BPG is committed to discovery and dissemination of knowledge
Retrospective Cohort Study Open Access
Copyright: ©Author(s) 2026. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution-NonCommercial (CC BY-NC 4.0) license. No commercial re-use. See permissions. Published by Baishideng Publishing Group Inc.
World J Clin Oncol. Nov 24, 2026; 17(11): 125117
Published online Nov 24, 2026. doi: 10.5306/wjco.125117
Development and internal validation of a disease-free survival model after radical resection of pathological T4 colorectal adenocarcinoma
Qing-Yuan Zeng, Xue-Lian Wu, Rong-Peng Chen, Guo-Qiang Wang, Department of Gastrointestinal Surgery, The Second Affiliated Hospital of Guangzhou Medical University, Guangzhou 510260, Guangdong Province, China
Qian-Ru Hang, Jiangsu Province Key Laboratory of Anesthesiology and Brain Science, Xuzhou Medical University, Xuzhou 221004, Jiangsu Province, China
ORCID number: Qian-Ru Hang (0009-0009-4578-5810); Rong-Peng Chen (0009-0000-6690-4457); Guo-Qiang Wang (0000-0002-5172-3749).
Co-first authors: Qing-Yuan Zeng and Xue-Lian Wu.
Co-corresponding authors: Rong-Peng Chen and Guo-Qiang Wang.
Author contributions: Zeng QY and Wu XL contributed equally to this work; Wang GQ and Chen RP designed the study; Chen RP, Zeng QY, and Wu XL collected and analyzed the data; Hang QR contributed to the statistical methodology and prepared the figures and tables; Zeng QY and Wu XL drafted the manuscript; Wang GQ and Chen RP critically revised the manuscript; Wang GQ reviewed the study methodology and findings; all authors interpreted the results, read and approved the final manuscript, and agree to be accountable for all aspects of the work. Wang GQ and Chen RP both contributed substantially to the study design and critical revision of the manuscript and share responsibility for the scientific content. Wang GQ additionally reviewed the study methodology and findings and serves as the contact for editorial correspondence.
AI contribution statement: ChatGPT was used solely for language refinement and improvement of the readability of the manuscript. All AI-generated suggestions were carefully reviewed and verified by the authors.
Institutional review board statement: The study was reviewed and approved by the Ethics Committee of The Second Affiliated Hospital of Guangzhou Medical University (Approval No. LYZX-2026-135-01).
Informed consent statement: The requirement for written informed consent was waived by the ethics committee because this retrospective study analyzed existing clinical data and involved no additional intervention.
Conflict-of-interest statement: All authors report no relevant conflicts of interest for this article.
STROBE statement: The authors have read the STROBE Statement-checklist of items, and the manuscript was prepared and revised according to the STROBE Statement-checklist of items.
Data sharing statement: De-identified analytic data may be made available by the corresponding author upon reasonable request and subject to institutional approval.
Corresponding author: Guo-Qiang Wang, MD, Department of Gastrointestinal Surgery, The Second Affiliated Hospital of Guangzhou Medical University, No. 250 Changgang East Road, Haizhu District, Guangzhou 510260, Guangdong Province, China. wgqjyh@126.com
Received: July 1, 2026
Revised: August 19, 2026
Accepted: October 22, 2026
Published online: November 24, 2026
Processing time: 140 Days and 21.5 Hours

Abstract
BACKGROUND

Postoperative outcomes vary substantially among patients with pathological T4 (pT4) colorectal adenocarcinoma despite radical resection. Pathological staging may not fully capture risk heterogeneity within this high-risk population.

AIM

To develop and internally validate a disease-free survival (DFS) prediction model after radical resection of pT4 colorectal adenocarcinoma.

METHODS

This single-center retrospective cohort included 440 of 690 patients screened from 2018 to 2022. Least absolute shrinkage and selection operator-Cox regression was used for predictor selection, followed by unpenalized Cox refitting in the full cohort. Conditional optimism correction used 1000 bootstrap resamples with selected terms fixed. Apparent time-dependent discrimination, calibration, and decision curves were evaluated at 36 months and 60 months. A TNM-only Cox comparator based on pT4 subtype and pN stage was used.

