LIU Xiu-ling, YU Hua-yi, QI Guo-ping, LU Wen-bin, LIU Qian, YANG Yu-wei, DENG Jian-zhong

Department of Oncology, Wujin Hospital Affiliated to Jiangsu University/Wujin Clinical College of Xuzhou Medical University/Changzhou Key Laboratory of Molecular Diagnostics and Precision Cancer Medicine, Changzhou 213017, China

Keywords:

ABSTRACT Objective: To construct a nomogram model for predicting the prognosis of colorectal cancer patients based on preoperative fibrinogen (Fbg), platelet to albumin ratio (PAR) and carbohydrate antigen 199 (CA199).Methods: In this retrospective study, we included 323 patients who underwent colorectal cancer (CRC) surgery in our hospital.The preoperative test indexes and relevant clinicopathological data of the patients were collected.According to the cut-off value of Fbg and PAR calculated from the receiver operating characteristic (ROC)curve, they were divided into the high and low expression groups.Univariate and multivariate Cox proportional hazards models were used to screen independent prognostic factors which were used to construct a nomogram.Results: According to ROC curve, the cut-off values of Fbg and PAR were 2.80 and 6.05.Multivariate Cox regression analysis showed that Fbg,PAR, CA199, TNM stage and grade were independent risk factors affecting the prognosis of patients (P ＜0.05).Fbg, PAR, and CA199 jointly built the risk score that was included into the nomogram.The area under the curve (AUC) in the ROC of the training and verification set was greater than 0.6.The calibration curve and ideal curve fit well.Conclusion: The nomogram based on Fbg, PAR and CA199 has well accuracy and can provide individualized prediction for the overall survival of CRC patients.

1.Introduction

At present, colorectal cancer (CRC) is the third most common tumor in the world and the second leading cause of cancer-related death[1].In recent years, with the change of life style, the incidence rate of CRC has increased year by year[2].Tumor stage is the most important prognostic factor, but the prognosis of patients with the same clinical stage is different.Therefore, it is necessary to judge the prognosis of patients individually so as to provide more effective treatment plans.Previous studies have shown that the plasma Fibrinogen (Fbg) level is significantly correlated with the survival of CRC patients and can be used as an effective prognostic marker[3].Platelets are fragments of nucleated cells produced by megakaryocytes in the bone marrow as well as in other organs,including the lungs.A large body of evidence now links platelets to cancer pathogenesis [4].Increased platelet count promotes cancer growth, infiltration and metastasis [5].Albumin is an important nutritional reference for cancer patients [6].Preoperative peripheral blood albumin levels have a non-negligible value on the prognosis of CRC patients [7].Carcinoembryonic antigen (CEA)and carbohydrate antigen 19-9 (CA199) are widely used in the diagnosis and prognosis assessment of malignant tumors in the digestive system [8].There are rare studies that combine all these factors to assess the prognosis of patients.This study combined inflammatory, nutritional, and coagulation indicators and tumor markers to construct a prognostic model for CRC patients, aiming to provide individualized prediction of prognosis for CRC patients through some economical and easily available indicators.

2.Materials and methods

2.1 General information

323 CRC patients who underwent surgery in our hospital from January 2016 to January 2019 were included for this retrospective analysis, including 201 males and 122 females,aged 24-93 years (median age: 68 years).The inclusion criteria of this study were: (1) postoperative and pathologically confirmed colorectal adenocarcinoma; (2) no combination of other systemic malignancies; (3) no preoperative anti-tumor related treatment;(4) complete data; (5) informed consent from patients or family members.Exclusion criteria: (1) postoperative pathology confirmed as colorectal squamous carcinoma or neuroendocrine carcinoma or other non-adenocarcinoma diseases; (2) combined with other systemic malignancies; (3) preoperative neoadjuvant chemotherapy or radiotherapy; (4) with severe coagulation abnormalities, local or systemic infections; (5) lost to follow-up.Pathological staging was defined according to the 7th edition of the American Joint Committee on Cancer (AJCC) TNM staging system.All patient data were obtained with the approval of our ethics committee.

