Exploration of immunocytochemical biomarkers related to central lymph node metastasis in papillary thyroid microcarcinoma

INTRODUCTION

Thyroid cancer is undergoing rapid growth as a malignant tumor on a global scale. Specifically, papillary thyroid microcarcinoma (PTMC), which is defined as papillary thyroid cancer (PTC) with a maximal tumor diameter not exceeding 1 cm, accounts for almost 50% of newly diagnosed cases of thyroid cancer.^[1]

Extensive literature has indicated that certain PTMCs consistently persist in a subclinical stage, exhibit a favorable prognosis, and can even coexist with the tumor; these PTMCs with good clinical outcomes are termed low-risk PTMCs.^[2] Although the criteria differ among various countries and regions, PTMCs that fulfill the criteria of no lymph node metastasis (LNM), no distant metastasis, no extrathyroidal invasion, no pathological high-risk subtype, and no tracheal or recurrent laryngeal nerve invasion are commonly defined as low-risk PTMCs.^[3] Low-risk PTMCs can be managed through active surveillance, ablation therapy, or surgery. Active surveillance and ablation therapy can circumvent surgical and post-operative complications, thereby significantly enhancing the quality of life for these patients.^[4] However, many countries and regions primarily choose surgical intervention when dealing with low-risk PTMCs.^[5] One of the major factors is that the existing clinical examination techniques in numerous countries are incapable of accurately determining whether the cervical lymph nodes have metastasized prior to treatment. PTMC cases demonstrate a low incidence of extrathyroidal invasion, high-risk pathological subtypes, tracheal or recurrent laryngeal nerve invasion, and distant metastasis, while PTMC cases have a substantial incidence of cervical LNM. Therefore, based on the definition of low-risk PTMC, the key determinant for classifying PTMC as low-risk is whether there are cervical LNMs.^[3] Nevertheless, the two main clinical assessment methods for preoperative cervical LNM have certain limitations: ultrasound examination has low accuracy in predicting LNM, especially in some institutions with poor medical standards; fine needle aspiration (FNA) examination is highly prone to accidentally injure large blood vessels and nerves when puncturing the central lymph nodes, and the puncture risk is extremely high.^[6] Consequently, if doctors follow international recommendations and administer active surveillance or ablation treatment to PTMC patients who show no clinically detectable LNM (cN0), there is a risk of delaying treatment for many patients who actually have LNM.

However, the approach of relying on surgical treatment sometimes results in unnecessary surgeries for numerous PTMC patients who do not actually have LNMs. A precise diagnosis is of utmost importance for precise treatment. Hence, it is vital to emphasize the need for new techniques in the medical setting to precisely determine the preoperative condition of cervical lymph nodes. This will permit the accurate diagnosis of low-risk PTMC and subsequently facilitate the implementation of tailored and personalized treatment, thus enhancing the patient’s chances of recovery and overall quality of life.

Employing clinical variables to predict PTMC cervical LNMs is a frequently employed strategy; however, its prediction accuracy is relatively low.^[7-9] The genetic information of tumor cells serves as the foundation for their biological functioning, and the advancement of high-throughput sequencing and bioinformatics technologies has made the investigation of molecular biomarkers a highly prospective approach for addressing this issue.

In our previous study,^[10] we discovered that the presence of squamous cell components in PTCs was correlated with a poor prognosis. Based on this finding, we postulated that biomarkers related to squamous cell components might be linked to central LNM (CLNM) in PTMC. Based on this hypothesis, this study aims to utilize bioinformatics analysis in conjunction with immunohistochemistry (immunocytochemistry) validation to investigate and confirm biomarkers associated with squamous cell components or other biomarkers with potential predictive value for CLNM in PTMC, thereby providing a reference for the clinical management of PTMC patients.

MATERIAL AND METHODS

Statistical analysis

A Chi-square test was employed to evaluate the disparity in immunohistochemical examination levels between the groups of PTMC patients with CLNM and those without metastasis using the statistical package for the social sciences (version 19.0, international business machines corporation, Armonk, NewYork, USA). If the P-value was >0.05, it was considered as statistical significance. For the chi-square test implementation, we selected between the Pearson Chi-square test and the continuity-corrected Chi-square test, taking into account both the total number of samples and the theoretical frequency. Additionally, we proceeded to calculate the effect size of our obtained result, which provides a more comprehensive understanding of the practical significance and magnitude of the associations detected in our study.

