Pre-treatment Controlling Nutritional Status (CONUT) Score and Carcinoembryonic Antigen Level Provide Tumour Progression and Prognostic Information in Gastric Cancer (cid:0) A Retrospective Study

Background This study explores the role of combining the controlling nutritional status (CONUT) score and the carcinoembryonic antigen (CEA) level on predicting tumour stage and prognosis in gastric cancer (GC) patients. Methods A total of 682 GC patients were included in this retrospective study. CONUT scores and CEA levels were combined to establish a new scoring system: CONUT-CEA score. Cut-off values for distinguishing patients between stage IV and non-stage IV were established by receiver operating characteristic curves. Cut-off values for predicting prognosis were determined by maximum x 2 method. Results The CONUT and CEA cut-off values for discriminating stage IV patients from non-stage IV patients were 2.0 and 5.58 ng/ml, respectively. Logistic regression model demonstrated that high CONUT-CEA score was related to advanced tumour stage. Among non-stage IV patients, CONUT and CEA cut-off values of 2.0 and 9.50 ng/ml predicted overall survival (OS), respectively. The Cox proportional risk model revealed that high CONUT-CEA score was notable related to decreased OS (2 vs 0: hazard ratios (HR) =2.358, 95% condence intervals (CI)=1.412–3.940, P=0.001) and decreased disease-free survival (2 vs 0: HR=1.980, 95% CI=1.072–3.656, P=0.003). Cox multivariate analysis, CONUT-CEA score-2 1.452, 1.046–2.016, P = 0.001; HR 1.980, 95%CI = 1.072–3.656, P = 0.003), TNM stage (II vs 2.820, 1.378–5.772, 3.300-13.335, P neurovascular


Background
Despite great progress in diagnosis and treatment, gastric cancer (GC) remains the fourth most common malignant tumour and the third leading cause of cancer death globally [1]. There are no symptoms or only nonspeci c symptoms present at the time of early-stage diseases; consequently, GC is generally diagnosed in advanced stages [2]. Furthermore, patients with the same tumour stage may have different outcomes because of their heterogeneity [3]. Accordingly, accurately evaluating the tumour progression and prognosis of GC patients is helpful for developing individualised treatment programmes to improve prognosis. Tumour progress depends not only on tumour characteristics but also on the nutritional status, systemic in ammation status and immune-compromised status of the patients [4].
Controlling nutritional status (CONUT), a newly developed scoring instrument, consists of three blood indexes: serum albumin (ALB) concentration, total number of peripheral blood lymphocytes and total cholesterol concentration. CONUT was initially reported as a screening instrument to evaluate the nutritional status of patients [5][6][7][8]. The prognostic values of CONUT for a variety of human tumours have been reported in recent years. It can be used to predict the prognosis of breast, bladder and lung cancer and the prognosis of GC patients [7,[9][10][11]. The effect of the CONUT score on the prognosis of GC patients was rst reported in 2018 [12]. Since then, other studies on the effect of CONUT on the prognosis of GC have been published, but the results were inconsistent. Some studies reported that the overall survival (OS) of GC patients with high CONUT scores was signi cantly lower than those with low CONUT scores [12][13][14][15], but the CONUT score cut-off values differed among studies. Liu X et al. [16] and Suzuki S et al. [15] found that the CONUT score can independently predict cancer-speci c survival (CSS). However, Kuroda D et al. [12] reported that the CONUT score had no signi cant effect on CSS. Moreover, these studies only analysed patients with Tumour-Node-Metastasis (TNM) stages I-III and did not include stage IV patients. The treatment and prognosis of stage IV and non-stage IV patients are different. Therefore, evaluating the preoperative clinical stage is key to choosing the appropriate treatment. In ammation, nutrition and other indicators may be altered in stage IV patients. Hence, the CONUT score may be able to predict preoperative stage IV GC.
Carcinoembryonic antigen (CEA), a structural antigen expressed on the surface of tumour cells, is one of the most commonly used biomarkers of GC [17]. CEA levels are often used in the early detection of cancer, and some studies have shown that CEA levels are related to preoperative predictions and can re ect tumour characteristics [18]. CEA levels have been used as a prognostic indicator in GC [19][20][21]. However, due to its low sensitivity and high false-positive rate, the role of CEA as an independent prognostic marker has always been controversial [22,23]. Although CEA levels have limited value as an independent prognostic factor, some research has shown that the combinations of CEA levels and other indicators can increase their prognostic sensitivity [24,25].
In this retrospective study, CONUT scores and CEA levels were combined to establish a new scoring system: CONUT-CEA score. This study analysed the role of combinations of pre-treatment CONUT score and CEA levels on tumour stage, and accessed the predictive value of the combined detection of the CONUT score and CEA levels on the prognosis of GC patients.  [24]. In the non-stage IV patients, 282 (50.5%) had proximal GC and 276 (49.5%) had distal GC; all the non-stage IV patients were subjected to radical operation. Seventy-four patients with stage IV GC were subjected to surgery, and the remaining 50 patients did not receive surgery and were diagnosed as stage IV by imaging examinations.
Stage IV GC was diagnosed as follows: (1) metastasis of the liver, lung, bone, pancreas and other organs; (2) peritoneal dissemination; and (3) metastasis of distant lymph nodes.