RESULTS

During a median follow-up of 51 months, 159 patients (36.1%) experienced a DFS event. Seven clinical variables represented by eight regression terms were retained. The apparent concordance index was 0.670 [95% confidence interval (CI): 0.628-0.711] and 0.652 after conditional optimism correction. Apparent time-dependent areas under the curve were 0.704 (95%CI: 0.650-0.756) at 36 months and 0.717 (95%CI: 0.644-0.780) at 60 months. The TNM-only comparator had apparent and conditional optimism-corrected C-indices of 0.609 and 0.598, respectively. Apparent calibration showed generally close agreement between predicted and observed risks, and decision curves suggested potential net benefit across portions of the evaluated threshold range.

CONCLUSION

The proposed model provided more detailed postoperative risk characterization than TNM staging alone, with moderate discrimination and reasonable apparent calibration. External validation is required before clinical use.

Key Words: Colorectal neoplasms; Disease-free survival; Prognostic model; Pathological T4; Nomogram; Internal validation

Core Tip: This single-center retrospective cohort study developed and internally validated a postoperative disease-free survival model for 440 patients with radically resected pathological T4 colorectal adenocarcinoma. Seven routinely available clinical and pathological variables were retained. The conditional optimism-corrected concordance index was 0.652, and the model showed higher discrimination than a TNM-only comparator within the development cohort. Apparent calibration was generally close, and decision-curve analysis suggested potential net benefit. The complete prediction equation and nomogram are provided. External validation and clinical-impact assessment are required before clinical use.



INTRODUCTION

Colorectal cancer is among the most frequently diagnosed malignancies worldwide and remains a leading cause of cancer-related death[1,2]. Radical resection is the cornerstone of curative-intent treatment for non-metastatic disease, with perioperative therapy tailored to tumor stage and clinical risk[3-7]. Nevertheless, postoperative recurrence, metastasis, and death remain important concerns. Disease-free survival (DFS) is therefore a clinically relevant endpoint for assessing prognosis after resection.

Patients with pathological T4 (pT4) colorectal cancer represent a high-risk but heterogeneous population. This category includes tumors that penetrate the visceral peritoneum (pT4a) or directly invade adjacent organs or structures (pT4b)[8]. Although these tumors share a locally advanced T classification, postoperative outcomes vary according to nodal involvement, pathological features, and patient characteristics[9,10]. Pathological stage remains fundamental to prognostic assessment, but may not fully characterize risk heterogeneity among patients with pT4 disease.

Routinely recorded clinical and pathological variables may provide complementary prognostic information. Perineural and lymphovascular invasion are associated with adverse outcomes[11,12], while preoperative carcinoembryonic antigen (CEA) and carbohydrate antigen 19-9 (CA19-9) have been investigated as prognostic markers[13]. Variation in tumor biology and metastatic potential further contributes to the heterogeneous clinical behavior of colorectal cancer[14,15]. Integrating several routinely available features may therefore provide more individualized postoperative risk estimates than anatomical stage alone.

Previous postoperative colorectal cancer prediction models have demonstrated the feasibility of integrating multiple clinical and pathological factors[16,17]. However, these models were developed in broader colorectal cancer populations, and their performance cannot be assumed to apply directly to a specifically defined pT4 cohort. Developing a model within this high-risk population may help characterize prognostic variation that remains after conventional pathological staging.

Accordingly, this study aimed to develop and internally validate a model for predicting 36-month and 60-month DFS after radical resection of pT4 colorectal adenocarcinoma using routinely recorded clinical and pathological variables. Least absolute shrinkage and selection operator (LASSO) selection was followed by Cox model refitting in the full analytic cohort. Model performance was assessed in terms of discrimination, calibration, and potential net benefit, and was compared with a TNM-only model. A complete prediction equation and nomogram were provided to facilitate individual risk estimation and subsequent independent validation.

