Complement C4-A and Plasminogen as Potential Biomarkers for Prediction of Papillary Thyroid Carcinoma

Background Early diagnosis and therapy of papillary thyroid carcinoma (PTC) is essential for reducing recurrence and improving the long-term survival. In this study, we aimed to investigate the proteome profile of plasma and screen unique proteins which could be used as a biomarker for predicting PTC. Methods Serum samples were collected from 29 PTC patients and 29 nodular goiter (NG) patients. Five PTC serum samples and five NG serum samples were selected for proteome profiles by proteomics. Eight proteins in PTC and NG serum samples were selected for confirmation by enzyme-linked immunosorbent assay analysis. Receiver operating characteristic curves was used to evaluate the diagnostic value of potential biomarkers. Results Complement C4-A (C4A) and plasminogen (PLG) were significantly lower in serum samples of PTC patients compared with NG patients. C4A was observed to have excellent diagnostic accuracy for PTC, with a sensitivity of 91.67% and specificity of 83.33%. The diagnostic value of PLG for PTC was demonstrated by a sensitivity at 87.50% and specificity at 75.00%. The AUC for C4A and PLG was 0.97 ± 0.02 and 0.89 ± 0.05. Conclusion C4A and PLG appeared to be excellent potential biomarkers for the prediction of PTC.


INTRODUCTION
Thyroid carcinoma is the most common endocrine malignancy with increasing incidences worldwide (1). Epidemiologic studies have shown that thyroid cancer is responsible for 567,233 new cases and 41,071 deaths in 2018 worldwide (2). The disease may rank as the third leading cancer in women (3). Papillary thyroid carcinoma (PTC) comprises more than 80% of thyroid cancer. Cervical lymph node metastases and aggressive subset are related to the risk of recurrence or death (4,5). The etiology of PTC is not well clarified, and its occurrence is unpredictable. Early and accurate diagnosis of PTC is essential for prevention of progression and recurrence.
Thyroid sonography and fine-needle aspiration biopsy (FNAB) are used for differentiating benign and malignant thyroid lesion. However, FNAB is invasive and may yield 20% indeterminate cytology (6). Many molecular testing techniques have been made to aid in the diagnosis of PTC, such as ThyroSeq mutational panel, the Afirma gene expression classifier, immunohistochemical stains, and proteomic analysis (7,8). However, these tests are based on biopsy samples and incapable of timely monitoring the occurrence of PTC. Therefore, accurate and timely biomarkers of serum protein are required to predicting PTC. With respect to serum protein for distinguishing PTC, Hu et al. (9) showed that serum vascular adhesion protein-1 with a 66.7% sensitivity and 77.4% specificity was profoundly downregulated in thyroid cancer group which included PTC and follicular thyroid carcinoma. Lu et al. (10) reported that serum complement C4-A/B increased in PTC patients by using weak cation exchange magnetic beads fractionation followed by matrix-assisted laser desorption/ ionization-time-of-flight mass spectrometry. However, the specificity and sensitivity for diagnosis is not high enough.
We aimed to screen for novel serum proteomic biomarkers for predicting PTC. Through screening assay using proteomics and enzyme-linked immunosorbent assay analysis (ELISA), we successfully identified that serum C4A and plasminogen (PLG) were proteins that met these requirements.

Study Population
A total of 29 PTC patients and 29 nodular goiter (NG) patients were enrolled in this study between September 2018 and July 2019. Twenty-nine PTC patients included 27 patients diagnosed as primary thyroid malignancy and two as suspicious for malignancy according to cytology result of FNAB. Twentynine NG patients with a maximum diameter of larger than 4 cm estimated by ultrasonography underwent thyroidectomy based on current Chinese guidelines. Pathological diagnosis with PTC or NG was confirmed by the histopathological result of resected specimens. We excluded those patients combined with other cancer types. The clinical characteristics and demographic information were collected from medical records, such as age, gender, tumor node metastasis (TNM) status, and multiplicity of tumor. The baseline characteristics of PTC cases and NG cases are presented in Tables 1, 2. We excluded patients with comorbid conditions such as other forms of cancer and inflammatory diseases. The present study was approved by the ethics committee of the West China Hospital of Sichuan University, and all patients enrolled in the study signed written informed consent.

Sample Collection
Five-milliliter peripheral venous blood sample was collected prior to any therapeutic intervention by EDTA-containing vacutainer tube. Samples were subjected to centrifugation at 1,600 rpm and 4°C for 10 min. The supernatant was aliquoted and stored at −80°C until use.

Screening Assay Using Proteomics
Five PTC serum samples and five NG serum samples were used to perform the screening assay by using proteomics. The proteomics were performed by Shanghai OE Biotech. Co., Ltd. according to the manufacturer's suggested instruction. Briefly, 10 µl serum sample was added to the resin slurry in the column, then incubated with gentle shaking for 60 min. Column was added to a 2-ml collection tube and centrifuge at 1,000×g for 2 min. The filtrate was collected for further processing. Protein concentration was determined using a bicinchonininc acid protein assay (11). The protein was digested into peptides using filter-aided sample preparation. The digested peptides were desalted using C18-Reverse-Phase SPE Column. Liquid chromatography (LC)-data-dependent acquisition (DDA)/dataindependent acquisition (DIA)-mass spectrometry (MS)/MS was performed by a QE mass spectrometer (Thermo, Waltham, MA, USA) equipped with an Easyspray source (Thermo, USA). The LC-MS/MS raw data were imported in Maxquant for labeling free quantification analysis. RAW files were imported into Spectronaut pulsar X to generate the spectral library. The DIA data were analyzed with Spectronaut Pulsar X.