2.2 Research methods

Relevant patient information was collected at the time of surgery including: gender, age, depth of tumor infiltration, lymph node metastasis, TNM stage, degree of differentiation, tumor size, tumor location, and tumor type.Laboratory tests were performed within 2 weeks before surgery, 2.5 mL of peripheral venous blood was routinely drawn and collected using a special blood collection tube for coagulation tests with sodium citrate anticoagulation, and plasma was further separated for Fbg detection by latex-enhanced immunoturbidimetry using the CS5100 automated coagulation analyzer.2.0 mL of peripheral venous blood was also collected with a vacuum blood collection tube anticoagulated with EDTA-K2 anticoagulant, and platelets were detected using an XN series blood cell analyzer.Albumin was detected by bromocresol green method using a fully automated biochemical analyzer, while CA199 was detected by chemiluminescence method using a fully automated chemiluminescent immunoassay.And CEA was detected by enzymelinked immunosorbent assay.The specific detection method is consistent with the published papers of our group [9].Fbg, platelets,albumin, CEA, CA199 were obtained from the medical records.And PAR represents the platelet/albumin ratio.

2.3 Patients follow-up

Follow up was carried out by telephone, outpatient follow-up or hospitalization.The start date was set as the operation date, and the follow-up was to December 31, 2021.Death was the end point of follow-up.Overall suevival (OS) in the data was defined as the period from the operation date to the last follow-up or death date.

2.4 Statistical processing

The data were analyzed by SPSS 22.0.We constructed the ROC curves which determine the cut-off value of Fbg and PAR.The patients were divided into high and low expression groups according to the cut-off values.χ2test was used to evaluate the relationship between Fbg, PAR and clinicopathological parameters of CRC patients.Prognostic factors were assessed using univariate and multivariate analysis (Cox proportional hazards regression model).Biomarkers as independent prognostic factors were used to construct new risk scores (risk score =Coef1* Fbg+Coef2* PAR+Coef3*CA199).Coef represents the regression coefficient of each index.Patients were divided into high and low risk groups according to the median value of the risk score, and further Kaplan Meier survival analysis was performed.323 patients were randomly divided into two groups through the R language "car" and "survival" packages,including 227 in the training set and 96 in the validation set.There was no statistical difference between the two groups.R studio 3.6.1 software and its "car", "rms", "pROC" and "timeROC" packages were used to construct nomograms, and the ROC curves and calibration curves were used for model evaluation.P value ＜ 0.05 was considered statistically significant.

3.Results

2.1 ROC curves drawing and cut-off values determination of Fbg and PAR

The area under the curve (AUC) of Fbg and PAR obtained by ROC curves were: 0.605(95%CI: 0.539-0.670), 0.604(95%CI: 0.536-0.672).cut-off values were: 2.80, 6.05.Sensitivities were: 76.2%,57.4%.Specificities were: 41.9%, 64.0%.The results were shown in Figure 1.Patients were divided into Fbg 2.80 group, Fbg ＞ 2.80 group, PAR 6.05 group and PAR＞ 6.05 group according to cut off values.

Fig 1 ROC curves of preoperative Fbg and PAR levels of CRC patients

3.2 Correlation between Fbg, PAR and clinicopathological parameters

The high and low expression of Fbg was significantly correlated with the depth of tumor invasion, tumor size, location and CEA, but not with gender, age, lymph node metastasis, stage, differentiation and CA199.The expression of PAR was significantly correlated with sex, depth of tumor invasion, tumor size and location, but not with age, lymph node metastasis, stage, degree of differentiation, CEA and CA199(Table 1).

3.3 Univariate and multivariate analysis related to OS

In univariate analysis, Fbg, PAR, CA199, tumor size, depth of tumor invasion, lymph node metastasis, TNM stage, and degree of differentiation were the factors affecting the prognosis of patients(P＜0.05).The above factors were included in multivariate Cox regression analysis by stepwise forward method.The results showed that Fbg, PAR, CA199, TNM stage and differentiation were independent risk factors affecting the prognosis of colorectal cancer patients (P＜0.05).See Table 2.