The pipeline of the “Materials and Methods” section is shown in Figure 1, which is created by Microsoft Office Word software (Microsoft Office 2017, Redmond, Washington, USA).

Close

Figure 1:

The pipeline of the “Materials and Methods” section. mRNA: Messenger ribonucleic acid, TCGA: The cancer genome atlas, pT1: Pathological tumor stage 1, PTC: Papillary thyroid carcinoma, DEG: Differentially expressed gene, PTMC: Papillary thyroid microcarcinoma.

Export to PPT

RESULTS

Hub genes related to LNM

A total of 132 instances of PTC with a pathological T stage of pT1 and a node stage were extracted from the TCGA database. This comprised 44 cases in the lymph node-positive group (pT1N1 PTC group) and 88 cases in the lymph node-negative group (pT1N0 PTC group). In accordance with the procedure described in the “Materials and Methods” section, a total of 292 DEGs were successfully identified. Among these, 87 genes were found to be up-regulated, while 205 genes were down-regulated. Figure 2 illustrates the heatmap and volcano plot of the DEGs.

Close

Figure 2:

(a) Heatmap of 292 DEGs retrieved from the TCGA database; (b) Volcano plot of 292 DEGs. The red color indicates the genes that have been upregulated, the green color represents the genes that have been downregulated, and the black color denotes genes without change. DEG: Differentially expressed gene, TCGA: The cancer genome atlas.

Export to PPT

In addition, the STRING database was employed to construct a PPI network for the identified DEGs. Subsequently, the string file was imported into the Cytoscape software to screen for the top 10 hub genes. Figure 3 presents the top 10 hub genes with the highest degree value. Among these genes, KRT5, KRT6A, KRT14, KRT6B, and KRT16 are upregulated, whereas CHGA, Neuropeptide Y, Calcitonin Related Polypeptide Alpha, Parathyroid Hormone and PAX2 are downregulated.

Close

Figure 3:

(a) PPI network constructed among DEGs with the criterion of interaction score >0.4; (b) Top 10 hub genes in PPI relationship pairs were ranked in the table and depicted in the graph. PPI: Protein-protein interaction, DEG: Differentially expressed gene.

Export to PPT

Immunohistochemical and immunocytochemical validation

The study screened 272 suitable cases out of a total of 2091 instances of PTC based on the criteria for inclusion and exclusion. After using propensity score matching to ensure equivalence in age, gender, and diameter between the two groups, a total of 104 cases with CLNM and 104 instances without LNM were selected for further immunohistochemical staining [Figure 4a, created by Microsoft Office Word software]. Propensity score matching demonstrated that the standardized mean difference (SMD) after matching was below 0.1 [Figure 4b]. Typically, an SMD <0.1 implies a favorable balance when assessing group differences, indicating that the disparity between the study groups is minimal.

Close

Figure 4:

(a) The screening process of cases for immunohistochemical and immunocytochemical validation; (b) The standardized mean difference between the two groups of cases before and after propensity score matching.

Export to PPT

Among the 208 cases, the biomarkers (CK5/6, CgA, and Pax-2) corresponding to the main hub genes (KRT5, KRT6A, CHGA, and PAX2) were immunohistochemically stained in postoperative thyroidectomy specimens. The immunohistochemical staining results revealed a statistically significant difference in the expression of CK5/6 between pN1a PTMC group and pN0 PTMC group (P = 0.002). PTMC cases with CLNM were more likely to be positive for CK5/6, the positive model mainly displayed spot-positive characteristics at the periphery of the tumor lesion, with a small part exhibiting strong patchy positivity [Figure 5a-d]. However, the expression levels of CgA (P = 0.208) and Pax-2 (P = 0.999) did not show any significant differences between the two groups [Table 1].

