The clinical value of a nomogram constructed from CEA,CA199,PT,FIB,tumor differentiation and TNM stage in colorectal cancer

Abstract

BACKGROUND:

The accurate Tumor-Node-Metastasis (TNM) staging of colorectal cancer (CRC) is of great guiding significance for the judgment of tumor progression and prognosis, and the formulation of treatment strategies.

OBJECTIVE:

The aim of this study was to construct a recurrence risk scoring (RRS) system and prognostic prediction model to improve the accuracy of staging, prognosis prediction, and clinical decision making in resectable CRC.

METHODS:

CRC patients who underwent radical resection were retrospectively enrolled into study. Multivariable Cox regression model was applied to screen for independent prognostic factors. The RRS system is composed of independent prognostic factors which was awarded 1point each. A prognostic model composed of RRS and TNM staging system (RRS-TNM model) was applied to predict postoperative recurrence.

RESULTS:

TNM stage, tumor differentiation, preoperative elevated Carcinoembryonic Antigen, Carbohydrate Antigen 199, Prothrombin Time and Fibrinogen were the independent prognostic biomarkers. 173 of 540 patients had recurrence. The 5-year cumulative recurrence rate (5-y CRR) and disease-free survival (DFS) of postoperative p-TNM stage I, II, and III were 12.7% and 104.8 months, 26.5% and 89.3 months, and 55.5% and 57.3 months, respectively. The 5-y CRR and DFS of preoperative Low-risk (RRS 0-1score), Middle-risk (RRS 2-3scores), and High-risk (RRS 4-5scores) groups were 13.9% and 101.1 months, 40.9% and 75.5 months, and 70.2% and 41.1 months. The AUC (area under ROC curve) of RRS system was not inferior to that of TNM staging system (0.713 vs. 0.666; $P=$ 0.093). The AUC (0.770) and C-index value (0.721) of RRS-TNM model were significantly better than both RRS and TNM staging system ( $P<$ 0.001).

CONCLUSIONS:

The RRS system accurately identifies CRC patients with high-risk recurrence preoperatively. Constructing a nomogram using the RRS system and TNM staging significantly improves the accuracy of staging and prognosis prediction, which is of great clinical significance for individualized clinical treatment and follow-up of CRC.

Keywords

Carcinoembryonic antigen carbohydrate antigen 199 prothrombin time fibrinogen colorectal cancer

1. Introduction

Colorectal cancer (CRC) is one of the most common tumors, with the third highest morbidity and the second highest mortality worldwide [1], and ranks fifth in terms of morbidity and mortality in China [2], one of the main reasons for the poor prognosis of CRC is its high postoperative recurrence rate [3, 4]. As the gold standard of treatment decision and prognosis assessment, the Tumor-Node-Metastasis (TNM) staging system is widely used in clinical management of CRC [5]. However, the prognosis of CRC patients with the same pathological TNM (p-TNM) stage may be quite different in the clinic [6, 7, 8, 9], indicating that p-TNM staging alone cannot accurately reflect the true staging and predict the prognosis of CRC. Therefore, finding effective biomarkers to improve the accuracy of tumor staging system and accurately predict postoperative recurrence is of great significance for the individualized treatment and prognosis improvement of CRC.

Tumor markers are widely used in clinical tumor screening and monitoring of postoperative recurrence and metastasis. Carcinoembryonic antigen (CEA) and Carbohydrate antigen 199 (CA199) are abnormally elevated in several tumors, particularly those from the digestive tract [10, 11]. Studies have suggested preoperative CEA maybe an efficient predictor of recurrence of CRC after curative surgery [12, 13, 14, 15, 16]. Increased preoperative CA199 level was possibly associated with poor differentiation and prognosis in CRC [17, 18]. However, it is still controversial whether tumor markers can be used as biomarkers to predict postoperative recurrence of CRC. It is urgent to search for better biomarkers to predict recurrence and improve the accuracy of TNM staging. Studies have shown that coagulation parameters are valuable for predicting postoperative recurrence of CRC [13, 19, 20]. Furthermore, elevated preoperative prothrombin time (PT) level has been reported as a predictor of poor prognosis in CRC [19, 21] and increased level of fibrinogen (FIB) was associated with metastases and poor prognosis in patients with resectable CRC [13, 19, 22, 23]. The degree of pathological differentiation has been shown to be a reliable prognostic marker in CRC patients and significantly related with the overall stage of TNM system [24].

Due to the high heterogeneity of CRC, a single biomarker may not elevated in all patients, leading to limitations in its validity for prognostic prediction [25, 26]. The combined detection of multiple biomarkers may make up for the shortcomings of single biomarker detection to a certain extent [27, 28, 29], and has become an important tool for accurate staging and monitoring of recurrence of CRC [17].

The aims of the present study were: 1) identify and validate a recurrence risk scoring (RRS) system based on the most reliable prognostic markers in order to improve the accuracy of postoperative recurrence prediction. 2) combining the RRS system with TNM staging system (RRS-TNM model) accurately distinguishes patients with different recurrence risks and predicts the postoperative recurrence of CRC.

2. Materials and methods

2.1 Patients

This is a retrospective cohort study. The patients were selected from those stage I-III CRC patients who underwent curative resection in Zhongshan Hospital of Xiamen University from January 1, 2010 to December 31, 2018. The Ethics Review Committee of Zhongshan Hospital of Xiamen University approved this study (ZSYY20191102). These patients were included when: 1) with TNM stage I, II, or III CRC; 2) underwent radical resection. The exclusion criteria were: 1) without detailed preoperative data of coagulation indexes and tumor markers; 2) with preoperative chemotherapy or chemoradiotherapy; 3) with history of malignant tumor; 4) with coagulation disorders or history of anticoagulants or thrombolytic treatment. Ultimately, a total of 540 CRC patients were included (Fig. 1). All patients were given appropriate treatment according to the treatment guidelines and no additional clinical trials were performed on them.