MATERIALS AND METHODS
Patients and study design

This retrospective cohort study included patients with pT4 colorectal adenocarcinoma who underwent radical resection at The Second Affiliated Hospital of Guangzhou Medical University between January 2018 and December 2022. Of 690 screened patients, 121 were excluded for peritoneal or distant metastasis, 25 for emergency or non-radical/palliative surgery, and 104 for missing essential clinical data or insufficient follow-up information required for model analysis or ascertainment of DFS time or event status. The final cohort included 440 patients (Figure 1). No statistical imputation was performed; model development was based on patients with the required predictor information and ascertainable DFS outcomes.

Figure 1
Figure 1 Study flow diagram. LASSO: Least absolute shrinkage and selection operator; pT4: Pathological T4.

The Ethics Committee of The Second Affiliated Hospital of Guangzhou Medical University approved the retrospective study in 2026 (approval No. LYZX-2026-135-01) and waived the requirement for written informed consent.

Data collection and outcome

Seventeen candidate variables were evaluated: Age, sex, body mass index, American Society of Anesthesiologists (ASA) classification, tumor site, preoperative bowel obstruction, CEA, CA19-9, albumin, neutrophil-to-lymphocyte ratio (NLR), pT4 subtype, pN stage, tumor size, histological type, differentiation, lymphovascular invasion, and perineural invasion. Laboratory values were recorded preoperatively. Categorical groupings are shown in Table 1, and coding and parameterization of the final model terms are provided in Supplementary Table 1.

Table 1 Baseline characteristics by disease-free survival event status, n (%).
Characteristic
All patients (n = 440)
No DFS event (n = 281)
DFS event (n = 159)
P value
Age, years64.00 (56.00, 70.00)64.00 (58.00, 70.00)65.00 (55.50, 72.00)0.398
Sex
    Male253 (57.5)174 (61.9)79 (49.7)0.017
    Female187 (42.5)107 (38.1)80 (50.3)
Body mass index22.79 (20.66, 24.77)22.59 (20.59, 24.67)23.03 (20.80, 24.90)0.262
ASA classification
    I-II367 (83.4)246 (87.5)121 (76.1)0.003
    III or higher73 (16.6)35 (12.5)38 (23.9)
Tumor site
    Right-sided colon111 (25.2)66 (23.5)45 (28.3)0.465
    Left-sided colon112 (25.5)71 (25.3)41 (25.8)
    Rectum217 (49.3)144 (51.2)73 (45.9)
Preoperative bowel obstruction
    No416 (94.5)271 (96.4)145 (91.2)0.035
    Yes24 (5.5)10 (3.6)14 (8.8)
Preoperative CEA, ng/mL3.84 (2.08, 9.93)3.46 (2.00, 8.33)4.42 (2.25, 11.34)0.065
Preoperative CA19-9, U/mL6.53 (2.50, 16.66)5.63 (2.44, 13.37)8.72 (2.79, 28.22)0.009
Preoperative albumin, g/L38.00 (35.00, 40.40)37.90 (34.90, 40.20)38.30 (35.25, 40.70)0.428
Neutrophil-to-lymphocyte ratio2.04 (1.55, 2.92)2.02 (1.51, 2.86)2.07 (1.59, 3.08)0.255
pT4 subtype
    pT4a329 (74.8)215 (76.5)114 (71.7)0.316
    pT4b111 (25.2)66 (23.5)45 (28.3)
pN stage
    pN0155 (35.2)120 (42.7)35 (22.0)< 0.001
    pN1179 (40.7)106 (37.7)73 (45.9)
    pN2106 (24.1)55 (19.6)51 (32.1)
Tumor size
    ≤ 5 cm274 (62.3)172 (61.2)102 (64.2)0.611
    > 5 cm166 (37.7)109 (38.8)57 (35.8)
Histological type
    Conventional adenocarcinoma405 (92.0)263 (93.6)142 (89.3)0.244
    Mucinous adenocarcinoma17 (3.9)8 (2.8)9 (5.7)
    Signet-ring cell carcinoma or other special type18 (4.1)10 (3.6)8 (5.0)
Differentiation grade
    Moderate/well differentiation365 (83.0)244 (86.8)121 (76.1)0.006
    Poor differentiation75 (17.0)37 (13.2)38 (23.9)
Lymphovascular invasion
    Negative278 (63.2)191 (68.0)87 (54.7)0.008
    Positive162 (36.8)90 (32.0)72 (45.3)
Perineural invasion
    Negative287 (65.2)192 (68.3)95 (59.7)0.087
    Positive153 (34.8)89 (31.7)64 (40.3)

