Assessment of Seven Clinical Tumor Markers in Diagnosis of Non-Small-Cell Lung Cancer

Background The correlation between tumor markers (TM) and TNM stage of non-small-cell lung cancer (NSCLC) has not been widely reported. Methods TM levels (CEA, CA125, CA15-3, CA19-9, CA72-4, CYFRA21-1, and SCC-Ag) in 424 cases of lung adenocarcinoma (LAC), 166 cases of lung squamous cell carcinoma (LSCC), and 103 cases of benign chest disease (BCD) were analyzed before treatment. Results By Kendall's tau-b correlation analysis, CEA, CA125, CA15-3, CA19-9, CA72-4, CYFRA21-1, and SCC-Ag were correlated with T stage of LAC (r = 0.235, p < 0.05; r = 0.298, p < 0.05; r = 0.254, p < 0.05; r = 0.063, p < 0.05; r = 0.080, p < 0.05; r = 0.268, p < 0.05; and r = 0.080, p < 0.05). CEA, CA125, CA15-3, CA19-9, CA72-4, and CYFRA21-1 were correlated with N stage of LAC (r = 0.356, p < 0.05; r = 0.415, p < 0.05; r = 0.340, p < 0.05; r = 0.117, p < 0.05; r = 0.175, p < 0.05; and r = 0.351, p < 0.05). CEA, CA125, CA15-3, CA19-9, CA72-4, and CYFRA21-1 were correlated with M stage of LAC (r = 0.365, p < 0.05; r = 0.353, p < 0.05; r = 0.293, p < 0.05; r = 0.135, p < 0.05; r = 0.169, p < 0.05; and r = 0.312, p < 0.05). CA125, CYFRA21-1, and SCC-Ag were correlated with T stage of LSCC (r = 0.202, p < 0.05; r = 0.233, p < 0.05; and r = 0.099, p < 0.05). CA125 and CYFRA21-1 were correlated with N stage of LSCC (r = 0.178, p < 0.05 and r = 0.284, p < 0.05). CA125, CA15-3, and CYFRA21-1 were correlated with M stage of LSCC (r = 0.214, p < 0.05; r = 0.152, p < 0.05; and r = 0.213, p < 0.05). Combining hazard ratio, AUC, sensitivity, specificity, NPV, and PPV, it was concluded that CEA and CYFRA21-1were the most related TM of LAC. SCC-Ag and CYFRA21-1 were the most related TM of LSCC. Conclusions CEA combined with CYFRA21-1 contributed to auxiliary diagnosis of LAC. CYFRA21-1 combined with SCC-Ag contributed to auxiliary diagnosis of LSCC. CEA, CA125, CA15-3, CA19-9, CA72-4, and CYFRA21-1 were correlated with primary tissue and metastasis of LAC. CA125 and CYFRA21-1 were correlated with primary tissue and metastasis of LSCC.


Introduction
Data released by the National Cancer Registry (NCCR) of China in 2018-2015 showed that primary lung cancer was the first cancer in morbidity and mortality for four consecutive years [1][2][3]. The diagnosis of early-stage non-small-cell lung cancer (NSCLC) was the first step toward successful clinical therapy and increased patient survival. Making full use of the existing means to diagnose early NSCLC has been concerned. Histopathological examination is the gold standard for the diagnosis of NSCLC. However, due to the invasiveness of histopathological examination, chest computed tomography and tumor markers (TM) test preceded histopathological examination. The TM test results were increasingly influencing decisions on tumor screening, initial treatment, and follow-up for NSCLC based on the advantages of minimal trauma, good repeatability, simplicity, and rapidity. TM can be used to assist diagnosis and differential diagnosis and to understand the possible pathological types of NSCLC. In the clinical diagnosis of NSCLC, carcinoembryonic antigen (CEA), cytokeratin 19 fragments antigen (CYFRA21-1), and squamous cell carcinoma antigen (SCC-Ag) have been used as reference TM for NSCLC. Other common TM also increased in NSCLC, including carbohydrate antigen 125 (CA125), CA15-3, CA19-9, and CA72-4. Previous studies have shown that TM level was correlated with pathological types, primary tissues [4], lymph node metastasis [5,6], and distant metastasis [7]. Combination of TM can improve the diagnostic accuracy of NSCLC [8][9][10]. However, the points of view on combination of TM were not consistent. The correlation between TM and TNM stage of NSCLC has not been widely reported. Our research came from daily clinical practice. The TM levels came from before treatment since TM vary significantly before and after treatment [11,12]. We chose benign chest diseases (BCD) that were easily confused with NSCLC as a control. We systematically discussed the characteristics of seven TM in pathological type, TNM stage, and diagnosis of NSCLC.