Figure 1.

Study design and research object selection. Abbreviations: CEA, carcinoembryonic antigen; CA199, carbohydrate antigen 199; PT, prothrombin time; FIB, fibrinogen; PLT, platelet.

2.2 Research indicators

The clinical and pathological characteristics of the included patients were collected from the medical records, including gender, age, tumor location, tumor differentiation grade, T stage (the infiltration depth of primary tumor), N stage (lymph node metastasis), TNM stage, vascular and neural invasion. Tumor markers and coagulation indicators were analyzed, including CEA, CA125, CA199, CA724, platelet (PLT), D-dimer, PT, FIB. Through statistical analysis, five indicators (differentiation, CEA, CA199, PT, FIB) were selected as the components of the RRS system. Then the RRS system was integrated with TNM staging system into a prognostic prediction model (RRS-TNM model).

In our study, we used the 5-year cumulative recurrence rate (5-y CRR) and disease-free survival (DFS) to evaluate the outcomes of patients, DFS was defined as the time from operation to recurrence or metastasis. And this study was reported according to STROBE checklist [30].

2.3 Statistical analysis

All statistical analyses were based on IBM SPSS STATISTICS 26 software. Receiver-operator characteristic (ROC) curve was used to confirm the optimal sensitivity and specificity of these indexes, between the value corresponding to the maximum Youden Index and upper normal limits, the cut-off value of each included indicators is the one with higher sensitivity. The relationship of research parameters and clinicopathological indicators was depicted by the Pearson chi-square test. The effects of parameters on recurrence and metastasis were described by Kaplan-Meier cumulative recurrence curve, Log-Rank test depicted the difference between subgroups of parameters. Univariate and multivariate Cox proportional hazard model were used to analyze the prognostic factors of recurrence and metastasis. Nomogram and calibration curves were used to evaluate the prognostic model. Area under ROC curve (AUC) and Concordance index (C-index) were used to assess the discriminative ability. Risk score distribution was used to describe the accuracy of this prognostic model for recurrence, decision curve analysis was applied to compare the net benefit of patients using TNM stage and prognostic model. $P$ values were two-sided, differences were statistically significant when $P<$ 0.05, and the confidence interval (CI) was 95%.

3. Results

3.1 Clinicopathological characteristics of the included patients

A total of 540 stage I-III CRC patients who underwent curative resection were enrolled in this study. There were 228 females and 312 males. The median age of included patients was 66.5-year, ranging from 29 to 92, and 358 of them (66.3%) were older than 60 years. 454 participators (84.1%) had colon cancers in this study. The pathological report showed that well tumor differentiation was found in 442 patients (81.9%). A total of 205 patients (38.0%) had T stage1-3 primary tumors; lymph node metastasis was found in 230 patients (42.6%). There were 66 patients (12.2%) with TNM I staging, 230 patients (42.6%) with TNM III staging. Vascular invasion occurred in 299 patients (55.4%), and neural invasion was caught in 401 patients (74.3%). By the end of the follow-up, 173 patients (32.0%) had recurrence and metastasis (Table 1).

Table 1
Clinicopathologic features and demographics of the included patients

	Patients ( $N=$ 540)	Percent (%)
Gender
Female	228	42.2
Male	312	57.8
Age (year)
$<$ 60	182	33.7
$\geqslant$ 60	358	66.3
Location
Colon	454	84.1
Rectum	86	15.9
Differentiation
Well	442	81.9
Poor	98	18.1
T stage
T1-3	205	38.0
T4	335	62.0
N stag
N-	310	57.4
N+	230	42.6
TNM stage
stage I	66	12.2
stage II	244	45.2
stage III	230	42.6
Vascular invasion
No	241	44.6
Yes	299	55.4
Neural invasion
No	139	25.7
Yes	401	74.3
Recurrence and metastasis
No	367	68.0
Yes	173	32.0

3.2 Identifying potential predictors of postoperative recurrence

During variable screening, we compared the AUCs of 4 cancer markers and 4 coagulation markers, the AUCs of CEA, CA125, CA199 and CA724 were 0.646, 0.529, 0.613, and 0.537, respectively; and the AUCs of PLT, D-dimer, PT and FIB were 0.514, 0.551, 0.630, and 0.610, respectively (Fig. 2a). We selected the four metrics with the largest AUC and according to the Youden Index, the cut-off values of CEA, CA199, PT and FIB were 5 ng/mL, 20.6 U/mL, 13 s and 3.8 g/L, respectively. Then we analyzed the survival differences at different levels of these indicators, the results showed that they all were significant (all $P$ values $<$ 0.05) (Fig. 2b–2f). This means that these four indicators can clearly distinguish patient prognosis.

Table 2
The association of clinicopathologic characteristics and demographics with the research indicators