Follow-up was conducted until June 2025. DFS was defined as the interval from radical resection to the first recorded recurrence, metastasis, or death from any cause. Patients without an event were censored at their last recorded disease-free follow-up. Median follow-up was estimated using the reverse Kaplan-Meier method.

Model development and evaluation

CEA, CA19-9, and NLR were transformed using ln(x + 1), and categorical variables were represented by indicator terms, yielding 20 candidate regression terms. After standardization, LASSO-Cox regression was used for variable selection[18]. The penalty was selected by ten-fold cross-validation stratified by DFS-event status, using the one-standard-error rule. Selected terms were refitted in an unpenalized Cox model on their original encoded scales using the full cohort. A nomogram was constructed to estimate 36-month and 60-month DFS event risks.

Discrimination was assessed using Harrell’s concordance index (C-index) and time-dependent area under the curve (AUC)[19]. C-index optimism correction used 1000 bootstrap resamples, with the selected terms held fixed and coefficients re-estimated. Calibration plots compared mean predicted risks with Kaplan-Meier estimates of observed event risk within quintiles of predicted risk at 36 months and 60 months[20,21].

Decision-curve analysis compared net benefit for the proposed model, a TNM-only Cox comparator containing pT4 subtype and pN stage, and treat-all and treat-none strategies[22]. Time-dependent AUCs and decision curves used inverse-probability-of-censoring weighting. Inverse probability of censoring weighting (IPCW) weights were based on Kaplan-Meier estimates of the censoring distribution. AUC confidence intervals (CI) were obtained from 1000 bootstrap resamples with censoring weights re-estimated. Calibration intercepts and slopes were estimated using IPCW complementary-log-log recalibration, with 95%CI obtained from 1000 bootstrap resamples with predictions fixed. Proportional-hazards assumptions were assessed using term-specific rank-transformed Schoenfeld tests. Selection stability was assessed using 500 bootstrap LASSO fits at the original selected penalty. AUCs, calibration plots, and decision curves reflected apparent performance in the development cohort; C-index optimism correction was conditional on the selected terms. Detailed calibration results, proportional-hazards diagnostics, selection-stability results, and sensitivity analyses are reported in Supplementary Tables 2-5.

Patients were divided into lower- and higher-risk groups using the median-centered linear predictor. Kaplan-Meier curves were compared using the log-rank test to assess risk-group separation within the development cohort.

Statistical analysis

Continuous variables were summarized as medians with interquartile ranges and categorical variables as numbers and percentages. Baseline characteristics were compared descriptively by DFS-event status using the Mann-Whitney U test or χ2 test, as appropriate. Baseline P values were not used for predictor selection. Analyses were performed using Python. Two-sided P values < 0.05 were considered statistically significant.

RESULTS
Patient characteristics

A total of 440 patients were included, comprising 223 patients with colon cancer and 217 with rectal cancer. There were 253 men (57.5%) and 187 women (42.5%). Baseline characteristics according to DFS-event status are summarized in Table 1.

The median follow-up estimated using the reverse Kaplan-Meier method was 51 months. During follow-up, 159 patients (36.1%) experienced a DFS event, and 281 (63.9%) were censored. Recorded outcome components included recurrence in 130 patients, distant metastasis in 123, local recurrence in 29, and death in 90; these categories overlapped.