Statistical Analysis
Continuous variables with a normal distribution were expressed as the mean ± standard deviation (SD). All statistical analyses were assessed by the Student's t-test using GraphPad Software Prism 5 (San Diego, CA, USA). A value of p < 0.05 was considered statistically significant. The sensitivity and specificity of C4A and PLG for diagnosis of PTC was evaluated by receiver operating characteristic (ROC) curve and the area under the curve (AUC). The ROC curves with an AUC ≥0.8 demonstrate good discriminatory power.

Screening of Serum Samples
The serum protein spectrum in five PTC and five NG patients were screened using proteomics. The analysis showed that four proteins were upregulated while eight proteins were downregulated in PTC group compared with NG group ( Table 3). Among them, we selected eight proteins for further confirmation analysis, including four upregulated and four downregulated with p-value ≤0.01.

Selected Protein Levels in PTC and NG Plasma
The levels of VAT1, SERPINA10, APOB, CFI, CD93, C4A, ECM1, and PLG were analyzed by ELISA in 24 PTC samples and 24 NG samples. We found that VAT1, SERPINA10, APOB, and CFI exhibited no significant differences between the PTC and NG samples ( Figures 1A-D). As shown in Figures 1E-H, the expression levels of CD93, C4A, ECM1, and PLG in the PTC group were significantly decreased compared with the NG group (p < 0.05).

Diagnostic Value of Serum C4A and PLG for PTC
Based on the quantitative results, the diagnostic values of serum C4A and PLG for PTC was evaluated by ROC curve. The cutoff values were chosen by considering the maximum sensitivity and specificity for the diagnosis of PTC. As shown in Figure 2 and Table 4, the AUC for C4A was 0.97 ± 0.02. When the cutoff was set at 10,884 ng/ml, the sensitivity of C4A was 91.67% and the specificity reached 83.33% (Figure 2A). In addition, the AUC for PLG was 0.89 ± 0.05. PLG was also observed to have an excellent diagnostic accuracy for PTC with a respective sensitivity and specificity of 87.50% and 75% for a cutoff value of 225.2 mg/ ml ( Figure 2B).

DISCUSSION
To the best of our knowledge, this is the first study investigating the expression levels of serum C4A and PLG in PTC and NG plasma by proteomics and ELISA. Based on our study, we found C4A and PLG levels were significantly lower in the PTC group compared with the NG group. C4A and PLG were observed to have an excellent diagnostic accuracy for PTC with high sensitivity and specificity. These findings implied that serum C4A and PLG may be potential biomarkers for predicting PTC. It is known that FNAB is frequently used in clinics for evaluating thyroid nodules.  (14). However, the subjects included were thyroid FNAB specimen. This biopsy is an invasive procedure. In the present study, peripheral venous blood sample were examined via proteomics and ELISA analysis. C4A was observed to have excellent diagnostic accuracy for PTC, with a respective sensitivity and specificity of 91.67% and 83.33%. The sensitivity and specificity of PLG was 87.50% and 75.00% for PTC diagnosis, respectively. C4A is one of inflammatory mediators which plays a role in the formation of immune complexes and regulation of binding of immune aggregates (15,16). In addition, C4 was observed high expression in the thyroid gland compared with the heart, pancreas, and thymus (17). Several studies have surveyed that abnormal C4A expression is associated with some cancers. Tikhonov (10). In contrast, we found that serum C4A in PTC group was significantly decreased with high sensitivity of 91.67% and specificity of 83.33% compared with the NG group. This result might be attributed to the different comparison. However, the mechanism of C4A affecting the tumorigenesis of PTC has not been demonstrated. Further mechanistic analysis is required to understand the biological role of C4A. PLG, a 92-94-kilodalton protein produced in the liver and circulated in the blood, is known to be involved in the degradation of extracellular matrix, cell migration, angiogenesis, oncogenesis, and metastasis (20,21). The PLG is associated autophagy of cancer (22). Fang et al. demonstrated that Plasminogen kringle 5 suppressed the growth gastric cancer by inhibiting angiogenesis and apoptosis (23). Yang et al. found that plasminogen kringle 5 can inhibit hepatocellular carcinoma via antiangiogenic activity (24). Tykhomyrov et al. reported that PLGtreated lung adenocarcinoma cells exhibited autophagy induction by upregulation of beclin-1 levels (25). In the present study, we observed that PTC group showed a great downregulation of PLG compared with the NG group, providing insight into the association between PLG and PTC. However, the mechanism by which PLG affects PTC is unknown. Further studies are warranted to exemplify.
Of course, we have to admit several limitations in present study. First, a small number of PTC and NG samples were included in the study. Second, due to financial constraints, we did not investigate serum protein in healthy controls. Third, we have not analyzed how C4A and PLG are involved in tumorigenesis of PTC. Therefore, well-designed trials with larger samples are required to confirm that C4A and PLG may be potential biomarkers for the diagnosis of PTC.
In conclusion, this study reveals that serum C4A and PLG may be potentially useful biomarkers for the prediction of PTC. C4A and PLG with a high sensitivity and specificity appear to be developed into a bedside test for rapidly predicting PTC.

DATA AVAILABILITY STATEMENT
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: Proteomexchange [accession: number:PXD029304].

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by the ethics committee of the West China Hospital of Sichuan University. The patients/participants provided their written informed consent to participate in this study.

AUTHOR CONTRIBUTIONS
ZL, JZ, and TW designed the research. YW, SZ, and DW performed the experiments and carried out statistical analysis. YW drafted the article. All authors contributed to the article and approved the submitted version.