	CEA (ng/mL)			CA199 (U/mL)			PT (s)			FIB (g/L)			Differentiation
	$<$ 5	$\geqslant$ 5	$P$	$<$ 20.6	$\geqslant$ 20.6	$P$	$<$ 13	$\geqslant$ 13	$P$	$<$ 3.8	$\geqslant$ 3.8	$P$	Well	Poorly	$P$
Gender			0.929			0.328			0.909			0.039			0.714
Female	81	147		148	80		137	91		133	95		185	43
Male	112	200		215	97		189	123		154	158		257	55
Age (years)			0.563			0.039			0.561			0.003			0.100
$<$ 60	62	120		133	49		113	69		113	69		142	40
$\geqslant$ 60	131	227		230	128		213	145		174	184		300	58
Location			0.698			0.481			0.346			0.024			0.009
Colon	163	291		308	146		278	176		230	224		363	91
Rectum	29	57		55	31		48	38		55	31		79	7
T stage			$<$ 0.001			$<$ 0.001			0.010			0.370			0.019
T1-T3	97	108		158	47		138	67		114	91		178	27
T4	96	239		205	130		188	147		173	162		264	71
N stage			$<$ 0.001			$<$ 0.001			0.035			0.040			0.001
N-	138	174		240	70		199	111		153	157		269	41
N+	57	173		123	107		127	103		134	96		173	57
TNM stage			$<$ 0.001			$<$ 0.001			0.046			$<$ 0.001			$<$ 0.001
stage I	43	23		55	11		47	19		46	20		63	3
stage II	93	151		185	59		152	92		107	137		206	38
stage III	57	173		123	107		127	103		134	96		173	57
Vascular invasions			0.032			$<$ 0.001			0.535			0.115			$<$ 0.001
No	98	143		184	57		149	92		119	122		213	28
Yes	95	204		179	120		177	122		168	131		229	70
Neural invasion			$<$ 0.001			0.045			0.854			0.980			0.112
No	67	72		103	36		83	56		74	65		120	19
Yes	126	275		260	141		243	158		213	188		322	79
Recurrence or metastasis			$<$ 0.001			$<$ 0.001			$<$ 0.001			$<$ 0.001			$<$ 0.001
No	158	209		272	95		250	117		215	152		318	49
Yes	35	138		91	82		76	97		72	101		124	49

CEA, carcinoembryonic antigen; CA199, carbohydrate antigen 199; PT, prothrombin time; FIB, fibrinogen.

Figure 2.

Variable screening based on ROC curves and Kaplan-Meier curves. (a) ROC curves of CEA, CA199, PT and FIB. Kaplan-Meier curves of 5-y CRR according to preoperative levels of CEA (b), CA199 (c), PT (d), FIB (e) and tumor differentiation (f). Abbreviations: ROC, receiver-operator characteristic; CEA, carcinoembryonic antigen; CA199, carbohydrate antigen 199; PT, prothrombin time; FIB, fibrinogen; 5-y CRR, 5-year cumulative recurrence rate; DFS, disease-free survival; 95%CI, 95% confidence interval.

The poorly tumor differentiation, elevation of CEA, CA199, PT and FIB was associated with recurrence and metastasis (all $P$ values $<$ 0.001) (Table 2). In addition, the univariate analysis results showed that tumor differentiation, T stage, N stage, TNM stage, vascular invasion, neural invasion, preoperative CEA, CA199, PT and FIB were prognostic factors for 5-yCRR and DFS. The multivariate analysis showed that tumor differentiation, TNM stage, CEA, CA199, PT and FIB were independent prognostic factors (Table 3). Therefore, we choose them as the research indicators to construct the RRS system and prognostic prediction system.

Table 3

Univariate and multivariate Cox proportional hazard analysis for DFS

Variables	HR	95% CI		$P$
		Lower bound	Upper bound
Univariate analysis
Gender (Male/Female)	0.941	0.697	1.271	0.692
Age ( $<$ 60/ $\geqslant$ 60)	1.243	0.895	1.727	0.195
Location (Colon/rectum)	1.335	0.913	1.952	0.136
Differentiation (well/poorly)	2.088	1.499	2.908	$<$ 0.001
TNM stage
Stage I	1 (Referent)
Stage II	2.254	1.025	4.954	0.043
Stage III	6.238	2.888	13.472	$<$ 0.001
Vascular invasion (-/+)	2.303	1.662	3.191	$<$ 0.001
Neural invasion (-/+)	2.017	1.365	2.980	$<$ 0.001
CEA ( $<$ 5/ $\geqslant$ 5)	2.681	1.848	3.891	$<$ 0.001
CA199 ( $<$ 20.6/ $\geqslant$ 20.6)	2.209	1.638	2.978	$<$ 0.001
PT ( $<$ 13/ $\geqslant$ 13)	1.753	1.296	2.370	$<$ 0.001
FIB ( $<$ 3.8/ $\geqslant$ 3.8)	1.472	1.087	1.992	0.012
Multivariate analysis: Method $=$ Enter
Differentiation (well/poorly)	1.741	1.235	2.452	0.002
TNM stage				0.001
Stage I	1 (Referent)
Stage II	1.341	0.581	3.094	0.492
Stage III	2.862	1.193	6.864	0.018
CEA ( $<$ 5/ $\geqslant$ 5)	2.163	1.470	3.185	$<$ 0.001
CA199 ( $<$ 20.6/ $\geqslant$ 20.6)	1.406	1.026	1.929	0.034
PT ( $<$ 13/ $\geqslant$ 13)	1.375	1.010	1.873	0.043
FIB ( $<$ 3.8/ $\geqslant$ 3.8)	1.455	1.059	2.000	0.021

CEA, carcinoembryonic antigen; CA199, carbohydrate antigen 199; PT, prothrombin time; FIB, fibrinogen; HR, Hazard Ratio; 95%CI, 95% confidence interval.