Predictor selection and model development

At the one-standard-error penalty of 0.05118, LASSO selection retained seven clinical variables represented by eight regression terms: CA19-9, sex, ASA classification, preoperative bowel obstruction, differentiation, lymphovascular invasion, and pN stage (Figure 2). The selected terms were refitted in an unpenalized Cox model, with coefficients, hazard ratios, and 95%CIs presented in Table 2.

Figure 2
Figure 2 Least absolute shrinkage and selection operator-Cox variable selection. A: Ten-fold cross-validation with standard-error bars and the minimum-error and one-standard-error penalties; B: Standardized coefficient paths.
Table 2 Final Cox model after least absolute shrinkage and selection operator selection.
Term or comparison
Coefficient
SE
HR (95%CI)
P value
ln(CA19-9 + 1)0.1220320.0500611.130 (1.024-1.246)0.015
Female vs male0.3092780.1625031.362 (0.991-1.873)0.057
ASA III or higher vs ASA I-II0.4726200.1953951.604 (1.094-2.353)0.016
Preoperative bowel obstruction: Yes vs no0.4233220.3016511.527 (0.845-2.758)0.161
Poor vs moderate/well differentiation0.3868380.1913371.472 (1.012-2.142)0.043
Lymphovascular invasion: Positive vs negative0.2498500.1704851.284 (0.919-1.793)0.143
pN1 vs pN00.5659600.2116121.761 (1.163-2.666)0.007
pN2 vs pN00.6964160.2289782.007 (1.281-3.143)0.002

Proportional-hazards testing indicated a departure for lymphovascular invasion (P = 0.009); results for all final model terms are provided in Supplementary Table 3.

Model performance

The apparent C-index was 0.670 (95%CI: 0.628-0.711), and the conditional bootstrap optimism-corrected C-index was 0.652. Apparent time-dependent AUCs were 0.704 (95%CI: 0.650-0.756) at 36 months and 0.717 (95%CI: 0.644-0.780) at 60 months (Figure 3 and Table 3).

Figure 3
Figure 3 Apparent time-dependent receiver operating characteristic curves at 36 months and 60 months. AUC: Area under the receiver operating characteristic curve; CI: Confidence interval.
Table 3 Discrimination of the proposed model and TNM-only comparator.
Model
Measure
Estimate
95%CI
Proposed modelHarrell C-index0.6700.628-0.711
Conditional bootstrap optimism-corrected C-index0.652Not estimated
Time-dependent AUC at 36 months0.7040.650-0.756
Time-dependent AUC at 60 months0.7170.644-0.780
TNM-only comparatorHarrell C-index0.6090.567-0.650
Conditional bootstrap optimism-corrected C-index0.598Not estimated
Time-dependent AUC at 36 months0.6210.564-0.677
Time-dependent AUC at 60 months0.5420.458-0.626

Apparent calibration plots showed generally close agreement between predicted and observed DFS event risks at both time points (Figure 4A and B). Calibration intercepts and slopes are reported in Supplementary Table 2. Decision curves suggested potential net benefit over treat-all and treat-none strategies across portions of the evaluated threshold range (Figure 4C and D).

Figure 4
Figure 4 Apparent calibration and decision curves. A: Calibration at 36 months; B: Calibration at 60 months. Predicted disease-free survival event risks are compared with observed risks by risk quintile, and dashed lines indicate ideal calibration; C: Decision-curve analysis at 36 months; D: Decision-curve analysis at 60 months. Decision curves compare the proposed model, the TNM-only comparator, and treat-all and treat-none strategies. The TNM-only comparator includes pathological T4 subtype and pN stage. DFS: Disease-free survival.

The TNM-only comparator had an apparent C-index of 0.609 and a conditional optimism-corrected C-index of 0.598, with apparent AUCs of 0.621 and 0.542 at 36 months and 60 months, respectively (Table 3). Paired bootstrap differences favored the proposed model: 0.061 (95%CI: 0.026-0.096) for the C-index, 0.083 (95%CI: 0.035-0.133) for the 36-month AUC, and 0.175 (95%CI: 0.113-0.238) for the 60-month AUC (Supplementary Table 5).