Clinical assessment and laboratory data
Medical records and laboratory results were retrospectively reviewed. Age, sex, smoking status, alcohol intake, clinical characteristics, lymphocytes, ALB and total cholesterol level were collected from the patients' medical records. All peripheral venous blood samples were collected in the morning after fasting for one night (at least 8 hours). The CONUT scores were calculated from the ALB concentration, total blood cholesterol level and total peripheral lymphocyte count (Table 1). CONUT score = Serum albumin score + Total lymphocyte count score + Total cholesterol score CONUT-CEA score was established by combining CONUT scores and CEA levels, including CONUT-CEA score-1 and CONUT-CEA score-2.
Cut-off values of CONUT scores and CEA levels for distinguishing between stage IV and non-stage IV patients were established, and the CONUT-CEA score-1 was calculated based on the above cut-off values. Cut-off values of CONUT scores and CEA levels for predicting OS in non-stage IV patients were established, and the CONUT-CEA score-2 was calculated based on the above cut-off values.

Follow-up
For the non-stage IV patients, follow-up was conducted every three months during the rst two years after resection, every six months during the next three years and every year thereafter. Postoperative follow-up included physical examination, laboratory examination, gastroscopy and chest/abdominopelvic computed tomography. The last follow-up was performed on September 1, 2019. The de nition of OS was from the date of radical operation to death or the last follow-up. The de nition of disease free survival (DFS) was from the date of radical operation to the date of local tumour recurrence and/or metastasis or the date of the last follow-up. The main endpoint was OS, and the secondary endpoint was DFS.

Statistical analyses
Cut  Table 2. The baseline clinical characteristics of the non-stage IV and stage IV cases are shown in Table 3. No signi cant differences were noted in sex, age, BMI, smoking status and alcohol intake (P > 0.05).

Relationship between pre-treatment tumour stage and CONUT/CEA
The median (IQR) CONUT score of the stage IV patients was 2 (1)(2)(3)(4), and that of the non-stage IV patients was 1 (0-3). The CONUT score of stage IV patients was signi cantly higher than that of non-stage IV patients (P = 0.002) ( Table 3). The median (IQR) CEA of the stage IV patients was 4.30 (1.77-11.93) ng/ml, and that of the non-stage IV patients was 2.28 (1.41-4.43) ng/ml. The CEA of the stage IV patients was signi cantly higher than that of the non-stage IV patients (P = 0.005) ( Table 3).
According to the ROC curves, the area under the cure (AUC) was 0.592 (95% CI: 0.535-0.649) for discriminating stage IV patients from non-stage IV patients based on the CONUT score, and the AUC was 0.651 (95% CI: 0.580-0.695) for discriminating stage IV patients from non-stage IV patients based on the CEA level ( Fig. 1a and b). According to the ROC analysis, the cut-off value of CONUT was 2.0, the sensitivity and speci city were 0.436 and 0.708, respectively. The cut-off value of CEA was 5.58 ng/ml, the sensitivity and speci city were 0.460 and 0.815, respectively.

Univariate and multivariate analyses of OS
As shown in Table 6, CONUT ≥ 2, CEA ≥ 9.5 ng/ml, high CONUT-CEA score-2, age ≥ 75 years, T stage, N stage, TNM stage, histological type and neurovascular invasion were established as signi cant prognostic factors of OS in the univariate analysis (all P < 0.05).   (Table 6).

Univariate and multivariate analyses of DFS
According to the univariate analysis (Table 6) (Table 6).