3.3 Construction of RRS system based on independent prognostic factors

The RRS system was composed of the total scores of the five independent prognostic factors mentioned above, and each positive independent prognostic factor gets 1 point. Consequently, the enrolled patients were assigned a score from 0 to 5 (six groups) based on the number of positive indicators, we performed a survival analysis for these six groups, the results of log-rank test indicated non-significant difference in prognosis between scores 0 and 1 ( $P=$ 0.41), scores 2 and 3 ( $P=$ 0.71), scores 4 and 5 ( $P=$ 0.22) (Supplementary Table 1). Thus, based on the RRS system, patients with a score of 0 to 1 were then assigned to the low-risk group ( $n=$ 184), the mid-risk group ( $n=$ 289) included those patients with a score of 2&3, and the patients in high-risk group had a score from 4–5 ( $n=$ 67). We applied Kaplan-Meier survival curves to assess the effects of different risk levels on postoperative recurrence of CRC. The 5-y CRR of mid-risk group was significantly higher than low-level group (40.9% vs. 13.9%, $P<$ 0.001), and the DFS was significantly shorter (75.5 vs. 101.1 months, $P<$ 0.001); and the 5-y CRR and DFS of high-risk group were significantly different with those of mid-risk ( $P<$ 0.001) (Fig. 3a). The above results indicate that RRS system can be used to accurately evaluate the high-risk recurrence population after CRC surgery.

Figure 3.

Kaplan-Meier curves of 5-y CRR according to different systems. (a) Kaplan-Meier curves of 5-y CRR according to the RRS system. (b) Kaplan-Meier curves of 5-y CRR according to the TNM staging system. (c) Kaplan-Meier curves of 5-y CRR according to the various risk levels in early-stage. (d) Kaplan-Meier curves of 5-y CRR according to the various risk levels in advanced-stage. (e, f) Kaplan-Meier curves of 5-y CRR according to the various risk levels in each TNM stage. Abbreviations: 5-y CRR, 5-year cumulative recurrence rate; DFS, disease-free survival; 95%CI, 95% confidence interval; RRS, recurrence risk scoring.

3.4 Predictive value of TNM staging system in CRC recurrence

The prognosis of patients with different stages was significantly different ( $P<$ 0.05), the results showed that 5-y CRR and DFS for TNM stages I, II, and III were 12.7% and 104.8 months, 26.5% and 89.3 months, and 55.5% and 57.3 months, respectively (Fig. 3b). Given the small number of patients with stage I in this study, we defined TNM stage I and II as early-stage CRC and TNM stage III as advanced-stage CRC. The 5-y CRR of advanced-stage group was 55.5%, which was significantly higher than that of early-stage group (23.7%; $P<$ 0.001), and the DFS in advanced-stage group significantly shorter than in early-stage group (57.3 vs 92.9 months). This suggests that TNM staging is also a good differentiator of patient prognosis.

3.5 Accurate screening of high-risk recurrent CRC patients at all stages of TNM using the RRS system as a biomarker

Prognostic differences at different risk levels for each TNM stage were also explored in depth. For early-stage CRC patients, the 5-y CRR and DFS in low-risk, medium-risk and high-risk groups were 10.5% and 104.3 months, 25.9% and 91.9 months, and 60.9% and 51.4 months, respectively (Fig. 3c). In advanced-stage CRC patients, the 5-y CRR and DFS in low-risk, medium-risk and high-risk groups were 21.8% and 86.7 months, 61.3% and 52.3 months, and 76.4% and 32.2 months, respectively (Fig. 3d). Interestingly, early-stage patients with higher risk had a worse prognosis than advanced-stage patients with lower risk. The differences in 5-y CRR and DFS between each risk level within early group were obvious (all $P$ value $<$ 0.05) (Fig. 3c). The prognoses of mid-risk patients with advanced-stage were significantly worse than those of low-risk patients with advanced-stage ( $P<$ 0.001), and better than those of high-risk patients with advanced-stage ( $P=$ 0.01) (Fig. 3d). The prognosis of early-stage patients with mid-risk was even comparable to that of advanced-stage patients with low-risk ( $P=$ 0.75) (Fig. 3e), and the prognosis of early-stage patients with high-risk is even as poor as that of advanced-stage patients with mid-risk ( $P=$ 0.72) (Fig. 3f). These results suggest that the prognosis of patients within the same TNM stage varies significantly and that the combination of RRS system and TNM staging system can better differentiate patients with different prognoses.

3.6 Construction and evaluation of a combined RRS-TNM model

Based on the results of the multivariate COX regression analysis, we formed the RRS-TNM model with all six independent prognostic factors, which is a combination of RRS system and TNM staging system. A nomogram was constructed for stage I-III CRC (Fig. 4a), calibration curves showed favorable constancy between the prediction of nomogram and actual conditions (Fig. 4b). The risk score distribution implied that this RRS-TNM model can well distinguish the prognosis of patients with different risk scores (Fig. 4c). The C-index value of the RRS-TNM model was 0.721 (95%CI, 0.701–0.742) and the AUC was the maximum in these three systems (0.770; 95%CI, 0.727–0.812), which was markedly higher than that of TNM staging system (0.666; 95%CI, 0.618–0.714; $P<$ 0.001) and RRS system (0.713; 95%CI, 0.667–0.759; $P<$ 0.001) (Fig. 4d), indicating positive discrimination. We further compared the net benefit of RRS-TNM model and TNM staging system using decision curve analysis, the results suggested that RRS-TNM model had a better net benefit than TNM staging system in both the 3-y CRR (Fig. 4e) and the 5-y CRR (Fig. 4f).

Figure 4.

Evaluation of the prognostic prediction model. (a) A nomogram for prognostic prediction of stage I to III CRC patients who underwent curative resection. (b) Calibration curves of 3-y and 5-y CRR for stage I to III CRC. (c) The risk score distribution and CRR status of this predictive model. (d) ROC curves of TNM staging system, RRS system and prognostic prediction model, and the comparison of AUCs of them. (e) Decision curve analysis for TNM staging system and predictive model (3-y CRR). (f) Decision curve analysis for TNM staging system and predictive model (5-y CRR). Abbreviations: ROC, receiver-operator characteristic; CEA, carcinoembryonic antigen; CA199, carbohydrate antigen 199; PT, prothrombin time; FIB, fibrinogen; 5-y CRR, 5-year cumulative recurrence rate; 95%CI, 95% confidence interval; AUC, Area under ROC curve; RRS, recurrence risk scoring.