Risk stratification and nomogram

Using the median centered linear predictor of -0.004203, patients were divided into lower- and higher-risk groups of 220 patients each. The higher-risk group had poorer DFS than the lower-risk group within the development cohort (log-rank P < 0.001; Figure 5).

Figure 5
Figure 5 Kaplan-Meier disease-free survival curves for groups defined by the median centered linear predictor. Shading indicates pointwise 95% confidence intervals.

A nomogram incorporating the seven selected clinical variables was constructed to estimate individual DFS event risks at 36 months and 60 months (Figure 6).

Figure 6
Figure 6 Nomogram for 36-month and 60-month disease-free survival event risk. Total predictor points are mapped to the 36-month and 60-month disease-free survival event-risk axes. The nomogram represents all eight regression terms across seven clinical-variable axes; pN stage is encoded by separate pN1 and pN2 indicator terms, with pN0 as the reference. DFS: Disease-free survival; ASA: American Society of Anesthesiologists; CA19-9: Carbohydrate antigen 19-9.
DISCUSSION

This study developed and internally validated a postoperative DFS prediction model specifically for patients with radically resected pT4 colorectal adenocarcinoma. By integrating seven routinely available clinical and pathological variables, the model captured prognostic heterogeneity that was not fully reflected by pT4 and pN stage alone. The conditional optimism-corrected C-index was 0.652, indicating moderate discrimination. Comparisons of the C-index and time-dependent AUCs consistently favored the proposed model over the TNM-only comparator, while calibration and decision-curve analyses further supported its ability to characterize risk within the development cohort. These findings suggest that routinely collected information beyond anatomical stage can contribute to more individualized postoperative risk assessment in this high-risk population.

The selected variables represent complementary dimensions of postoperative prognosis. Nodal stage reflects regional disease burden, whereas differentiation and lymphovascular invasion characterize pathological aggressiveness; these features are consistent with the prognostic importance of nodal involvement and adverse pathology reported in T4 and locally advanced colon cancer[9,10,23]. Preoperative CA19-9 provides additional information from a circulating tumor marker and has previously been associated with recurrence after colorectal cancer resection[13]. ASA classification and preoperative bowel obstruction extend the model beyond tumor pathology by incorporating the patient's general condition and clinical presentation. Female sex was also retained during model development, although its CI included unity after Cox refitting; therefore, this finding should not be interpreted as establishing female sex as an independent adverse prognostic factor.

Previous postoperative colorectal cancer nomograms have integrated clinical and pathological factors across broader surgical populations and randomized-trial cohorts[16,17], and similar approaches have been reported for rectal cancer[24]. In contrast, the present study focuses specifically on patients with pT4 colorectal adenocarcinoma, a clinically high-risk group that remains heterogeneous despite sharing the same locally advanced T category. This narrower focus allows prognostic variation within pT4 disease to be examined directly rather than inferred from models developed across multiple stages. In addition, the final predictors are routinely available in clinical practice, and the complete prediction equation, predictor coding, baseline survival estimates, and nomogram are provided to facilitate individual risk estimation and subsequent external validation.

Comparison with the TNM-only model provides evidence that the additional clinical and pathological variables contributed prognostic information beyond anatomical staging alone. The paired differences in C-index and time-dependent AUCs consistently favored the proposed model. This finding is plausible because patients with pT4 disease may differ substantially in nodal burden, tumor aggressiveness, physiological condition, clinical presentation, and tumor-marker levels even when classified within the same broad pathological stage. Integrating these features therefore provides a more detailed description of postoperative risk than pT4 and pN stage alone. The model should be considered an adjunct to conventional pathological staging rather than a replacement for it, and the present analysis does not establish superiority over other multivariable prognostic tools that were not evaluated in this cohort.