Tab 1 The relationship between Fbg、PAR and clinicopathological parameters of CRC patients(n, %)

2.4 The construction of prognostic risk scores

Patients were divided into high- and low-risk groups according to the median risk scores constructed by Fbg, PAR and CA199.The Kaplan Meier survival analysis results were shown in Figure 2.Among the risk score grouped patients, the median OS of patients in the high-risk group was significantly lower than that of patients in the low-risk group (P＜0.001).

2.5 Establishment of nomogram model for prognosis prediction

The risk score and clinicopathological parameters (TNM stage and differentiation degree) as independent prognostic factors were used as predictors to draw the nomogram.The scores corresponding to each index were added to obtain the total score.The corresponding survival rate was obtained according to the intersection of the vertical line drawn downward from the total score and the coordinate axis corresponding to the 1, 3 and 5-year survival rate.The nomogram results were shown in Figure 3.The risk score had the highest impact on prognosis, followed by TNM stage and differentiation.

Fig 2 Survival curves of CRC patients

2.6 Evaluate the prediction efficiency of nomogram model

The nomogram was verified by internal verification, and measured by ROC curves and calibration curves.The area under ROC curves(AUC) ＞ 0.6 indicated that it had good prediction efficiency.The AUC corresponding to the 5-year survival rate in the training and validation set were 0.687 and 0.713 respectively, and the 5-year AUC corresponding to the risk score was higher than the AUC corresponding to the TNM staging system and the degree of differentiation in the seventh edition, indicating that the model had good discrimination.The calibration curve showed the relationship between the 5-year survival rate predicted by the nomogram and the actual 5-year survival rate of the patient.The calibration curves in the training set and the validation set had high consistency with the ideal curve, and the difference between the predicted outcome events and the actual outcome events was small.See Figure 4.

Tab 2 Univariate and multivariate analyses of prognosis in CRC patients

Fig 3 Nomogram prediction model for overall survival of CRC patients

Fig 4 ROC curves and calibration curves of nomogram prediction model

4.Discussion

At present, as the incidence rate of CRC increases year by year and the number of patients with CRC is younger, it is very important to find a new, economic and practical biomarker to guide the further diagnosis and treatment of CRC patients, reduce the risk of recurrence and disease progression, and prolong the survival period.The results of this study showed that the risk score composed of Fbg, PAR and CA199 was a prognostic factor for patients after CRC surgery, and can provide individualized survival prediction for patients.

The occurrence, development and treatment of malignant tumors are often accompanied by abnormal coagulation function, which further causes thrombosis after the activation and release of coagulation factors, thus causing tumor progression [10].At the same time, malignant tumors can promote the formation of Fbg and further cause the increase of blood viscosity and vascular resistance.Fbg, also known as coagulation factor 1, is produced by hepatocytes and can be converted into fibrin under the action of thrombin and deposited in the extracellular matrix, providing a good angiogenic bed for tumor cells.Fbg can also act as an inflammatory regulator and is closely related to the proliferation, angiogenesis, invasion,metastasis and the formation of inflammatory microenvironment of malignant tumors [11].It has been found that Fbg can act as a mediator of tumor proliferation, and its mechanism of promoting tumor cell proliferation is to combine various growth factors such as (TGF-b, FGF, VEGF), endothelial cell adhesion molecules, and also to promote metastasis by maintaining the adhesion and survival of tumor cell emboli in the vascular system of target organs[12,13].Relevant studies have proved that Fbg is a marker of poor prognosis in CRC, lung cancer, breast cancer and other related solid tumors[14-16].