Materials and Methods
2.1. Patients. This study was approved by the ethics committee of the Guangxi Medical University Affiliated Tumor Hospital with an ethics approval number of LW2018011. When the patients suspected they had lung cancer, they were first admitted to the thoracic tumor surgery department. We systematically reviewed all patients who had been hospitalized in the thoracic tumor surgery department from October 2014 to August 2017. Cases with other malignancies and/or no pathology and/or no image were excluded. NSCLC and BCD were all first diagnosed. The diagnosis of NSCLC was established by histopathology. NSCLC was staged according to the 2009 seventh edition of the tumor-nodes-metastasis (TNM) classification for lung cancer [13][14][15]. The diagnosis of BCD was established by histopathology or pathogeny. BCD cases were not treated with any therapy. NSCLC cases had not been treated with any antitumor therapy, such as surgery, chemotherapy, radiotherapy, biological therapy, endocrine therapy, Chinese medicine treatment, hyperthermia, and radiofrequency ablation therapy. Complete and detailed case information were available for all cases. A total of 424 cases of lung adenocarcinoma (LAC), 166 cases of lung squamous cell carcinoma (LSCC), and 103 BCD patients were included in this study (Table 1).

Statistical
Analysis. IBM SPSS21.0 and MedCalc 18.2 software were applied for statistical analysis. The levels adopted for significance were p < 0 05 (two tailed). The calculation of sample content: n = μ ɑ 2 p(1-p)/δ 2 , the sample size of the case group and the control group were all 96, and the sample size of the study met the requirements. The normality test used skewness coefficient (sk) and kurtosis coefficient (ku). The TM data was not normal distribution. Mann-Whitney U test of nonparametric rank sum test was used to compare the median of two groups. Chi-square test (χ 2 ) was used to compare the positive rates of two groups. The correlation between TM level and TNM stage of NSCLC was analyzed by Kendall's tau-b correlation analysis. The levels of TM may be influenced by T stage, N stage, M stage, and other factors, so the correlation coefficient was low. In this paper, the correlation analysis was based on p < 0 05 not the correlation coefficient. With BCD as a control, binary logistic regression was used to calculate the hazard ratios of TM in NSCLC. Applying the receiver operating characteristic curve (ROC curve) to calculate the area under the curve (AUC) of TM in NSCLC, z test was used to compare AUC. Sensitivity, specificity, negative predictive value (NPV), and positive predictive value (PPV) were calculated.

Patients' Characteristics.
A total of 590 NSCLC patients and 103 BCD patients were included in the study. Table 1 summarizes patients' characteristics. Of note, the proportion of advanced NSCLC was higher than that of early NSCLC at initial diagnosis (Table 1).

