Risk factors for recurrence and metastasis of stage IIIC colorectal cancer: A two-center retrospective cohort study

Abstract

BACKGROUND

Colorectal cancer (CRC) is a highly common malignancy worldwide. Specifically, stage III CRC patients still have significant risk of recurrence and metastasis after undergoing radical surgery. Currently, prognostic factors for overall stage III CRC are already relatively clear, but research specifically focusing on risk factors for recurrence and metastasis in the stage IIIC subgroup is still limited.

AIM

To investigate the risk factors influencing postoperative recurrence and metastasis in patients with stage IIIC CRC, providing a basis for individualized clinical diagnosis and treatment.

METHODS

A two-center retrospective cohort study was conducted involving 255 patients with pathologically confirmed stage IIIC CRC who underwent curative surgery between July 2019 and June 2022. Clinicopathological data and follow-up outcomes were collected. Multivariate Cox proportional hazards regression models were utilized to identify independent prognostic factors. Disease-free survival (DFS) were estimated using the Kaplan-Meier method and log-rank test.

RESULTS

Multivariate analysis identified preoperative carcinoembryonic antigen (CEA) ≥ 5.0 μg/L (HR = 1.871, 95%CI: 1.267-2.763, P = 0.002), metastatic lymph node ratio (MLR) ≥ 0.51 (HR = 2.132, 95%CI: 1.388-3.275, P = 0.001), and positive root lymph node (PRL; HR = 1.602, 95%CI: 1.072-2.393, P = 0.021) as independent risk factors for diminished DFS. Notably, patients exhibiting all three high-risk factors (CEA ≥ 5.0 μg/L, MLR ≥ 0.51, and PRL) demonstrated a severely reduced median DFS of 16.4 months and a 3-year recurrence rate of 95%.

CONCLUSION

Preoperative CEA ≥ 5.0 μg/L, MLR ≥ 0.51, and PRL are independent risk factors for postoperative recurrence and metastasis in stage IIIC CRC. These clinical determinants are pivotal for identifying high-risk populations and formulating individualized monitoring and treatment strategies.

Key Words: Colorectal cancer; Stage IIIC; Recurrence and metastasis; Metastatic lymph node ratio; Prognostic factors

Core Tip: Patients with stage IIIC colorectal cancer still face a high risk of recurrence and metastasis after undergoing radical surgery. This two-center retrospective study confirmed that preoperative carcinoembryonic antigen ≥ 5.0 μg/L, metastatic lymph node ratio ≥ 0.51, and positive root lymph node metastasis are independent risk factors for recurrence and metastasis. Patients possessing all three risk factors simultaneously constitute a high-risk subgroup. Identifying these key risk factors is conducive to formulating individualized surveillance and treatment strategies.

INTRODUCTION

According to the 2022 global cancer statistics, colorectal cancer (CRC) remains the third most common malignancy worldwide[1]. Based on the 8^th edition of the TNM staging system, released in 2017 by the Union for International Cancer Control and the American Joint Committee on Cancer, stage IIIC CRC encompasses several subcategories, including T4aN2a, T3-T4aN2b, and T4bN1-N2. In contrast, stage IV disease is defined by the presence of distant metastasis, namely M1, regardless of the corresponding T or N classification[2]. Data from the Surveillance, Epidemiology, and End Results database (January 2010 to December 2020) indicate that stage III cases account for 26.38% of all CRC diagnoses, with stage IIIC representing 21.39% of that subset; stage IV cases constitute 20.92% of all CRC cases. Survival statistics show that stage IIIC CRC has a 3-year survival rate of 55.1% and a 5-year survival rate of 42.3%. Conversely, stage IV CRC exhibits a significantly poorer prognosis, with 3-year and 5-year survival rates of only 23.8% and 13.0%. These findings highlight the prognostic disparity between stage IIIC and stage IV CRC.

Radical surgery remains the cornerstone of treatment for stage IIIC CRC. However, despite successful surgical resection, the risk of recurrence and metastasis persists, posing significant challenges to long-term survival. Understanding the risk factors associated with postoperative recurrence and metastasis in stage IIIC CRC is critical for identifying high-risk patients and developing personalized monitoring and intervention strategies. Previous studies have identified several key risk factors for recurrence and metastasis in stage III CRC, including molecular mutations, microsatellite instability status and elevated serum tumor markers [such as carcinoembryonic antigen (CEA) and carbohydrate antigen 19-9 (CA19-9)], among others[3-6]. However, research specifically addressing the risk factors for recurrence and metastasis in stage IIIC CRC remains limited. This study aims to investigate the potential risk factors associated with recurrence and metastasis following radical surgery for stage IIIC CRC, with the goal of identifying high-risk populations and providing evidence for tailored clinical management strategies.