Platelets significantly participate in the growth and metastasis of cancer and affect the therapeutic effect of cancer patients [5].The mechanism of its growth promotion is to make focal adhesion kinase (FAK) extravasate into the tumor microenvironment.The important condition for cancer metastasis is cancer cells in circulating blood.Circulating tumor cells can form platelet tumor cell aggregates by interacting with platelets, release growth factors and small molecules that promote growth and invasion to protect them from NK cells and TNF mediated cell death, and also interfere with the recognition of tumor cells by NK cells [17].Platelets can promote the further progress of cancer.Meanwhile, cancer cells can promote the occurrence and development of tumor by inducing local and systemic platelet activation, adhesion, aggregation and stimulating angiogenesis.There is interaction between tumor and platelets[18].Moreover, relevant experimental studies in mice have shown that platelets can form an early metastatic microenvironment of immunosuppression and severe inflammatory response by aggregating granulocytes to tumor sites [19].Previous studies have shown that prostate cancer patients with higher platelet / lymphocyte ratio have an increased risk of recurrence and metastasis [20].Therefore, preoperative high platelet count is associated with poor prognosis of tumor patients.Cancer cells and platelets can further inhibit liver albumin synthesis through the release of inflammatory factors.The decrease of albumin is always related to the increase of C-reactive protein, which often causes the aggravation of systemic inflammatory response [21].At the same time, albumin can inhibit histone H4 induced platelet activation and aggregation, and exert anticoagulant effect by binding arachidonic acid [22].When the level of albumin in the body decreases, these functions will be weakened, which will promote tumor progression.Albumin is an important component of PAR, which is abundant in plasma and often synthesized in liver.It is a functional protein, which reflects the synthetic ability and nutritional status of the body's liver, and plays an anti-tumor role through anti-oxidation [23].Relevant studies systematically evaluated the significance of pretreatment albumin in the prognosis evaluation of various cancers including breast cancer,gastric cancer, CRC, liver cancer, non-small cell lung cancer and so on.Therefore, hypoalbuminemia is significantly associated with poor prognosis in patients with malignant tumors [24].

Previous studies have shown that PAR is related to clinicopathological parameters such as tumor invasion depth,lymph node metastasis, differentiation degree and stage of cancer patients, and it is an independent risk factor affecting the prognosis of cancer patients.Moreover, the survival rate of patients with low expression of PAR in gastric cancer, non-small cell lung cancer and locally advanced pancreatic cancer is significantly prolonged [25-27].In this study, we found that PAR was related to gender, tumor invasion depth, tumor size and location, and the expression of PAR was related to the prognosis of tumor patients.This finding is also consistent with the above study.Previous studies have reported that fibrinogen / albumin is related to TNM stage, preoperative tumor marker level and other clinicopathological parameters, which can predict the prognosis of rectal cancer patients.The results show that the overall survival and disease-free survival of patients in the low fibrinogen / albumin group are longer than the higher expression group, which can be used as independent prognostic factors in multivariate analysis [28].Serum CA199 has been proved to be a marker for diagnosis and prognosis of colorectal cancer [29].Relevant literatures have reported that nomograms based on factors that can predict the overall survival of patients are helpful for prognosis prediction and clinical decision-making of CRC patients [30].In this study, the new risk score constructed by three independent prognostic factors (Fbg, PAR and CA199) were combined with TNM stage and degree of differentiation to construct a nomogram.Compared with traditional TNM stage, the nomogram can provide more accurate and individualized prognosis assessment for CRC patients.In addition,the discrimination and calibration evaluation of the model shows that the model has a good prediction effect, and the AUC of the risk score is higher than TNM stage and differentiation degree, which further shows that the risk score has a better prognostic evaluation efficiency.Therefore, we believe that the nomogram model based on risk score, TNM stage and differentiation degree may better predict the prognosis of patients after CRC surgery.

In conclusion, a new, economical and practical risk score can evaluate the prognosis of patients with CRC.High risk is one of the poor prognostic factors of patients.However, there are still some limitations in this study.First, this study is a retrospective study with a single center and a small number of samples.The level of evidence-based medicine is not high, and it still needs to be verified by large samples and prospective studies; Secondly, the medication of some patients with underlying diseases may also affect the values of Fbg and PAR.In conclusion, the nomogram model has potential clinical significance in CRC patients.

Author's contribution:

Liu Xiu-ling, Yang Yu-wei and Deng Jian-zhong were responsible for the research design, data analysis and manuscript writing;Yang Yu-wei, Yu Hua-yi and Qi Guo-ping are responsible for data collection and collation; Lu Wen-bin and Liu Qian are responsible for statistical processing and feasibility analysis; Liu Xiu-ling and Deng Jian-zhong are responsible for paper review and quality control.

All authors declare no conflict of interest.