4. Discussion

The accuracy of the TNM staging system, which consists of tumor infiltration depth, lymph node metastasis, and distant metastasis, is limited by the level of medical personnel and medical equipment. The differences in prognoses of CRC patients with the same TNM stage strongly appeal a novel staging system or an effective supplement to the TNM stage to achieve more accurate classification [31]. The progression of CRC may be accompanied by elevated levels of tumor markers and anomalous coagulation. Tumor differentiation, preoperative CEA, CA199, PT and FIB levels showed some effect in differentiating recurrent and metastases after radical resection in CRC patients. Many studies have suggested that pretreatment tumor markers including CEA and CA199 are prognostic indicators of CRC [12, 16, 17, 18, 32, 33, 34], these biomarkers can provide a certain predictive value for postoperative recurrence and metastasis of CRC; moreover, coagulation indexes are also associated with prognoses of patients [19, 21, 22, 23, 35, 36, 37]; but they don’t have satisfactory sensitivity and specificity.

In response to this clinical question, we selected several indicators significantly relating to prognosis to form a RRS system, then used this system to accurately classify patients with various prognoses, and combined with the TNM staging system to achieve higher accuracy for prognostic prediction.

In our study, we found that pathological grade of tumor differentiation, preoperative levels of CEA, CA199, PT and FIB were closely related to the lymph node metastasis, TNM stage and postoperative recurrence (all $P$ values $<$ 0.05). The results of variable filtering pointed out that they were the independent prognostic factors for postoperative recurrence of CRC, which is consistent with previous reports [19, 33, 34, 37]. Be based on this, we assigned scores to five independent prognostic factors to construct RRS system. The AUC of this RRS system was slightly, but not significantly, higher than TNM staging system (0.713 vs. 0.660; $P=$ 0.09), suggesting that these two systems were comparable in their performance in differentiating the prognosis of CRC patients.

Then we evaluated the prognostic differences between different risk levels. The results of statistical analysis indicated that patients with high-risk had a significantly worse prognosis than mid-risk patients ( $P<$ 0.001), and mid-risk patients had a visibly higher risk of recurrence than low-risk patients ( $P<$ 0.001). At the same time, in the subgroup analysis of early-stage and advanced-stage, the prognosis of patients with various risk levels also showed the same conclusion. Among early-stage patients, the 5-year CRR was nearly 6 times higher in the high-risk group than that in the low-risk group, so this result had important implications for identifying high-risk population among the early-stage patients. In advanced-stage patients, prognostic differences between each risk-level were also quite pronounced, suggesting that advanced high-risk patients usually had a poor prognosis. To summarize, the RRS system not only accurately distinguishes patients with different prognosis, but also, in combination with the TNM staging system, recognizes the differences in prognosis among patients with each TNM stage, which helps to better manage patients (more accurate preoperative staging, more appropriate treatment option, more individualized postoperative treatment, and follow-up plan). Accurate staging of early-stage CRC patients is particularly important, both to identify high-risk patients so that more aggressive treatment regimens can be administered to ensure a lower risk of postoperative recurrence and adequate and effective follow-up to ensure a lower risk of postoperative recurrence, and to enable differentiation of low-risk patients to avoid excessive medical intervention. This RRS system has an appreciable discrimination in early-stage CRC, with significant prognostic differences between each risk level, indicating that this system can meet the above requirements.

We also used all six independent prognostic factors to compose an RRS-TNM model. The nomogram, as a predictive method, was widely used in recent years [38, 39, 40], we utilized nomogram to visualize survival analysis results. The C-index of RRS-TNM model was 0.721 and the AUC was 0.770, whereas, the AUCs of TNM staging system and RRS system was 0.666 and 0.713, respectively, which were both lower than that of RRS-TNM model. Moreover, the smooth calibration curves showed the consistency between predicted and actual probabilities, and decision curve analysis of the 3-y and 5-y CRR for both TNM staging system and RRS-TNM model also demonstrated a higher net benefit for the RRS-TNM model, indicating that using this model for prognosis prediction can achieve relatively realistic results. It also illustrates that the TNM staging system together with our constructed RRS system can significantly improve the accuracy of TNM staging system for prognosis prediction.

TNM stage is an indispensable parameter for describing the cancer. [8]; and the postoperative pathology reports include items on the tumor differentiation grade; CEA, CA199, PT, and FIB are common entries in preoperative examinations. Therefore, this model was constructed without the need to perform additional tests on the patients. Moreover, the levels of CEA, CA199, PT and FIB are objective data and are not influenced by the subjective judgement of medical staff, making this system more reproducible and actionable, and making the system a powerful complement to the TNM staging system. In addition, the RRS system, which consists of preoperative objective indicators, can be used to predict patient prognosis before surgery and serve as a good supplement to preoperative staging to guide the development of treatment plans. Although some literature has shown the value of these metrics, combining them into one model to predict prognosis is the first of its kind.

In a nutshell, the RRS system constructed in this study can be used clinically to supplement and modify the TNM staging system, and the prognostic model can be used after surgery to predict the prognosis of patients in order to develop corresponding treatment and follow-up plans.

Our study has certain limitations. First, the number of cases included in this study was small and the study was a single-center retrospective cohort study, and second, the recurrence rate of stage I CRC patients in this study was slightly higher than previously reported [41], which we attribute to the instability of the results due to the small sample. Therefore, this research work provides clinicians with a novel model, which has reference significance for the evaluation and prediction of stage I to III CRC prognosis, but its effectiveness and reliability need to be verified by further large sample studies.