MATERIALS AND METHODS

Data

This two-center retrospective cohort study examined the clinical data of 255 patients with stage IIIC CRC who underwent postoperative pathological confirmation at the Department of Colorectal Surgery of The First Affiliated Hospital of Xiamen University and The First Affiliated Hospital of Fujian Medical University between July 2019 and June 2022.

Inclusion criteria: (1) Pathologically confirmed diagnosis of CRC; (2) Underwent curative surgery for CRC; (3) Pathological TNM staging of IIIC; and (4) Complete clinicopathological and follow-up data available.

Exclusion criteria: (1) Concurrent malignancies or preoperative imaging/colonoscopy indicating multiple origins of CRC; (2) Coexisting severe cardiovascular, respiratory, neurological, or hematological diseases that significantly affect prognosis; (3) Familial adenomatous polyposis with malignant transformation; (4) Positive surgical margins; (5) Received neoadjuvant chemotherapy or radiotherapy prior to surgery; or (6) Lost to follow-up or died due to non-cancer-related accidents.

The study was approved by the Ethics Committee of the First Affiliated Hospital of Xiamen University [Approval No. (2025) Scientific Research Ethics Review (016)].

Observation variables

Patient demographics, intraoperative details, and postoperative adjuvant treatments were recorded. Variables included gender, age, body mass index, preoperative bowel obstruction status, preoperative tumor markers (CEA and CA19-9), preoperative albumin levels, primary tumor location (right colon, left colon, and rectum), pathological T stage, Ki-67 index, vascular invasion, perineural invasion, tumor size, histological differentiation, macroscopic appearance, metastatic lymph node ratio (MLR), calculated as the number of metastatic lymph nodes divided by the total number of dissected lymph nodes, and chemotherapy regimens (XELOX, mFOLFOX6).

For right-sided colon cancer patients, a D3 radical resection was performed, involving the ligation of superior mesenteric artery branches at the left root of the superior mesenteric vein, defining lymph nodes No. 203, No. 213, and No. 223 as root lymph nodes. For left-sided colon cancer, depending on the tumor location (descending colon, descending-sigmoid junction, or sigmoid colon), No. 253 was defined as the root lymph node; if the tumor was located at the splenic flexure, No. 223 and No. 253 were defined as root lymph nodes. For rectal cancer, No. 253 was defined as the root lymph node. Clinically and histologically, a positive root lymph node (PRL) was defined as the presence of one or more metastatic colorectal adenocarcinoma deposits within these designated apical lymph node stations. Regarding the extent of lymphadenectomy, while current guidelines recommend examining a minimum of 12 lymph nodes for adequate staging, achieving this threshold was not a strict inclusion criterion for our study, provided that pathological evaluation definitively confirmed stage IIIC disease. Histological confirmation of PRL was achieved through standard hematoxylin and eosin staining of the dissected lymph nodes. All pathological specimens were evaluated by two independent, experienced gastrointestinal pathologists. Patients were subsequently classified into the “recurrence and metastasis” group (107 cases) and the “no recurrence and metastasis” group (148 cases) based on their follow-up outcomes. A comparative analysis of clinical data was conducted to identify prognostic risk factors.

Follow-up

Patients were followed up every 3 months on average via outpatient visits or telephone contact to obtain postoperative survival and recurrence information. The final follow-up date was December 30, 2025. Freedom from recurrence or metastasis was defined as the absence of any tumor progression from the surgery date to the last follow-up. The primary endpoint of this study was disease-free survival (DFS), defined as the interval from the date of curative surgery to the first documented tumor recurrence or metastasis, the occurrence of a new primary malignancy, or death due to tumor progression. Recurrence and metastasis were confirmed via comprehensive evaluations including chest computed tomography (CT), contrast-enhanced abdominal CT, serum tumor markers, positron emission tomography-CT, and/or pathological biopsy.

