Introduction

Serum carcinoembryonic antigen (CEA) is widely recognized and recommended as a reliable tumor marker for colorectal cancer. It is therefore used for making the diagnosis, predicting the prognosis, and monitoring treatment [1,2,3]. It is also used for clinicopathological characterization of lung cancer [4,5,6].

In cases of resectable non-small-cell lung cancer (NSCLC), the use of the preoperative (pre CEA) [7] and postoperative (post CEA) [8] CEA values is meaningful for predicting postoperative prognostic outcomes. However, little is known about the significance of the combined use of the pre- and postoperative CEA values for predicting the postoperative prognosis.

According to previous reports, the ratio of postoperative to preoperative serum CEA value (CEA ratio) is associated with the postoperative survival in patients with resected colorectal adenocarcinoma [9]. We therefore hypothesized that the CEA ratio would also be useful for predicting the prognosis of patients undergoing radical surgery for primary lung adenocarcinoma.

Study design and participants

This retrospective clinical study was approved by the Kagoshima University Hospital Ethics Committee (Research approval number: 220211). All data on the clinical characteristics of patients and pathology reports were extracted from the department's database and patient records. Of the 923 patients diagnosed with and operated on for NSCLC at our hospital between 2010 and 2016, 133 with lung adenocarcinoma with preoperative CEA > 5.0 ng/ml who underwent lobectomy for curative intent were included. The characteristics of the 133 patients are summarized in Table 1. Pathological tumor-node-metastasis (pTNM) staging was recorded for all patients based on the 7th edition of the American Joint Committee on Cancer (AJCC)/Union for International Cancer Control (UICC) classification. The preoperative serum CEA value was the most recent serum CEA value up to 4 weeks (2.2 ± 0.8 weeks) before surgery, while the postoperative serum CEA value was that measured 4–8 weeks (5.6 ± 1.2 weeks) after surgery before adjuvant chemotherapy. The reference value for the serum CEA value was ≤ 5.0 ng/ml. In this study, smokers quit smoking for at least 2 months before lobectomy and continued to quit after lobectomy.

Table 1. A comparison of clinical characteristics
Full size table

The following patients were not included in this study: those who had been treated for malignancy within the past 5 years, those with no preoperative or postoperative serum CEA measurements, dialysis patients, those who received preoperative chemotherapy or radiotherapy, those who underwent sublobar resection, bilobectomy and pneumonectomy, those whose follow-up information was incomplete, and those who died of other diseases.

Statistical analyses

The diagnostic value of the CEA ratio for the postoperative survival was evaluated using a receiver operating characteristic (ROC) analysis, and the cutoff value was defined by selecting the point on the ROC curve where sensitivity + specificity is the maximum value. Patient over-all survival (OS) curves were plotted using the Kaplan–Meier method, and the difference between the survival curves was determined using the log-rank test. A univariate analysis of the Cox proportional hazards model was used to select significant factors among patient characteristics, serum CEA values, pathological factors, presence or absence of ground-glass opacification (GGO) and presence or absence of epidermal growth factor receptor (EGFR) mutation. In multivariate analyses, a stepwise Cox proportional hazards model was used with factors with p < 0.05 in univariate analyses.

All statistical analyses were performed using the EZR (Saitama Medical Center, Jichi Medical University, Saitama, Japan) and SPSS (Dr SPSS II for Windows, Standard Version 26.0; SPSS Inc., Chicago, IL, USA) software programs. Statistical significance was set at P < 0.05.

Results

The median postoperative follow up period was 73.5 (range 0.6–148.3) months. During the postoperative follow-up period, 52 patients died. According to ROC curves showing the diagnostic potential of post CEA for the survival of 133 patients, the best cutoff value of post CEA was 6.2 ng/ml (sensitivity: 38.5%, specificity: 87.7%, area under the curve [AUC]: 0.648, 95% confidence interval [CI]: 0.551–0.745). When the 133 patients were divided into groups based on the post CEA (≤ 6.2 or > 6.2 ng/ml), the 5-year OS rates were 81% (n = 105) and 43% (n = 28), respectively (p ≤ 0.001) (Fig. 1A, B). Similarly, according to ROC curves showing the diagnostic potential of the CEA ratio for the survival of 133 patients, the best cutoff value of the CEA ratio was 0.58 (sensitivity: 50.0%, specificity: 77.8%, AUC: 0.619, 95% CI 0.519–0.72). When the 133 patients were divided into groups based on the CEA ratio (≤ 0.58 or > 0.58), the 5-year OS rates were 79% (n = 94) and 60% (n = 39), respectively (p ≤ 0.001) (Fig. 1C, D).

Fig. 1
figure 1

ROC curves and survival curves for all patients. ROC curve using the post CEA values for all patients (A). Survival curves for all patients were stratified into two groups: post CEA ≤ 6.2 ng/ml and post CEA > 6.2 ng/ml (B). ROC curve using the CEA ratio for all patients (C). Survival curves for all patients were stratified into two groups: CEA ratio ≤ 0.58 and CEA ratio > 0.58 (D)

Full size image

The usefulness of the CEA ratio was further evaluated by an ROC analysis in patients with post CEA > 6.2 ng/ml (n = 28) and those with post CEA ≤ 6.2 ng/ml (n = 105) (Fig. 2A). The AUC was 0.83 (95% CI 0.652–1) for patients with post CEA > 6.2 ng/ml, with the best cutoff value of the CEA ratio being 0.39 (sensitivity: 94.7%, specificity: 66.7%), whereas the AUC was only 0.457 (95% CI 0.328–0.585) for patients with post CEA ≤ 6.2 ng/ml (Fig. 2B, C). In other words, the CEA ratio was useful in patients with post CEA > 6.2 ng/ml but not in those with post CEA ≤ 6.2 ng/ml.

