3.1 Classification and composition of lung cancer patients
This study included 255 lung cancer patients (53.4%), 103 patients with benign lung diseases (21.5%), and 120 healthy controls (25.1%). Among lung cancer patients, 44 had lung squamous cell carcinoma (17.3%), 183 had lung adenocarcinoma (71.8%), and 28 had small cell lung cancer (SCLC) (11.1%) (Table 1).
3.2 Comparison of tumor markers in lung cancer, benign lung diseases, and healthy controls
The positive rates, expression levels and the median concentrations of serum CEA, CYFRA21-1, NSE, proGRP, and SCC-Ag in patients with lung cancer were significantly higher than those in patients with benign lung diseases and healthy controls (p< 0.001). Besides, the positive rates of CYFRA21-1, CEA, NSE, SCC-Ag and proGRP were 58%, 43.5%, 31.3%, 19.8%, and 17.3%, respectively, and were not related to the patient’s sex, ALT, and AST levels (p>0.05) (Table 2). What’s more, there was a difference in age among the three groups (p<0.001), which might have been caused by confounding factors. Therefore, it’s necessary to be corrected in a multivariate model.
3.3 The diagnostic value of tumor markers in different types of lung cancer
Patients with different pathological types of lung cancer showed differences in serum CYFRA21-1, NSE, proGRP, and SCC-Ag levels and positive rates. CEA levels were not statistically significant among the three groups (p>0.05). The level of CYFRA21-1 in lung squamous cell carcinoma (median 6.26) was higher than that in lung adenocarcinoma (median 3.60) and small cell lung cancer (median 5.64) (p<0.05), while the levels of NSE and proGRP in small cell lung cancer (median 535.85) were significantly higher than those in lung squamous cell carcinoma (median 42.30) and lung adenocarcinoma (median 44.34) (p <0.001). The level of SCC-Ag in the squamous cell carcinoma group (median 2.20) was higher than that in the lung adenocarcinoma (median 0.80) and small cell carcinoma (median 0.75) groups (p<0.001) (Table 3), while compared to the small cell lung cancer group, there was no statistically significant difference in the level of SCC-Ag in the non-small cell lung cancer group (p>0.05) (Table 4). The levels of CYFRA 21-1, NSE, and proGRP levels in the small cell lung cancer group (median: 5.64, 32.50, 535.85, respectively) were significantly higher than those in the non-small cell lung cancer group (median: 3.71, 13.20, 44.33, respectively) (p <0.05) (Table 4).
3.4 The relationship between the level of markers and the occurrence of lung cancer
The quartile range of the independent variable were divided into four groups and introduced into the model as a grade variable (Table 5). After quartiles were divided, a non-conditional logistic regression model was constructed. The higher the levels of CEA, CYFRA21-1, NSE, and proGRP, the higher the incidence of lung cancer; on the contrary, the level of SCC-Ag was not correlated with the occurrence of lung cancer (p>0.05) (Table 6).
3.5 The combination of biomarkers in diagnosis of lung cancer
The ROC curve was used to analyze the diagnostic efficacy of these four risk factors alone or in combination with the diagnosis of lung cancer (Figures 1-4). The areas under the curves of CEA, CYFARA21-1, NSE, and proGRP were arranged from largest to smallest: CEA (0.746) = CYFARA21-1 (0.746) > proGRP (0.734) > NSE (0.631). We confirmed the critical value according to the principle of the largest Youden index, and then calculated the sensitivity and specificity: CEA 3.53 ng/mL, diagnostic sensitivity 51.50%, specificity 89.30%; CYFARA21-1 3.24 ng/mL, diagnostic sensitivity 55.10%, specificity 87.50%; NSE 10.86 ng/mL, diagnostic sensitivity 75.50%, specificity 47.60%; proGRP 47.91 pg/mL, diagnostic sensitivity 49.50%, specificity 84.50%. Among the single tumor markers, CYFRA21-1 had the highest diagnostic value in the lung cancer group, with a sensitivity of 55.10% and a Youden index of 0.426. The CEA+CYFRA21-1 lung cancer group had the highest diagnostic value in the combination of two tumor markers, with a sensitivity of 59.50% and a Youden index of 0.550. Among the three tumor markers, the CEA+CYFRA21-1+proGRP lung cancer group had the highest diagnostic value, with a sensitivity of 61.20% and a Youden index of 0.583. When the four tumor markers were combined, the sensitivity was 64.30%, the specificity was 93.20%, and the Youden index was 0.575 (Table 7).
3.6 The combination of biomarkers in the diagnosis of small cell lung cancer
We used the ROC curve to analyze the diagnostic efficacy of a single index or combined detection for small cell lung cancer (Figures 5-6). The diagnostic value of CYFARA21-1 combined with proGRP, NSE combined with proGRP, and the three combined detections of small cell lung cancer (AUC of 0.928, 0.925, and 0.924, respectively) was better than that of proGRP alone (AUC of 0.904) (Table 8).
3.7 Tumor markers play an important role in guiding the prognosis
The postoperative follow-up results of lung cancer patients undergoing surgery (excluding those who lost follow-up) showed that a total of 59 lung cancer patients survived and 9 died. There was no significant difference in gender, age, CEA, proGRP, and SCC-Ag between the survival group and the death group (p>0.05), indicating that the survival of lung cancer patients after surgery was not related to these factors. However, we found that the levels of CYFAR21-1 and NSE in the survival group were significantly lower than those in the death group. (p<0.05), indicating that CYFAR21-1 and NSE had a certain value in prognostic evaluation (Table 9).