Statistical analysis

Analyses were performed using SPSS software (version 29.0). Unordered categorical variables were compared using the χ² test, while ordinal categorical variables were analyzed using the Mann-Whitney U test. Multivariate analysis was performed using Cox proportional hazards regression models. DFS and survival curves were estimated via the Kaplan-Meier method and compared using the log-rank test. Receiver operating characteristic (ROC) curve analysis was performed to evaluate the predictive value of MLR. The optimal cut-off value for MLR was determined to be 0.51 by calculating the maximum Youden’s index (sensitivity + specificity - 1). Variables demonstrating statistical significance (P < 0.05) in univariate analysis were incorporated into a multivariate Cox proportional hazards regression model. A two-sided P value < 0.05 was considered statistically significant.

RESULTS

Univariate analysis of recurrence and metastasis in stage IIIC CRC

Univariate analysis revealed that age ≥ 60 years, CEA ≥ 5.0 μg/L, CA19-9 ≥ 37 kU/L, pathological perineural invasion, T stage, MLR ≥ 0.51, and PRL were significant factors influencing postoperative recurrence and metastasis (P < 0.05; Table 1). MLR was defined as the number of positive lymph nodes divided by the total number of dissected lymph nodes, using the optimal cutoff value of 0.51 determined by the ROC curve analysis (Figure 1).

Figure 1 Receiver operating characteristic curve plotted to discriminate metastatic lymph node ratio. AUC: Area under the curve.

Table 1 Comparison of clinical data between non-recurrence/metastasis group (n = 148) and recurrence/metastasis group (n = 107), n (%).

Variable	Non-recurrence/metastasis (n = 148)	Recurrence/metastasis (n = 107)	P value
Gender
Female	85 (57.4)	54 (50.5)	0.270
Male	63 (42.6)	53 (49.5)
Age (year)
< 60	76 (51.4)	38 (35.5)	0.012
≥ 60	72 (48.6)	69 (64.5)
BMI (kg/m2)
< 18	10 (6.8)	2 (1.9)	0.163
18-23.9	76 (51.4)	54 (50.5)
≥ 24	62 (41.9)	51 (47.7)
Preoperative intestinal obstruction
No	48 (32.4)	24 (22.4)	0.080
Yes	100 (67.6)	83 (77.6)
CEA (μg/L)
< 5	99 (66.9)	49 (45.8)	0.001
≥ 5	49 (33.1)	58 (54.2)
Tumor location
Right colon	28 (18.9)	26 (24.3)	0.141
Left colon	31 (20.9)	30 (28.0)
Rectum	89 (60.1)	51 (47.7)
CA19-9 (kU/L)
< 37	116 (78.4)	66 (61.7)	0.004
≥ 37	32 (21.6)	41 (38.3)
ALB (g/L)
< 35	23 (15.5)	14 (13.1)	0.583
≥ 35	125 (84.5)	93 (86.9)
Pathological Ki-67
Low	76 (51.4)	56 (52.3)	0.877
High	72 (48.6)	51 (47.7)
Vascular invasion
No	75 (50.7)	41 (38.3)	0.051
Yes	73 (49.3)	66 (61.7)
Neural invasion
No	63 (42.6)	32 (29.9)	0.039
Yes	85 (57.4)	75 (70.1)
Root lymph nodes
Negative	109 (73.6)	45 (42.1)	< 0.001
Positive	39 (26.4)	62 (57.9)
Tumor size (cm)
< 5	95 (64.2)	79 (73.8)	0.103
≥ 5	53 (35.8)	28 (26.2)
Histological differentiation
Others	100 (67.6)	69 (64.5)	0.607
Poorly-differentiated	48 (32.4)	38 (35.5)
Pathological appearance
Infiltrative	10 (6.8)	6 (5.6)	0.933
Ulcerative	97 (65.5)	71 (66.4)
Protruding	41 (27.7)	30 (28.0)
MLR
< 0.51	96 (64.9)	40 (37.4)	< 0.001
≥ 0.51	52 (35.1)	67 (62.6)
pT stage
3	73 (49.3)	28 (26.2)	0.001
4a	68 (45.9)	71 (66.4)
4b	7 (4.7)	8 (7.5)
Chemotherapy regimens
Xelox	130 (87.8)	91 (85.0)	0.518
Folfox	18 (12.2)	16 (15.0)

BMI: Body mass index; CEA: Carcinoembryonic antigen; CA19-9: Carbohydrate antigen 19-9; ALB: Preoperative albumin level; MLR: Metastatic lymph node ratio; pT: Pathological T.

