A combined biomarker panel shows improved sensitivity and specificity for detection of ovarian cancer

Abstract Background Combined biomarkers can improve the sensitivity and specificity of ovarian cancer (OC) diagnosis and effectively predict patient prognosis. This study explored the diagnostic and prognostic values of serum CCL18 and CXCL1 antigens combined with C1D, FXR1, ZNF573, and TM4SF1 autoantibodies in OC. Methods CCL18 and CXCL1 monoclonal antibodies and C1D, FXR1, ZNF573, and TM4SF1 antigens were coated with microspheres. Logistic regression was used to construct a serum antigen‐antibody combined detection model; receiver‐operating characteristic curve (ROC) was used to evaluate the diagnostic efficacy of the model; and the Kaplan‐Meier method and Cox regression models were used for survival analysis to evaluate the prognosis of OC. Data from The Cancer Genome Atlas (TCGA) and Genotype‐Tissue Expression (GTEx) projects and online survival analysis tools were used to evaluate prognostic genes for OC. The CIBERSORT immune score was used to explore the factors influencing prognosis and their relationship with tumor‐infiltrating immune cells. Results The levels of each index in the blood samples of patients with OC were higher than those of the other groups. The combined detection model has higher specificity and sensitivity in the diagnosis of OC, and its diagnostic efficiency is better than that of CA125 alone and diagnosing other malignant tumors. CCL18 and TM4SF1 may be factors affecting the prognosis of OC, and CCL18 may be related to immune‐infiltrating cells. Conclusions The serum antigen‐antibody combined detection model established in this study has high sensitivity and specificity for the diagnosis of OC.


| INTRODUC TI ON
Ovarian cancer (OC) is one of the most frequent malignant diseases that seriously threatens women's life and health. Its incidence ranks third among malignant tumors of the female reproductive tract. 1 However, the mortality rate of OC ranks first. 2 In 2019, there were about 21,750 new cases of OC in the United States, with 13,940 deaths and a mortality rate exceeding 60%. 3 Due to the hidden incidence of OC, the lack of typical clinical symptoms, and early diagnosis methods, about 70% of OC patients are already in the middle and advanced stages when diagnosed. [4][5][6] Among the current technical approaches for non-invasive diagnosis of OC, a pelvic examination is not sufficiently sensitive to detect ovarian masses, and the level of serum tumor marker CA125 is elevated in 90% of patients with advanced disease, but in only 50% of patients with stage I tumors. 5 Therefore, improving the early detection rate of OC and screening out factors that influence the prognosis of OC is critical. At present, no suitable biomarkers that can be used for early diagnosis, curative effect detection, and prognostic assessment of OC have been identified. [7][8][9][10][11] Current studies have shown that combining multiple biomarkers can not only improve the sensitivity and specificity of early diagnosis of OC but also predict the choice of effective treatment methods and prognosis. [12][13][14] With the progress of tumor immunotherapy, the correlation between immunity and the tumor has received considerable attention.
The level of immune cell infiltration in the tumor is associated with tumor growth, progression, and patient outcome and has become the focus of research in recent years. Some scholars have proposed a method to calculate the composition of immune cells from the gene expression profile of complex tissues. This method has been verified by flow cytometry in colorectal cancer, lung cancer, and follicular lymphoma and can be used in large-scale analysis of gene expression profiles. 15,16 Serological analysis of recombinantly expressed cDNA clone (SEREX) technology was performed to identify serum IgG autoantibodies C1D, TM4SF1, FXR1, and ZNF573 that are significant for the diagnosis of OC, 17 and the surface-enhanced laser desorption/ ionization-time of flight mass spectrometry (SELDI-TOF-MS) technology was used to identify serum antigens CCL18 and CXCL1 for OC. 18 Through the application of a liquid suspension chip detection system, a combined detection method of serum CCL18 and CXCL1 antigens and C1D, TM4SF1, FXR1, and ZNF573IgG autoantibodies for the diagnosis of OC was successfully established. Its sensitivity and specificity were higher compared with the conventional enzymelinked immunosorbent assay (ELISA) method. 19,20 It was concluded that the diagnostic model with combined detection of serum antigen and antibody is better than serum CA125 in the diagnosis of OC. 21 A model for diagnosing ovarian malignant tumors was successfully constructed and validated by conducting preliminary experiments.
However, no effective indicators were found to predict the survival and prognosis of OC. In addition, because the sample size and the number of samples included in the study were not effective, and there were fewer samples from other malignant tumor groups, we expanded the sample size and the types of malignant tumor control groups. The combined detection model of serum antigen and antibody was used to diagnose OC and verify other malignant tumors to compare their diagnostic efficiency. Moreover, the overall survival (OS) of patients with OC was evaluated. The differences in the expression of the six genes in normal ovarian tissue and OC tissue were analyzed based on the Cancer Genome Atlas (TCGA) and the Genotype-Tissue Expression (GTEx) databases. The relationship between the expression of these genes in OC tissues and the OS time was also evaluated. Finally, the immune score of the CIBERSORT algorithm was used to further analyze the relationship between the established prognostic indicators and infiltrating immune cells.

| Detection method and model building
In this experiment, the liquid chip technology was prepared by a mixed suspension of fluorescent coded microspheres. After mixing the coded microspheres for different detection substances, a small amount of the test sample was added. The target and cross-linked molecules on the surface of microspheres precisely bind in the suspension. The two-color laser is employed to detect red classification fluorescence on microspheres and green reporter fluorescence on the reporter molecule, allowing for the determination of the type and quantity of bound detection substance.