5. Conclusion

In conclusion, pathological degree of tumor differentiation and the preoperative levels of tumor markers (CEA and CA199) and coagulation indexes (PT and FIB) are positively related to the prognosis of stage I to III CRC. The performance of RRS system is comparable to that of TNM staging system, demonstrating that the combination of these five indicators can provide an accurate classification of CRC patients’ prognoses. The prognostic model constructed from RRS and TNM staging system can accurately predict the prognosis of patients, which provides a better predictive value than TNM staging system.

Author contribution

Conception: JCC, YYZ, and KW.

Interpretation or analysis of data: KW, LLM, LYC, and YTJ.

Preparation of the manuscript: KW, YYZ, and LLM.

Revision for important intellectual content: KW and YYZ.

Supervision: JCC and YYZ.

Ethical approval

The Ethics Review Committee of Zhongshan Hospital of Xiamen University approved this study (ZSYY20191102).

Supplementary data

The supplementary files are available to download from https://dx-doi-org.web.bisu.edu.cn/10.3233/CBM-230116.

sj-pdf-1-cbm-10.3233_CBM-230116.pdf - Supplemental material

Supplemental material, sj-pdf-1-cbm-10.3233_CBM-230116.pdf

Footnotes

Acknowledgments

This work was supported by the National Natural Science Foundation of China (81502039), the Natural Science Foundation of Fujian Province (2020J011214), the Health and Family Planning Commission of Fujian Province for Youth Research Project (2015-2-49), the Scientific Research Foundation for Returned Scholars from Ministry of Human Resources and Social Security (2015142), and the Digestive Center of Xiamen University Foundation. Thanks to Dr. Xujin Wei for his great support.

Conflict of interest

The authors declare that they have no competing interests.

References

Sung

Ferlay

Siegel

R.L.

Laversanne

Soerjomataram

Jemal

and Bray

, Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries, CA Cancer J Clin 71 (2021), 209–249. doi: 10.3322/caac.21660.

Chen

Zheng

Baade

P.D.

Zhang

Zeng

Bray

Jemal

X.Q.

and He

, Cancer statistics in China, 2015, CA Cancer J Clin 66 (2016), 115–32. doi: 10.3322/caac.21338.

Benson

A.B.

, 3rd Venook

A.P.

Cederquist

Chan

Chen

Y.J.

Cooper

H.S.

Deming

Engstrom

P.F.

Enzinger

P.C.

Fichera

Grem

J.L.

Grothey

Hochster

H.S.

Hoffe

Hunt

Kamel

Kirilcuk

Krishnamurthi

Messersmith

W.A.

Mulcahy

M.F.

Murphy

J.D.

Nurkin

Saltz

Sharma

Shibata

Skibber

J.M.

Sofocleous

C.T.

Stoffel

E.M.

Stotsky-Himelfarb

Willett

C.G.

C.S.

Gregory

K.M.

and Freedman-Cass

, Colon Cancer, Version 1.2017, NCCN Clinical Practice Guidelines in Oncology, J Natl Compr Canc Netw 15 (2017), 370–398. doi: 10.6004/jnccn.2017.0036.

Schmoll

H.J.

Van Cutsem

Stein

Valentini

Glimelius

Haustermans

Nordlinger

, van de Velde Balmana

Regula

Nagtegaal

I.D.

Beets-Tan

R.G.

Arnold

Ciardiello

Hoff

Kerr

Kohne

C.H.

Labianca

Price

Scheithauer

Sobrero

Tabernero

Aderka

Barroso

Bodoky

Douillard

J.Y.

El Ghazaly

Gallardo

Garin

Glynne-Jones

Jordan

Meshcheryakov

Papamichail

Pfeiffer

Souglakos

Turhal

and Cervantes

, ESMO Consensus Guidelines for management of patients with colon and rectal a personalized approach to clinical decision making, Ann Oncol 23 (2012), 2479–2516. doi: 10.1093/annonc/mds236.

Edge

S.B.

and Compton

C.C.

, The American Joint Committee on Cancer: the 7th edition of the AJCC cancer staging manual and the future of TNM, Ann Surg Oncol 17 (2010), 1471–4. doi: 10.1245/s10434-010-0985-4.

Chen

J.H.

Zhai

E.T.

Yuan

Y.J.

K.M.

J.B.

Peng

J.J.

Chen

C.Q.

Y.L.

and Cai

S.R.

, Systemic immune-inflammation index for predicting prognosis of colorectal cancer, World J Gastroenterol 23 (2017), 6261–6272. doi: 10.3748/wjg.v23.i34.6261.

Fan

X.J.

Wan

X.B.

X.H.

P.H.

Chen

D.K.

Wang

P.N.

Jiang

Wang

D.H.

Chen

Z.T.

Huang

Wang

J.P.

and Wang

, Phosphorylated p38, a negative prognostic biomarker, complements TNM staging prognostication in colorectal cancer, Tumour Biol 35 (2014), 10487–95. doi: 10.1007/s13277-014-2320-3.

Nagtegaal

I.D.

Quirke

and Schmoll

H.J.

, Has the new TNM classification for colorectal cancer improved care? Nat Rev Clin Oncol 9 (2011), 119–23. doi: 10.1038/nrclinonc.2011.157.

Dahl

Myklebust

M.P.

Dale

J.E.

Leon

Serup-Hansen

Jakobsen

Pfeiffer

Loes

I.M.

Pfeffer

Spindler

K.G.

Guren

M.G.

Glimelius

and Johnsson

, Evaluation of the stage classification of anal cancer by the TNM 8th version versus the TNM 7th version, Acta Oncol 59 (2020), 1016–1023. doi: 10.1080/0284186X.2020.1778180.

10.

Park

C.H.

Lizarraga

A.D.

Lee

Y.T.