Multivariate analysis of recurrence and metastasis in stage IIIC CRC

Variables demonstrating statistical significance in the univariate analysis were incorporated into a Cox proportional hazards regression model. The independent variables included age, CEA, CA19-9, pathological perineural invasion, T stage, MLR, and root lymph node metastasis status. The results identified CEA ≥ 5.0 μg/L (HR = 1.871, 95%CI: 1.267-2.763, P = 0.002), MLR ≥ 0.51 (HR = 2.132, 95%CI: 1.388-3.275, P = 0.001), and PRL (HR = 1.602, 95%CI: 1.072-2.393, P = 0.021) as independent risk factors for recurrence and metastasis in stage IIIC CRC (Table 2).

Table 2 Multivariate analysis of risk factors for postoperative recurrence or metastasis in stage IIIC colorectal cancer.

Variable	B	SE	Wald χ2	P value	HR	95%CI
Age (year)	0.217	0.214	1.031	0.310	1.242	0.817-1.888
CEA (μg/L)	0.626	0.199	9.909	0.002	1.871	1.267-2.763
CA19-9 (kU/L)	0.290	0.203	2.042	0.153	1.337	0.898-1.990
Neural invasion	0.196	0.220	0.797	0.372	1.217	0.791-1.873
PRL	0.471	0.205	5.292	0.021	1.602	1.072-2.393
MLR	0.757	0.219	11.945	0.001	2.132	1.388-3.275
pT stage			5.381	0.068
T4a vs T3	0.532	0.234	5.143	0.023	1.702	1.075-2.694
T4b vs T3	0.556	0.416	1.787	0.181	1.744	0.772-3.944

CEA: Carcinoembryonic antigen; CA19-9: Carbohydrate antigen 19-9; PRL: Positive root lymph node; MLR: Metastatic lymph node ratio; pT: Pathological T.

Survival curve comparison

Survival curve comparisons revealed that the 3-year DFS for patients with CEA ≥ 5.0 μg/L was significantly lower than for those with CEA < 5.0 μg/L (P < 0.001, χ² = 11.396; Figure 2A). Patients with MLR ≥ 0.51 exhibited a significantly lower 3-year DFS compared to those with MLR < 0.51 (P < 0.001, χ² = 17.780; Figure 2B). Similarly, patients with PRL demonstrated a significantly reduced 3-year DFS compared to those with negative root lymph nodes (P < 0.001, χ² = 17.018; Figure 2C).

Figure 2 Kaplan-Meier curves for 3-year disease-free survival of prognostic factors following radical surgery in stage IIIC colorectal cancer. A: Carcinoembryonic antigen (CEA) ≥ 5 µg/L vs CEA < 5 µg/L, P < 0.001; B: Metastatic lymph node ratio (MLR) ≥ 0.51 vs MLR < 0.51, P < 0.001; C: Positive root lymph node vs negative root lymph node, P < 0.001. CEA: Carcinoembryonic antigen; MLR: Metastatic lymph node ratio.

Survival analysis

The 255 included patients had an average follow-up period of 42.8 months. Among the cohort, 148 patients (58.0%) remained free of recurrence or metastasis at 3 years with a 3-year DFS of 32.0 months. Notably, patients presenting with the co-occurrence of all three independent risk factors (CEA ≥ 5.0 μg/L, MLR ≥ 0.51, and PRL) experienced a 3-year recurrence rate of 95.0% and a severely reduced median DFS of only 16.4 months.

DISCUSSION

Recurrence and metastasis following curative surgery for CRC significantly compromises the patient prognosis. This study identified preoperative CEA ≥ 5.0 μg/L, MLR ≥ 0.51, and PRL as independent risk factors and recognizing these factors is instrumental in tailoring individualized treatment and surveillance protocols to optimize overall patient prognosis.

Elevated CEA is a well-established predictor of poor prognosis in CRC[7]. Zhang et al[8] found that patients with preoperative CEA elevation exhibit significantly lower 3-year DFS compared to those with normal baseline levels. CEA levels reflect the biological behavior and aggressiveness of the tumor, where dynamic fluctuations may signal therapeutic evasion and the initiation of metastasis[9]. Furthermore, the combination of CEA with fecal occult blood testing or gut microbiota metabolites can enhance early detection rates[10]. Additionally, integrating circulating tumor cell (CTC) capture with CEA assays has shown promise in identifying subclinical micrometastases, potentially reshaping staging protocols and therapeutic decision-making[11].