Discussion
In the present study, CONUT scores and CEA levels were combined to establish the CONUT-CEA score, which is a new scoring system used for predicting the preoperative stage and postoperative prognosis of GC patients. The present study demonstrated that the CONUT-CEA score plays an important role in discriminating patients with stage IV GC from those with non-stage IV GC, and high CONUT-CEA score was notable related to poor OS and DFS in non-stage IV GC patients. To our knowledge, this study is the rst to determine the role of the CONUT-CEA score in the staging and prognostic evaluation of GC.
A CONUT cut-off value of 2 was used to preoperatively determine whether the patient was stage IV or non-stage IV. The present study also showed high CONUT levels (≥ 2) was associated with worse OS and DFS. The CONUT cut-off value in the present study was the same as that reported by Ryo S et al. [14]. The CONUT score is calculated according to the ALB concentration, total blood cholesterol level and total peripheral lymphocyte count. In addition to re ecting nutritional status, ALB can re ect systemic in ammation [26]. Proin ammatory cytokines decrease ALB by regulating the synthesis of albumin in the liver [27,28]. The total number of lymphocytes re ects the host's immune response to the tumour [29]. Low peripheral blood lymphocyte counts lead to insu cient immune responses of the host to cancer cells, leading to cancer progression [30][31][32]. The total cholesterol concentration is related to tumour progression and the prognosis of various cancers [33,34]. Tumour growth requires cholesterol consumption, leading to cholesterol lowering [35].
The CONUT score can re ect nutritional status, the systemic in ammation status and immune response. Nutritional damage and immunosuppression in cancer patients promote the chronic in ammatory response of cancer cells [36][37][38]. Additionally, systemic in ammation and malnutrition are the main reasons for the reduction of interleukin-2 and interleukin-3 levels, leading to the impairment of tumour immune function and the increase of tumour cell proliferation [36,38]. The increased production of vascular endothelial cell growth factors is related to chronic in ammation, malnutrition and immunosuppression in GC patients [39].
CEA is a glycosylated protein with a glycosylated form of salivary fucosylation [40]. As a selectin ligand, CEA promotes the metastasis of cancer cells [41,42]. CEA levels are closely associated with tumour burden, so the elevation of CEA levels may be correlated with the pathological stage and prognosis of cancer. Some researches have found that CEA levels are related to pathological stage and can predict prognosis and recurrence in GC, and some studies also reported that high CEA levels above the upper limit of normal indicate a poor outcome in GC; however, the cut-off values of CEA remain unclear. Eui Soo Han et al. [43] found that the CEA levels of GC patients with extensive peritoneal seeding were signi cantly higher than those of other stages, but the ROC curve did not determine the cut-off values of CEA. Xiaowen Liu et al. [44] found that the increase of serum CEA levels was related to the GC pathological stage and lymph in ltration, but they did not show the cut-off value of CEA. A study conducted by Park et al. [45] found a high recurrence rate in CEApositive GC patients. Takahashi et al. [46] founded that patients with high CEA levels, particularly a CEA ratio more than twice the normal upper limit, experience more frequent cancer recurrence. Jong-Chan Lee et al. [47] also found that patients with CEA levels more than twice the normal limit had a poor prognosis. Jun Xiao et al. [48] reported that a cut-off value of 30.02 ng/ml could be applied to distinguish between patients with a poor prognosis and good prognosis.
In the present study, according to the ROC analysis results, CEA cut-off value of 5.58 ng/ml can be used to judge whether patients were in stage IV preoperatively. When assessing prognosis, the present study unveiled that CEA levels ≥ 9.50 ng/ml was related to worse OS and DFS in non-stage IV patients after radical operation. Moreover, the CEA levels were nearly twice as high as normal limits, which was close to the results reported by Kim et al. [49] and Jong-Chan Lee [47] et al.
Tumour progression is known to be associated with tumour characteristics, host nutritional status, systemic in ammation status and immunocompromised status. Therefore, identifying parameters that can re ect both tumour characteristics and host state can provide a better prognostic value. Meanwhile, CEA levels were found to be related to invasion of tumour cells [39][40][41], and the CONUT score can re ect the nutritional, in ammation and immune response state. The multivariate logistic analysis of the present study showed that high CONUT-CEA score was related to advanced tumour stage. So we believe that the CONUT score combined with the CEA level can improve the preoperative diagnosis of GC stage. Moreover, the current study demonstrated that the CONUT-CEA score was a more effective candidate prognostic biomarker in patients undergoing surgical resection of GC than the CONUT score or CEA level alone displayed in Table 6.
There were some limitations in this study. First of all, this was a retrospective study at one centre, so the possibility of selection bias cannot be completely controlled. Second, we cannot assess the in uence of the postoperative CONUT score and CEA level on the prognosis of GC. Third, different nutritional supports after surgery were inevitable, which may confuse our results.

Conclusions
In summary, the CONUT score and CEA level are useful for the differential diagnosis of stage IV and non-stage IV GC, and the combination of these measurements is more helpful than each factor alone. Additionally, the CONUT score and CEA level are helpful for forecasting OS and DFS in GC patients after radical gastrectomy. Assessing preoperative CONUT-CEA score may help develop effective strategies for GC treatment. CONUT and CEA, as inexpensive and convenient markers, can play an important role in treatment decisionmaking and follow-up in GC. Availability of data and materials

Abbreviations
The datasets used and/or analysed during the current study are available from the rst or correspondence authors on reasonable request.

Ethics approval and consent to participate
The present study was approved by the Ethics Committee of the First A liated Hospital of Fujian Medical University (approval no: MRCTA, ECFAH of FMU [2019] 200). Written informed consent for data collection and analysis was obtained from the respective patients.

Consent for publication
Not applicable.