From a clinical perspective, the nomogram translates the fitted model into individualized estimates of 36-month and 60-month DFS event risk. Such estimates may provide a useful framework for postoperative prognostic discussion if their accuracy is confirmed in independent populations. The separation of Kaplan-Meier curves further demonstrates that the model can stratify patients with different DFS outcomes within the development cohort. Decision-curve analysis also suggested potential net benefit across portions of the evaluated threshold range. These findings support further evaluation of the model for risk-adapted postoperative assessment, although external validation is required before specific treatment or surveillance decisions can be based on its predictions.

Several limitations should be considered. First, the retrospective, single-center design and exclusion of 104 patients because of missing essential clinical variables or insufficient follow-up information may have introduced selection bias. Second, the inclusion of both colon and rectal cancers introduces treatment-related heterogeneity. The study was designed to develop a single model for the overall pT4 colorectal adenocarcinoma cohort rather than separate colon- and rectum-specific models. Chemotherapy regimens, immunotherapy exposure, surgical approach, circumferential margin status, and RAS/BRAF alterations were incompletely characterized, and information on neoadjuvant treatment was limited. These factors may account for additional prognostic variation not captured by the current model and may affect its performance in other populations.

Finally, model performance was assessed within the development cohort. The bootstrap correction re-estimated coefficients while holding the selected terms fixed and therefore did not reproduce the complete model-building process; some optimism may consequently remain. The observed departure from the proportional-hazards assumption for lymphovascular invasion and the greater uncertainty of estimates at 60 months also warrant consideration. Independent external validation is therefore needed to assess discrimination, calibration, and transportability across different patient populations and care settings.

CONCLUSION

The proposed model integrated seven routinely available clinical and pathological variables to characterize postoperative DFS risk in patients with radically resected pT4 colorectal adenocarcinoma. It provided moderate discrimination, reasonable apparent calibration, and better discrimination than a TNM-only comparator within the development cohort. External validation and clinical-impact assessment are required before clinical use.