Yoon

K.J.

and Yoo

T.K.

, Increased Carcinoembryonic Antigen (CEA) Level Is Highly Associated with Low Skeletal Muscle Mass in Asymptomatic Adults: A Population-Based Study, J Clin Med 11 (2022). doi: 10.3390/jcm11175009.

11.

Luo

Jin

Deng

Cheng

Fan

Gong

Qian

Huang

Liu

and Yu

, Roles of CA19-9 in pancreatic cancer: Biomarker, predictor and promoter, Biochim Biophys Acta Rev Cancer 1875 (2021), 188409. doi: 10.1016/j.bbcan.2020.188409.

12.

Baqar

A.R.

Wilkins

Staples

Angus Lee

C.H.

Oliva

and McMurrick

, The role of preoperative CEA in the management of colorectal cancer: A cohort study from two cancer centres, Int J Surg 64 (2019), 10–15. doi: 10.1016/j.ijsu.2019.02.014.

13.

Liu

Ding

Cai

and Zhang

, Prothrombin time (PT) and CEA as prognostic predictive biomarkers for postoperative recurrence after curative resection in patients with stage I-III colorectal cancer: a retrospective cohort study, Updates Surg 74 (2022), 999–1009. doi: 10.1007/s13304-022-01268-8.

14.

Becerra

A.Z.

Probst

C.P.

Tejani

M.A.

Aquina

C.T.

Gonzalez

M.G.

Hensley

B.J.

Noyes

Monson

J.R.

and Fleming

F.J.

, Evaluating the Prognostic Role of Elevated Preoperative Carcinoembryonic Antigen Levels in Colon Cancer Patients: Results from the National Cancer Database, Ann Surg Oncol 23 (2016), 1554–61. doi: 10.1245/s10434-015-5014-1.

15.

Das

Kalita

and Pal

, Predictive and prognostic biomarkers in colorectal cancer: A systematic review of recent advances and challenges, Biomed Pharmacother 87 (2017), 8–19. doi: 10.1016/j.biopha.2016.12.064.

16.

Grunnet

and Sorensen

J.B.

, Carcinoembryonic antigen (CEA) as tumor marker in lung cancer, Lung Cancer 76 (2012), 138–43. doi: 10.1016/j.lungcan.2011.11.012.

17.

Gao

Wang

Zhou

Sheng

Qian

S.Y.

and Huo

, Evaluation of Serum CEA, CA19-9, CA72-4, CA125 and Ferritin as Diagnostic Markers and Factors of Clinical Parameters for Colorectal Cancer, Sci Rep 8 (2018), 2732. doi: 10.1038/s41598-018-21048-y.

18.

Stiksma

Grootendorst

D.C.

and van der Linden

P.W.

, CA 19-9 as a marker in addition to CEA to monitor colorectal cancer, Clin Colorectal Cancer 13 (2014), 239–44. doi: 10.1016/j.clcc.2014.09.004.

19.

Zhang

Luo

Kuang

Mao

Dai

and Wang

, Prediction of Poor Outcomes in Patients with Colorectal Cancer: Elevated Preoperative Prothrombin Time (PT) and Activated Partial Thromboplastin Time (APTT), Cancer Manag Res 12 (2020), 5373–5384. doi: 10.2147/CMAR.S246695.

20.

Giessen

Nagel

Glas

Spelsberg

Lau-Werner

Modest

D.P.

Michl

Heinemann

Stieber

and Schulz

, Evaluation of preoperative serum markers for individual patient prognosis in stage I-III rectal cancer, Tumour Biol 35 (2014), 10237–48. doi: 10.1007/s13277-014-2338-6.

21.

Lee

Huh

S.J.

S.Y.

Koh

M.S.

Kim

S.H.

Lee

J.H.

Han

J.Y.

Choi

H.J.

Kim

S.J.

and Kim

H.J.

, Clinical significance of coagulation factors in operable colorectal cancer, Oncol Lett 13 (2017), 4669–4674. doi: 10.3892/ol.2017.6058.

22.

Tang

Liu

Wang

Zhao

and Liu

, High preoperative plasma fibrinogen levels are associated with distant metastases and impaired prognosis after curative resection in patients with colorectal cancer, J Surg Oncol 102 (2010), 428–32. doi: 10.1002/jso.21668.

23.

Sun

Han

Song

Gao

Yang

Wang

and Wang

, Prognostic Value of Preoperative Fibrinogen for Predicting Clinical Outcome in Patients with Nonmetastatic Colorectal Cancer, Cancer Manag Res 12 (2020), 13301–13309. doi: 10.2147/CMAR.S275498.

24.

Derwinger

Kodeda

Bexe-Lindskog

and Taflin

, Tumour differentiation grade is associated with TNM staging and the risk of node metastasis in colorectal cancer, Acta Oncol 49 (2010), 57–62. doi: 10.3109/02841860903334411.

25.

Nicholson

B.D.

Shinkins

Pathiraja

Roberts

N.W.

James

T.J.

Mallett

Perera

Primrose

J.N.

and Mant

, Blood CEA levels for detecting recurrent colorectal cancer, Cochrane Database Syst Rev 2015 (2015), CD011134. doi: 10.1002/14651858.CD011134.pub2.

26.

Sorensen

C.G.

Karlsson

W.K.

Pommergaard

H.C.

Burcharth

and Rosenberg

, The diagnostic accuracy of carcinoembryonic antigen to detect colorectal cancer recurrence – A systematic review, Int J Surg 25 (2016), 134–44. doi: 10.1016/j.ijsu.2015.11.065.

27.

Wang

Tian

Wang

Sun

Shi

Zhao

and Jing

, The diagnostic value of serum tumor markers CEA, CA19-9, CA125, CA15-3, and TPS in metastatic breast cancer, Clin Chim Acta 470 (2017), 51–55. doi: 10.1016/j.cca.2017.04.023.