Close

Figure 5:

(a) PTMCs with lymph node metastases were more likely to display spot-positive CK5/6 expression (immunohistochemical staining, ×400); (b) A small number of PTMCs with lymph node metastases were intensely patchy-positive for CK5/6 (immunohistochemical staining, ×400); (c) CK5/6 positive sites with lymph node metastases were usually predominantly located at the periphery of the tumor lesion (immunohistochemical staining, ×100). (d) CK5/6 negative staining result (immunohistochemical staining, ×100). (e and f) Presentation of positive and negative results of CK5/6 immunocytochemical staining conducted in preoperative thyroid FNA liquid-based samples (immunocytochemical staining, ×800). (g) The staining results of CK5/6 in immunohistochemistry and immunocytochemistry. PTMC: Papillary thyroid microcarcinoma, CK5/6: Cytokeratin 5/6, FNA: Fine needle aspiration, IHC: Immunohistochemistry, ICC: Immunocytochemistry, CLNM: Central lymph node metastasis.

Export to PPT

Subsequently, CK5/6 immunocytochemical staining was conducted in preoperative thyroid FNA liquid-based samples, and the results further verified that CK5/6 was more likely to be positively expressed in PTMCs with CLNM (P = 0.010) [Figure 5e-g].

Table 1: Immunohistochemical staining results related to CLNM in papillary thyroid microcarcinoma.

Biomarkers	CLNM		P	Effect size (Phi coefficient)	Odds ratio (95% confidence interval)
	−	+
CK5/6			0.002	0.214	3.296 (1.489–6.951)
−	94	77
+	10	27
CgA			0.208	0.087	1.709 (0.7673–3.920)
−	94	88
+	10	16
Pax-2			0.999	0.070	+infinity (0.1111 –+infinity)
−	104	103
+	0	1

CLNM: Central lymph node metastasis, CK5/6: Cytokeratin 5/6, CgA: Chromogranin A, Pax-2: Pair box gene 2

DISCUSSION

Numerous studies have documented molecular markers correlated with the onset and progression of PTC.^[11] Ruiz and colleagues performed an analysis of DEGs related to LNM in PTC and identified a panel of 25 genes that can be utilized to differentiate between the pN0 and pN1 groups of PTC samples. These genes exhibited a sensitivity of 86% for detecting LNM.^[12] Zhang et al. reported that an 11-gene signature displayed good discriminative performance for lymph node metastasis (LNM) prediction (area under the curve [AUC] = 0.847), and TIPARP mRNA was upregulated in PTC with LNM versus with-out (AUC = 0.692).^[13] Moreover, Napoli et al. identified that four micro ribonucleic acids (miR-154-3p, miR-299-5p, miR-376a-3p, and miR-302E) exhibited a significant differential expression in cytological samples of PTC with LNM and predicted the positive nodal status with relatively good performance (AUC = 0.66–0.72).^[14] In general, a combination of DEGs is capable of determining whether PTC has LNM. This method demonstrates reliability and scientific rigor, making it a valuable tool for predicting PTMC LNM. Nevertheless, the genes implicated in LNM in PTC may vary depending on the tumor diameter. Hence, it is crucial to investigate the specific genes associated with LNM in PTMC. Li et al. identified 20 DEGs in the group of patients with LNMs (N1) and those without (N0) in PTMC. Among these DEGs, FN1 demonstrated good performance in differentiating between N1 and N0 patients, with an AUC value of 0.690.^[15] The study carried out by Lin et al. revealed that interferon-stimulated gene 15 was highly expressed in PTMC with LNM (P < 0.001).^[16] However, these studies either exhibited insufficient prediction accuracy or had a limited validation of cases, and have not explicitly focused on low-risk PTMC, which holds the utmost therapeutic significance.