Elevated CEA is a well-established predictor of poor prognosis in CRC. Zhang et al[8] demonstrated that patients with preoperative CEA elevation exhibit significantly lower 3-year DFS rates compared to those with normal baseline levels. CEA levels mirror the biological behavior and aggressiveness of the tumor, where dynamic fluctuations may signal therapeutic evasion and the initiation of metastasis. Furthermore, the combination of CEA with fecal occult blood testing or gut microbiota metabolites can enhance early detection rates. Additionally, integrating CTC capture with CEA assays has shown promise in identifying subclinical micrometastases, potentially reshaping staging protocols and therapeutic decision-making.

Lymph node metastasis serves as a critical prognostic determinant in CRC[12]. A Danish nationwide cohort study demonstrated that both the total number of retrieved lymph nodes and the MLR are independent prognostic indicators; notably, MLR out-performed the standard N staging in predicting overall survival for stage III CRC[13]. Given that the retrieved lymph node count is heavily contingent upon surgical technique and the meticulousness of pathological examination, MLR provides a more standardized prognostic assessment[14]. This is particularly relevant as 30%-50% of CRC patients have fewer than 12 harvested lymph nodes, making MLR a vital supplementary metric[3,15,16]. The optimal cutoff value for MLR was determined to be 0.51 via ROC curve analysis. An MLR ≥ 0.51 effectively predicts early recurrence and metastasis in CRC, serving as an independent prognostic indicator. Furthermore, an elevated MLR correlates with mutations in genes such as TP53 and KRAS, which can disrupt cellular signaling cascades, reshape the tumor microenvironment, and bolster the invasive potential of cancer cells[17,18].

Our analysis further identified PRL as an independent risk factor. The infiltration of root lymph nodes signifies a successful breach of both local intestinal architecture and lymphatic barriers, reflecting a sophisticated adaptation to the host immune microenvironment and subsequent evasion of immunosurveillance. Such involvement likely facilitates systemic dissemination through the mesenteric vasculature into the broader circulatory system[19]. This is exemplified by the cohort study of Lee et al[20], which revealed that in sigmoid and rectal cancer, the recurrence and 5-year survival profiles of patients with inferior mesenteric lymph node (station 253) metastasis were virtually identical to those with para-aortic lymph node (M1 stage) metastasis. Such parity suggests that RPL functions not merely as regional spread, but as a surrogate for systemic dissemination. This insight corroborates our hypothesis that the presence of PRL identifies a highly specific subgroup of stage III patients whose tumor burden and recurrence risk are very high. While M1 disease is formally classified as stage IV[21], conventional imaging modalities frequently fail to detect occult peritoneal, hepatic, or nodal metastases in the 5-10 mm range[22]. Consequently, leveraging clinicopathological determinants to predict such subclinical micro-dissemination becomes imperative for accurate risk stratification. High-risk stage IIIC patients in our cohort (CEA ≥ 5.0 μg/L, MLR ≥ 0.51, and PRL) faced a 95% recurrence rate within 3 years, necessitating aggressive therapy and rigorous surveillance. Our data identify a subset of high-risk stage IIIC patients, defined by the concurrence of CEA ≥ 5.0 μg/L, MLR ≥ 0.51, and PRL, who faced an alarmingly high 3-year recurrence rate of 95%. Such a profound risk profile suggests that conventional protocols may be insufficient, necessitating a shift toward aggressive interventional regimens and rigorous long-term monitoring.

This study has several limitations. First, its retrospective design may involve inherent selection bias and uncontrolled confounders. Second, the bi-center, single-country nature of the cohort limits the extrapolation of results to more diverse populations. Third, the minimum follow-up duration was relatively brief, which may have insufficiently captured late-stage recurrences and thus underestimated long-term metastatic rates. Fourth, the MLR cutoff (≥ 0.51) was identified through post-hoc ROC analysis of our data, necessitating further external validation to ensure its broad applicability. Ultimately, prospective multicenter trials are required to optimize and confirm the efficacy of this risk-stratification model.

CONCLUSION

Preoperative CEA ≥ 5.0 μg/L, MLR ≥ 0.51, and PRL are independent risk factors for postoperative recurrence and metastasis in stage IIIC CRC. Patients exhibiting all three factors constitute a high-risk cohort with a 95% 3-year recurrence rate. These clinical determinants are pivotal for identifying high-risk populations and formulating individualized monitoring and treatment strategies.