Fig. 2
figure 2

Scatter plots and ROC curves for all patients. Scatterplot of the CEA ratio and post CEA values in all patients still alive (〇) or who had died ( ×). Patients were stratified into two groups: post CEA ≤ 6.2 ng/ml and post CEA > 6.2 ng/ml (A). The ROC curve for the CEA ratio showed a relatively high AUC in patients with post CEA > 6.2 ng/ml (B) but showed a relatively low AUC in patients with post CEA ≤ 6.2 ng/ml (C)

Full size image

Based on the above ROC curve analysis, all patients were classified into three groups using the post CEA and CEA ratio: group a with post CEA ≤ 6.2 ng/ml (n = 105), group b with post CEA > 6.2 ng/ml and a CEA ratio < 0.39 (n = 7), and group c with post CEA > 6.2 ng/ml and a CEA ratio ≥ 0.39 (n = 21) (Fig. 3A). According to the survival analysis, group c showed a worse prognosis than the other two groups (group a vs. c; p < 0.001, group b vs. c; p = 0.02), while groups a and b showed comparable prognostic outcomes to each other (p = 1.000) (Fig. 3B).

Fig. 3
figure 3

Based on the ROC analysis in Fig. 2C, all cases were classified into 3 groups: post CEA ≤ 6.2 ng/ml (group a), post CEA > 6.2 ng/ml and CEA ratio < 0.39 (group b), and post CEA > 6.2 and CEA ratio ≥ 0.39 (group c) (A). Group c showed the worst survival among the three groups (B)

Full size image

The results of the univariate and multivariate Stepwise Cox proportional hazards analyses are summarized in Table 2. According to univariate analyses, the pre CEA, post CEA, CEA ratio, pN status, and presence or absence of GGO were related to survival outcomes. A multivariate analysis further showed that the CEA ratio and pN status were independent prognostic factors.

Table 2 Hazard ratios and 95% CI from univariate and multivariate Stepwise Cox proportional hazard models of all cases (OS)
Full size table

Discussion

Many reports have examined the prognosis of NSCLC using CEA values before and after surgery for NSCLC [6, 8, 10]. However, the results have been inconsistent regarding the usefulness of the pre CEA value [7, 8], probably because patients whose CEA had been normalized postoperatively might show a relatively favorable prognosis [11]. However, there are many reports that the prognosis is poor when both the pre CEA value and post CEA values are high [8, 11].

The present study included patients with abnormally high pre CEA values. As expected, patients with a high post CEA value had a poor prognosis. We further investigated whether or not the CEA ratio, in addition to the post CEA value, can improve the accuracy of predicting the postoperative survival. Consequently, the CEA ratio was useful in patients with post CEA values > 6.2 ng/ml but not in those with post CEA values ≤ 6.2 ng/ml. Patients with post CEA ≤ 6.2 ng/ml may include patients with nonspecifically elevated CEA. In such cases, the CEA ratio may not accurately reflect the prognosis because the change in pre to post CEA is little.

In general, although serum CEA values are routinely measured perioperatively, there are no specific guidelines concerning the use of serum CEA values in therapeutic decision-making for patients with high CEA values. If the CEA ratio can be determined to be a reliable marker for identifying patients with a high risk of recurrence, it can be used to identify patients requiring postoperative adjuvant chemotherapy. As this was a retrospective study and the number of patients included was small, a larger prospective study should be conducted to verify the reliability of the CEA ratio.

According to the literature, the statistical utility of the CEA ratio has been verified in colon cancer [9] however not in lung cancer [10]. To our knowledge, only one report is available regarding the usefulness of the CEA ratio in the prognostic assessment of lung cancer. Tomita et al. reported that post CEA was more useful than the CEA ratio for predicting the prognosis [10]. However, whether or not our study results are indeed inconsistent with Tomita’s results is inconclusive, as Tomita et al. did not evaluate the usefulness of the CEA ratio in combination with post CEA. In addition, Tomita’s study differed from ours with regard to patient selection criteria: Tomita et al. included patients undergoing sublobar resection, patients with non-adenocarcinoma, and patients undergoing incomplete resection. In addition, unlike Tomita et al., we evaluated the presence or absence of GGO [12] in the primary lesion and presence or absence of EGFR mutation, which are known to affect the prognosis of lung adenocarcinoma.

A multivariate Cox proportional hazards analysis showed that the CEA ratio and pN status were independent prognostic factors among variables identified as significant by a univariate analysis (pre and post CEA, CEA ratio, pN status, and presence of GGO). A multivariate analysis may indicate that the CEA ratio is superior to the presence or absence of GGO in predicting the postoperative survival A possible explanation for this result is that a high CEA ratio may indicate the presence of a subclinical residual tumor, which may lead to the development of subsequent recurrence. Therefore, the CEA ratio can be used for the early diagnosis of recurrence in the early postoperative period, which is useful in making decisions concerning postoperative anticancer treatment.

There are some limitations in the current study. As this was a retrospective study and the number of patients was small, we believe that a larger prospective study is needed. Among the included patients, 79 patients with a history of smoking had quit smoking. The CEA values may be affected by smoking. The effect of smoking on serum CEA value has been reported in the past [13], however because many lung adenocarcinoma patients have a history of smoking, we did not exclude patients with a history of smoking from the current study. Furthermore, in this study, only 41 of 93 patients (44.1%) eligible for adjuvant chemotherapy actually received it. Therefore, it was not possible to evaluate its impact on the current study.

Conclusion

The CEA ratio may be a useful prognostic marker in patients who undergo lobectomy for lung adenocarcinoma and show postoperative CEA > 6.2 ng/ml. Further studies are warranted to validate the utility of this marker in making decisions for adjuvant therapy.