| Follow-up time
One hundred fifty OC patients from January 2014 to June 2018 whose serum samples were available were followed up by tele-

| CIBERSORT score
The

| Comparison of serum levels of CCL18 and CXCL1 antigens, and C1D, TM4SF1, FXR1, and ZNF573 IgG autoantibodies
We have listed the detection process of fluorescence analysis technology ( Figure 1A). A dose-response standard curve of was drawn to detect serum CCL18 and CXCL1 antigens ( Figure 1B), as well as a dose-response standard curve of C1D, TM4SF1, FXR1, and ZNF573 IgG protein to detect serum IgG autoantibodies ( Figure 1C). After quantitative calculation, the levels of serum CCL18 and CXCL1 antigens, and C1D, TM4SF1, FXR1, and ZNF573 IgG autoantibodies were significantly higher in patients with OC than in those with cervical cancer, liver cancer, breast cancer, gynecological benign tumor patients, and healthy women ( Figure 2).

| Comparison of the efficacy of individual markers and combined detection models in the diagnosis of OC
Patients with OC (n = 300) were regarded as the positive group and which was higher than the AUC for detection of OC using CA125 alone (0.818, p < 0.001) ( Figure 3B).

| Comparison of diagnostic efficacy of combined detection for ovarian cancer and other cancers
The AUC of serum antigen-antibody combined detection of OC (n = 300) was also higher than that of the other malignant tumors, The results showed that the positive rate, positive predictive value, and positive likelihood ratio of this model were higher for the diagnosis of OC than cervical cancer, breast cancer, and liver cancer (p < 0.001) ( Table 2).

| Comparison of the diagnostic efficacy of serum antigen-antibody combined detection model and CA125 alone in the detection of early-stage OC (stage Ⅰ-Ⅱ)
The  The results indicated that the combined detection of serum antigen and antibody was significantly better than CA125 alone in the diagnosis of early-stage OC (p < 0.001).

| Relationship between the expression of CCL18 and TM4SF1 and immune cell infiltration
The 374 cases of OC tissues obtained from TCGA were divided into high-expression groups and low-expression groups according to the median of CCL18 and TM4SF1 expression as the cut-off value. The CIBERSORT algorithm was utilized to analyze the difference in the distribution of tumor-infiltrating immune cells between the two groups (Figure 7). There was substantial infiltration of M2 macrophages (0.297 ± 0.12%), resting memory CD4+ T cells

| DISCUSS ION
Ovarian cancer is the fifth most common cause of cancer-related deaths in women. 23 The high mortality rate may be due to the insidi- and regulate the immune system to promote tumor development. 24 Therefore, inflammatory factors could be used as novel diagnostic and predictive biomarkers for OC. Preliminary studies in our laboratory indicated that the combined detection of CCL18 and CXCL1 chemokines and C1D, FXR1, TM4SF1, and ZNF573 autoantibodies has great potential for early diagnosis of OC. 20,21 Nuclear nucleic acid-binding protein C1D plays a role in multiple cellular processes, transcriptional regulation, genome stability monitoring, DNA repair, and RNA processing, all of which are necessary to maintain the life cycle of the host. 25 When the tumor microenvironment changes, C1D induces the production of autoantibodies in the serum of patients with epithelial ovarian cancer (EOC). The production of autoantibodies may be due to the ectopic expression of C1D, which stimulates the immune system. 26,27 Fragile X mental retardation syndrome-related protein 1 (FXR1) has been confirmed to be amplified in lung cancer, breast cancer, head and neck cancer, and OC. 28  showed that patients with a higher ratio of M1/M2 macrophages had a better prognosis, indicating that macrophages play an important role in the progression of OC. 45 In the current study, we used the CIBERSORT algorithm to determine the proportion of infiltrating immune cells and found higher infiltration of CD8+ T cells and M1 type macrophages in the ovarian serous cystadenocarcinoma tissues in TCGA data in the high CCL18 expression group compared with their corresponding low-expression groups. This may be because the prognosis of the high CCL18 expression group appeared slightly better than that of the low CCL18 expression group. Based on findings from previous studies and our observations on the improved prognosis of OC in the high-expressing CCL18 group and the low-expressing CCL18 group, it is reasonable to assume that the expression of CCL18 in the tumor environment may attract and activate immune cells, thereby enhancing the immune response against malignant tumors. Therefore, CCL18 may also be an independent prognostic factor for survival in OC.
Although TM4SF1 may be a key factor influencing the prognosis of OC in our study, Yang and colleagues found that the positive expression of TM4SF1 protein was not an independent factor for prognosis. 34 Qiang et al. 46 showed that the expression of TM4SF1 in colorectal cancer tissues was significantly higher than that in In summary, combined detection of CCL18 and CXCL1 chemokines, and C1D, TM4SF1, FXR1, and ZNF573 autoantibodies can improve the specificity and sensitivity of OC diagnosis, and its diagnostic efficiency is higher than that of other malignant tumors.
CCL18 may serve as a novel biomarker for the early diagnosis and prognosis of OC. Although these possibilities require further verification, they still provide new avenues for the future clinical diagnosis and prognosis of OC.

CO N FLI C T O F I NTE R E S T
The authors declare no potential conflicts of interest regarding research, authorship, and/or publication of this article.

E TH I C S S TATEM ENT
The ethical review was approved by the Institute Ethics Committee of Guangxi Medical University Cancer Hospital (LW2021094) and individual consent for this retrospective analysis was waived.

DATA AVA I L A B I L I T Y S TAT E M E N T
The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.