Discussions
TM was also expressed in BCD. Increased TM in BCD may be correlated with inflammation and disease severity [16]. Seven TM in BCD are out of the normal range to varying degrees. In this study, the parallel positive rate of seven TM in BCD was 41.7% and the tandem positive rate was 55.3%. The parallel positive rate of seven TM in LAC was 86.5% and the tandem positive rate was 245.3%. The parallel positive rate of seven TM in LSCC was 93.4% and the tandem positive rate was 212.1%. It was shown that TM in BCD mainly increased by one, and that in NSCLC mainly increased by two or more. The median of seven TM in BCD was within the normal range. The median of CEA and CYFRA21-1 in LAC and the median of CYFRA21-1 and SCC-Ag in LSCC had approached or exceeded the normal range. The median and positive rates of CEA, CA125, CA15-3, CA19-9, and CYFRA21-1 in LAC were higher than those in BCD (p < 0 05). The median and positive rates of CEA, CA125, CYFRA21-1, and SCC-Ag in LSCC were higher than those in BCD (p < 0 05). It was shown that TM was lowly expressed in BCD and highly expressed in NSCLC. The expression of CEA in primary tissue and lymph node metastasis of LAC was higher than that in other NSCLC types [17]. CEA was highly expressed in nonlepidic dominant histologic subtype [18], adenocarcinoma tissues coexisting with bullae or honeycomb cysts [19] and welldifferentiated adenocarcinoma tissues [20]. The expression of CEA was also correlated with TNM stage of NSCLC [21,22]. CEA was correlated with T stage, N stage, and M stage of LAC (r = 0 235, p < 0 05, r = 0 356, p < 0 05, and r = 0 365, p < 0 05). It was shown that the level of CEA was correlated with primary tissue, lymphatic metastasis, and distant metastasis of LAC. However, CEA was not correlated with TNM stage of LSCC (p > 0 05). The correlation between CA19-9, CA72-4, and TNM stage in LAC and LSCC were similar to that of CEA; the correlation coefficients were lower. CA19-9 increased in patients with infectious lung disease [23]. NSCLC, especially advanced NSCLC, was often accompanied by severe infection. Therefore, CA19-9 can help identify NSCLC stage. In addition, CA19-9 was highly expressed in certain pathological types of LAC [24][25][26]. CA72-4 had not been reported in NSCLC.
The expression of SCC-Ag was correlated with differentiation degree of squamous cell carcinoma tissue. SCC-Ag was highly expressed in well-differentiated squamous cell carcinoma tissues and lowly expressed in poorly differentiated squamous cell carcinoma tissues [31]. The level of SCC-Ag was correlated with T stage of both LAC and LSCC (r = 0 080, p < 0 05; r = 0 099, p < 0 05). The level of SCC-Ag may be correlated with the highly differentiated squamous cell carcinoma of the lung. Due to the low sensitivity, SCC-Ag provided no additional value when used in combination with CYFRA21-1 to diagnose NSCLC [32]. Compared with CYFRA21-1, SCC-Ag has lower sensitivity (39.8%) but higher specificity (96.1%) for LSCC. SCC-Ag was still a widely used TM to identify pathological types of NSCLC and still an effective method for the identification of benign and malignant solitary pulmonary nodules [33]. CA15-3 was a soluble form of mucin-1 serum shedding [34]. Mucin-1 expression may be associated with nonsquamous cell carcinoma tissue [35]. CA15-3 level was correlated with T stage, N stage, and M stage of LAC (r = 0 254, p < 0 05; r = 0 340, p < 0 05; and r = 0 293 , p < 0 05). CA15-3 was also correlated with M stage of LSCC (r = 0 168, p < 0 05). CA15-3 in LAC may be correlated with primary tissue, lymph node metastasis, and distant metastasis. CA15-3 in LSCC may indicate distant metastasis. However, Liu et al. [36] reported that mucin-1 mRNA in NSCLC peripheral blood cannot be used as a reliable TM of metastasis. The value of CA15-3 in diagnosis of NSCLC still needs further study.

Data Availability
The data availability statement is listed in the manuscript.

Ethical Approval
This study was approved by the ethics committee of the Guangxi Medical University Affiliated Cancer Hospital with an ethics approval number of LW2018011.

Conflicts of Interest
The authors have no conflicts of interest to declare.