively immature-did not react with TTF-1 (Figures 2(e) and 2(f)).Positive control (thyroid tissue) presented clear and strong nuclear staining.Negative control does not react.The differences between HCC and all of ICC, RCC, MAC, and HB were statistically significant (Table 3).As for different histopathological patterns of HCC, cytoplasmic staining was presented in 56.2% of trabecular pattern (154/274), including 2 cases of clear  4).

Clinicopathological Correlations. Expression of TTF-
1 correlated significantly with the grade of HCC, which was stronger in well-differentiated carcinoma (grade I + II) (64.5%; 151/234) than in poorly differentiated ones (grade III + IV) (21.5%; 14/65).The difference between high-grade and low-grade carcinoma was significant (P = .01).There was no correlation between TTF-1 expression and sex, age, serum AFP, tumor size, number of nodules, and vein or bile duct tumor thrombus (Table 5).

Discussion
TTF-1 is a 38-kDa nuclear protein initially identified as a mediator of thyroid-specific gene transcription, the code gene located on chromosome 14q13.It is a member of the NKx2 family of homeodomain transcription factors and is expressed in thyroid and lung tissues [11].Expression of TTF-1 was characteristically seen in the nuclei of tumor cells origin from thyroid and lung.Cytoplasmic staining was reported occasionally and presumed as a nonspecific staining pattern previously [12,13].However, the diagnostic utility of cytoplasmic reactivity of TTF-1 has drawn the interest of some researchers recently.Rates of cytoplasmic reactivity of TTF-1 ranged from 64% up to 93% in HCC in the literature [1,6,7], which was totally different from those found in lung and thyroid tumors, suggesting the potential application of TTF-1 antibody in the diagnosis of HCC.Nevertheless, there was not a consensus of opinion in terms of the mechanism of cytoplasmic staining of TTF-1, based on the similar staining type between TTF-1 and HPA; Pang et al. [14] draw conclusion that different epitopes to HPA and TTF-1 share the same protein molecule in hepatocyte mitochondria using immunoelectron microscopic.While a recent study by Gu et al. [15] demonstrates that TTF-1 immunoreactivity in the hepatocyte cytoplasm is due to an approximately 160-kd protein and this unique protein is neither an alternative splicing product of TTF-1 nor does it express in thyroid and lung tissues in light of RNA and protein analyses; in our study cytoplasmic TTF-1 staining in HCC and normal hepatocytes is a reliable reaction, as endogenous peroxides were excluded by using 3% hydrogen peroxide in the process, and artifact from endogenous biotin in hepatocytes could be eliminated in the biotin-free detection system.In addition, external positive and negative controls, as well as internal negative control such as bile duct epithelium, endothelial cells, and Kupffer cells, were used to verify the specificity of the antibody.Although similar results have been described before in the literature, the case numbers were not large enough to convince the specificity of cytoplasmic TTF-1 staining in hepatocytes.Our study includes 299 HCCs using tissue microarray to detect the expression of TTF-1 and finds that the positive staining type of TTF-1 in HCC, hepatic tissue, and adjacent nonneoplastic hepatic tissue is exclusively cytoplasmic reactivity, without any nuclear staining case detected.Our study shows that 55.2% (165/299) HCCs are positive for TTF-1, which are significantly higher than that in ICC (6.4%), MAC (3.5%), and RCC (0%).In addition to this, 60% (3/5) metastatic HCCs showed positive reactivity.And the expression rate of TTF-1 in adjacent non-neoplastic hepatic tissue and normal hepatic tissue is up to 99.0% and 95.8%, respectively.These results proved that cytoplasmic staining of TTF-1 is specific to HCC and hepatocytes.It can be used as a candidate marker in the diagnosis and differential diagnosis of HCC.Even the application of detecting the original tissue of metastatic carcinoma would be possible.The staining type, cytoplasmic or nuclear, could do favor to distinguish the original tissue.
Nevertheless, the expression rate of TTF-1 in HCC in our study (55.2%) was lower compared with previous studies which may attribute to difference in number of cases in the studies, and different antibody agent applied as well.Furthermore, we investigated the correlations between the expression of TTF-1 in HCC and their clinicopathological features, which show that tumor grades are parallel with the level of TTF-1, with the expression rates being lower in highgrade cases (poorly differentiated) compared with low-grade cases (well differentiated), 21.5% versus 64.5%, respectively.However, there is no correlation between TTF-1 expression and sex, age, level of serum AFP, tumor size, number of nodules, and vein or bile duct tumor thrombus.In addition to this, an interesting phenomenon is found in HB.TTF-1 only expresses in fetal cells, but not embryonal cells, which are less differentiated in morphology than fetal cells in HBs [16].It seems that expression rate of TTF-1 in HCC and HB (neoplastic component) correlates with differentiation stage of tumor.However, profound senses of these findings are needed to be explored in the future.
In summary, our study focuses on the utility of TTF-1 in the diagnosis and differential diagnosis of HCC with a large number of cases using tissue microarray.Cytoplasmic TTF-1 staining is sensitive and specific to HCC.The clinicopathological correlations are analyzed in our study, which may help to understand what role cytoplasmic staining of TTF-1 plays in HCC exactly.In all, TTF-1 antibody may serve as a candidate for making diagnosis and differential diagnosis of HCC.

Table 1 :
Clinical and pathologic information.

Table 5 :
Clinicopathologic correlations between biologic factors and patients with HCC.