The current study explored the genes that affect LNM through the application of bioinformatics analysis, immunohistochemistry, and immunocytochemistry. Overall, a comprehensive analysis identified a total of 292 DEGs, and the top 10 hub genes were uncovered from a dataset of 122 instances of pT1 PTCs. Further immunohistochemical and immunocytochemical validation of the primary hub genes demonstrated notable disparities in CK5/6 between the PTMC group with CLNM and those without LNM. During this analytical process, the Chi-square test was adopted to evaluate the disparities in immunohistochemical examination levels between PTMC patients with and without CLNM. When conducting the test, either the Pearson Chi-square test or the continuity-corrected Chi-square test was selected based on the total number of samples and the theoretical frequency, thereby revealing the differences among the groups. This study fully acknowledges the potential interference of confounding factors on CK5/6 expression and CLNM development. Therefore, in terms of patient characteristics, factors such as age, gender, and tumor size that might affect the research results were taken into account. During the sample screening stage, the propensity score matching method was utilized to accurately match PTMC patients with and without CLNM according to age, gender, and tumor diameter. The SMD between groups was stringently controlled below 0.1, effectively diminishing the confounding effects of these factors and enhancing the comparability between groups. This guarantees that the differences in CK5/6 expression are attributed to the CLNM status rather than the disparities in patients’ basic characteristics, all of which have laid a relatively solid statistical foundation for the research conclusions. It is noteworthy that the cases utilized for verification were those that fulfilled the criteria of low-risk PTMC, with the exception of the status regarding LNM. From a therapeutic perspective, the selection of these cases is of utmost clinical significance, as the presence or absence of LNM in these cases will lead to significantly different treatment strategies.

CK5/6 is a frequently utilized antibody in the field of clinical pathology. It is predominantly detected in squamous epithelium, ductal epithelial basal cells, myoepithelial cells, and mesothelial cells in normal tissues. In thyroid cancer, CK5/6 can be employed to assist in the diagnosis of thyroid squamous cell carcinoma and can also be used to diagnose the squamous cell component in PTC, although the latter is less used in clinical practice due to the paucity of literature documenting the role of the squamous component in PTC.

Squamous cell carcinoma of the thyroid is postulated to have originated from PTC, the former is associated with a poorer prognosis and a higher incidence of LNMs.^[1,17] Consequently, the presence of a squamous cell component in PTC may imply an elevated likelihood of LNM, a hypothesis that is further supported by some studies.^[18] Interestingly, there are reports suggesting that in basal-like breast cancer cell lines, CK5 knockdown resulted in a reduction in LNMs in vivo compared to control cells,^[19] although this was not directly observed in PTC. In the context of thyroid cancer, despite the dearth of direct evidence, it is plausible that CK5/6 could be implicated in processes such as epithelial-mesenchymal transition (EMT).^[19] EMT is a crucial process in cancer metastasis, during which epithelial cells acquire mesenchymal traits, endowing them with the ability to migrate and invade neighboring tissues. CK5/6 may potentially contribute to maintaining the epithelial phenotype or modulating EMT-related signaling pathways, thereby exerting an influence on the metastatic potential of PTMC cells. Moreover, CK5/6 has been proposed as a possible stem cell marker.^[20] Stem cells are known to interact with other molecules or signaling cascades involved in lymphangiogenesis,^[21] the formation of new lymphatic vessels that facilitate the spread of cancer cells to regional lymph nodes. These investigations and hypotheses have, to some extent, demonstrated that CK5/6 is capable of predicting LNM in PTMC, thereby validating the reliability of our research findings.

The novelties of this study are summarized as follows: (1) This study represents the first instance of using commonly employed antibodies in clinical practice to predict the presence of CLNM in PTMC. CK5/6 is a widely utilized antibody in daily pathology work. The findings of this study can be promptly translated into clinical practice to facilitate the prediction of the status of PTMC central lymph nodes. (2) For cases that underwent immunohistochemical and immunocytochemical verification, we employed strict inclusion criteria and propensity score matching to select 208 PTMC cases from 2091 cases. These cases fulfilled the criteria for low-risk PTMC except for LNM.