28.

Bagaria

Sood

Sharma

and Lalwani

, Comparative study of CEA and CA19-9 in esophageal, gastric and colon cancers individually and in combination (ROC curve analysis), Cancer Biol Med 10 (2013), 148–57. doi: 10.7497/j.issn.2095-3941.2013.03.005.

29.

Lakemeyer

Sander

Wittau

Henne-Bruns

Kornmann

and Lemke

, Diagnostic and Prognostic Value of CEA and CA19-9 in Colorectal Cancer, Diseases 9 (2021). doi: 10.3390/diseases9010021.

30.

von Elm

Altman

D.G.

Egger

Pocock

S.J.

Gotzsche

P.C.

Vandenbroucke

J.P.

and Initiative

, The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies, J Clin Epidemiol 61 (2008), 344–9. doi: 10.1016/j.jclinepi.2007.11.008.

31.

Bruni

Angell

H.K.

and Galon

, The immune contexture and Immunoscore in cancer prognosis and therapeutic efficacy, Nat Rev Cancer 20 (2020), 662–680. doi: 10.1038/s41568-020-0285-7.

32.

You

Sheng

Yan

Chen

Gong

Zheng

Chen

Wang

Tan

Xie

and Wang

, The difference in prognosis of stage II and III colorectal cancer based on preoperative serum tumor markers, J Cancer 10 (2019), 3757–3766. doi: 10.7150/jca.31660.

33.

Takakura

Ikeda

Imaoka

Urushihara

and Itamoto

, An elevated preoperative serum carbohydrate antigen 19-9 level is a significant predictor for peritoneal dissemination and poor survival in colorectal cancer, Colorectal Dis 17 (2015), 417–25. doi: 10.1111/codi.12865.

34.

Huang

S.H.

Tsai

W.S.

You

J.F.

Hung

H.Y.

Yeh

C.Y.

Hsieh

P.S.

Chiang

S.F.

Lai

C.C.

Chiang

J.M.

Tang

and Chen

J.S.

, Preoperative Carcinoembryonic Antigen as a Poor Prognostic Factor in Stage I-III Colorectal Cancer After Curative-Intent Resection: A Propensity Score Matching Analysis, Ann Surg Oncol 26 (2019), 1685–1694. doi: 10.1245/s10434-019-07184-3.

35.

Son

H.J.

Park

J.W.

Chang

H.J.

Kim

D.Y.

Kim

B.C.

Kim

S.Y.

Park

S.C.

Choi

H.S.

and Oh

J.H.

, Preoperative plasma hyperfibrinogenemia is predictive of poor prognosis in patients with nonmetastatic colon cancer, Ann Surg Oncol 20 (2013), 2908–13. doi: 10.1245/s10434-013-2968-8.

36.

Lin

Liu

Qiu

Zhang

Liang

Chen

Qin

and Zou

, Clinical significance of plasma D-dimer and fibrinogen in digestive cancer: A systematic review and meta-analysis, Eur J Surg Oncol 44 (2018), 1494–1503. doi: 10.1016/j.ejso.2018.07.052.

37.

Hong

Shen

Chen

and Hua

, Preoperative plasma fibrinogen, but not D-dimer might represent a prognostic factor in non-metastatic colorectal cancer: A prospective cohort study, Cancer Biomark 19 (2017), 103–111. doi: 10.3233/CBM-160510.

38.

Balachandran

V.P.

Gonen

Smith

J.J.

and DeMatteo

R.P.

, Nomograms in oncology: more than meets the eye, Lancet Oncol 16 (2015), e173–80. doi: 10.1016/S1470-2045(14)71116-7.

39.

Pan

Y.X.

Chen

J.C.

Fang

A.P.

Wang

X.H.

Chen

J.B.

Wang

J.C.

Y.Z.

Chen

M.S.

Zhang

Y.J.

Q.J.

and Zhou

Z.G.

, A nomogram predicting the recurrence of hepatocellular carcinoma in patients after laparoscopic hepatectomy, Cancer Commun (Lond) 39 (2019), 55. doi: 10.1186/s40880-019-0404-6.

40.

Zhang

Chen

and Song

, A nomogram for predicting overall survival in patients with low-grade endometrial stromal sarcoma: A population-based analysis, Cancer Commun (Lond) 40 (2020), 301–312. doi: 10.1002/cac2.12067.

41.

Boland

C.R.

and Goel

, Prognostic Subgroups among Patients with Stage II Colon Cancer, N Engl J Med 374 (2016), 277–8. doi: 10.1056/NEJMe1514353.

The clinical value of a nomogram constructed from CEA,CA199,PT,FIB,tumor differentiation and TNM stage in colorectal cancer

Abstract

BACKGROUND:

OBJECTIVE:

METHODS:

RESULTS:

CONCLUSIONS:

Keywords

1. Introduction

2. Materials and methods

2.1 Patients

2.3 Statistical analysis

3. Results

3.1 Clinicopathological characteristics of the included patients

Table 1 Clinicopathologic features and demographics of the included patients

Table 2 The association of clinicopathologic characteristics and demographics with the research indicators

3.5 Accurate screening of high-risk recurrent CRC patients at all stages of TNM using the RRS system as a biomarker

3.6 Construction and evaluation of a combined RRS-TNM model

5. Conclusion

Author contribution

Ethical approval

Supplementary data

sj-pdf-1-cbm-10.3233_CBM-230116.pdf - Supplemental material

Footnotes

Acknowledgments

Conflict of interest

References

Table 1
Clinicopathologic features and demographics of the included patients

Table 2
The association of clinicopathologic characteristics and demographics with the research indicators