This study exhibits the following limitations: (1) This study is a single-center study, and multicenter investigations are warranted for further validation. (2) The bioinformatics analysis was conducted using mRNA sequencing data of pT1 PTCs (≤2 cm), while immunohistochemical and immunocytochemical staining validation was performed on specimens of PTMC patients (≤1 cm). Owing to the absence of maximum diameter information about the tumor in the TCGA database, we could only select pT1 PTC in the bioinformatics analysis. In immunohistochemical and immunocytochemical verification, we selected PTMC, which is of greater clinical relevance. (3) There may be other confounding factors that could have affected CK5/6 expression or CLNM development. For example, patient-related factors were not thoroughly examined in relation to CK5/6 expression. Moreover, the immunohistochemical and immunocytochemical staining methods possess inherent limitations, such as inter-observer variability in the assessment of staining intensity and percentage of positive cells. Despite using two experienced pathologists and a standardized scoring system, a certain level of subjectivity may still exist. (4) The study did not investigate the functional consequences of CK5/6 expression in PTMC cells, which could yield further insights into its role in LNM.

Looking ahead to future research directions in the field of clinical medicine, actively promoting multi-center and prospective studies holds profound significance. By extensively incorporating samples from diverse regions and various clinical backgrounds, the universality and stability of the conclusions drawn from this study can be comprehensively verified across a wide range of medical settings. This will facilitate an in-depth analysis of the precise efficacy and long-term prognostic value of CK5/6 as a biomarker in clinical decision-making, thereby laying a solid empirical foundation for formulating unified and efficient clinical guidelines. In the realm of basic medicine, focusing on CK5/6 gene knockout and rigorous validation at the cell line level is of crucial importance. Through the application of precise gene editing techniques to construct knockout models, an in-depth exploration into the impact mechanisms of CK5/6 gene deletion on key tumor biological behaviors such as cell proliferation, migration, invasion, and EMT can be conducted. This will elucidate its functional role in the lymphatic metastasis process of papillary thyroid microcarcinoma from the molecular origin, providing robust support for macroscopic clinical findings through microscopic mechanism analysis at the cellular level. Thus, a closely intertwined and synergistic research framework between basic and clinical research will be formed, opening up new avenues and injecting innovative impetus into the conquest of lymphatic metastasis in thyroid cancer.

The findings of this study harbor potential clinical implications. CK5/6 could be integrated into existing diagnostic protocols for PTMC. For instance, in cN0 PTMC patients, CK5/6 immunohistochemical staining of preoperative thyroid FNA samples could be performed. A positive CK5/6 expression might initiate further evaluation, such as a more detailed preoperative imaging study, or entail consideration of a more aggressive treatment option, such as prophylactic central lymph node dissection.

In comparison to established markers like BRAF or RET/PTC mutations linked to CLNM in thyroid cancer, CK5/6 offers a unique perspective. The combination of CK5/6 analysis with these genetic markers may provide a more comprehensive evaluation of the CLNM risk in thyroid cancer patients. When combined with other biomarkers, CK5/6 can augment the accuracy of CLNM prediction. Future studies could investigate the development of a biomarker panel comprising CK5/6 and other relevant markers to enhance PTMC patient risk stratification.

SUMMARY

This study employed bioinformatics analysis to detect hub genes and subsequently performed immunohistochemical and immunocytochemical staining of these genes in PTMC. CK5/6 has been established as a potential immunohistochemical and immunocytochemical biomarker for distinguishing PTMC with CLNM from PTMC without LNM. Future research with larger scale, prospective design, and in-depth exploration at the cell line level will contribute to further enhancing the credibility of research conclusions.

AVAILABILITY OF DATA AND MATERIALS

The data analyzed in this study can be accessed through the corresponding author upon reasonable request.

ABBREVIATIONS

CgA: Chromogranin A

CK5/6: Cytokeratin 5/

CLNM: Central lymph node metastasis

DEG: Differentially expressed gene

FDR: False discovery rate

FNA: Fine needle aspiration

ICC: Immunocytochemistry

IHC: Immunohistochemistry

LNM: Lymph node metastasis

mRNA: Messenger ribonucleic acid

Pax-2: Pair box gene 2

PPI: Protein-protein interaction

pT1: Pathological tumor stage 1

PTC: Papillary thyroid carcinoma

PTMC: Papillary thyroid microcarcinoma

SMD: Standardized mean difference

STRING: Search tool for the retrieval of interacting genes

TCGA: The cancer genome atlas