References
1.  Bray F, Laversanne M, Sung H, Ferlay J, Siegel RL, Soerjomataram I, Jemal A. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2024;74:229-263.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Cited by in Crossref: 16785]  [Cited by in RCA: 18110]  [Article Influence: 9055.0]  [Reference Citation Analysis (55)]
2.  Dekker E, Tanis PJ, Vleugels JLA, Kasi PM, Wallace MB. Colorectal cancer. Lancet. 2019;394:1467-1480.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Cited by in Crossref: 4063]  [Cited by in RCA: 3734]  [Article Influence: 533.4]  [Reference Citation Analysis (40)]
3.  Argilés G, Tabernero J, Labianca R, Hochhauser D, Salazar R, Iveson T, Laurent-Puig P, Quirke P, Yoshino T, Taieb J, Martinelli E, Arnold D; ESMO Guidelines Committee. Localised colon cancer: ESMO Clinical Practice Guidelines for diagnosis, treatment and follow-up. Ann Oncol. 2020;31:1291-1305.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Cited by in Crossref: 1145]  [Cited by in RCA: 1047]  [Article Influence: 174.5]  [Reference Citation Analysis (9)]
4.  Benson AB, Venook AP, Adam M, Chang G, Chen YJ, Ciombor KK, Cohen SA, Cooper HS, Deming D, Garrido-Laguna I, Grem JL, Haste P, Hecht JR, Hoffe S, Hunt S, Hussan H, Johung KL, Joseph N, Kirilcuk N, Krishnamurthi S, Malla M, Maratt JK, Messersmith WA, Meyerhardt J, Miller ED, Mulcahy MF, Nurkin S, Overman MJ, Parikh A, Patel H, Pedersen K, Saltz L, Schneider C, Shibata D, Shogan B, Skibber JM, Sofocleous CT, Tavakkoli A, Willett CG, Wu C, Gurski LA, Snedeker J, Jones F. Colon Cancer, Version 3.2024, NCCN Clinical Practice Guidelines in Oncology. J Natl Compr Canc Netw. 2024;22:e240029.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Cited by in Crossref: 206]  [Cited by in RCA: 260]  [Article Influence: 130.0]  [Reference Citation Analysis (6)]
5.  Benson AB, Venook AP, Adam M, Chang G, Chen YJ, Ciombor KK, Cohen SA, Cooper HS, Deming D, Garrido-Laguna I, Grem JL, Haste P, Hecht JR, Hoffe S, Hunt S, Hussan H, Johung KL, Joseph N, Kirilcuk N, Krishnamurthi S, Malla M, Maratt JK, Messersmith WA, Meyerhardt J, Miller ED, Mulcahy MF, Nurkin S, Parikh A, Patel H, Pedersen K, Saltz L, Schneider C, Shibata D, Shogan B, Skibber JM, Sofocleous CT, Tavakkoli A, Willett CG, Wu C, Jones F, Gurski L. NCCN Guidelines® Insights: Rectal Cancer, Version 3.2024. J Natl Compr Canc Netw. 2024;22:366-375.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Cited by in Crossref: 172]  [Cited by in RCA: 228]  [Article Influence: 114.0]  [Reference Citation Analysis (3)]
6.  Hofheinz RD, Fokas E, Benhaim L, Price TJ, Arnold D, Beets-Tan R, Guren MG, Hospers GAP, Lonardi S, Nagtegaal ID, Perez RO, Cervantes A, Martinelli E; ESMO Guidelines Committee. Localised rectal cancer: ESMO Clinical Practice Guideline for diagnosis, treatment and follow-up. Ann Oncol. 2025;36:1007-1024.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Cited by in Crossref: 8]  [Cited by in RCA: 151]  [Article Influence: 151.0]  [Reference Citation Analysis (3)]
7.  Baxter NN, Kennedy EB, Bergsland E, Berlin J, George TJ, Gill S, Gold PJ, Hantel A, Jones L, Lieu C, Mahmoud N, Morris AM, Ruiz-Garcia E, You YN, Meyerhardt JA. Adjuvant Therapy for Stage II Colon Cancer: ASCO Guideline Update. J Clin Oncol. 2022;40:892-910.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Cited by in Crossref: 253]  [Cited by in RCA: 267]  [Article Influence: 66.8]  [Reference Citation Analysis (9)]
8.  Weiser MR. AJCC 8th Edition: Colorectal Cancer. Ann Surg Oncol. 2018;25:1454-1455.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Cited by in Crossref: 834]  [Cited by in RCA: 779]  [Article Influence: 97.4]  [Reference Citation Analysis (10)]
9.  Eom T, Lee Y, Kim J, Park I, Gwak G, Cho H, Yang K, Kim K, Bae BN. Prognostic Factors Affecting Disease-Free Survival and Overall Survival in T4 Colon Cancer. Ann Coloproctol. 2021;37:259-265.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Full Text (PDF)]  [Cited by in Crossref: 10]  [Cited by in RCA: 19]  [Article Influence: 3.8]  [Reference Citation Analysis (0)]
10.  Kang JH, Son IT, Kim BC, Park JH, Kim JY, Kim JW. Recurrence-Free Survival Outcomes Based on Novel Classification Combining Lymphovascular Invasion, Perineural Invasion, and T4 Status in Stage II-III Colon Cancer. Cancer Manag Res. 2022;14:2031-2040.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Full Text (PDF)]  [Cited by in Crossref: 2]  [Cited by in RCA: 13]  [Article Influence: 3.3]  [Reference Citation Analysis (4)]
11.  Knijn N, Mogk SC, Teerenstra S, Simmer F, Nagtegaal ID. Perineural Invasion is a Strong Prognostic Factor in Colorectal Cancer: A Systematic Review. Am J Surg Pathol. 2016;40:103-112.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Cited by in Crossref: 204]  [Cited by in RCA: 194]  [Article Influence: 19.4]  [Reference Citation Analysis (6)]
12.  Yuan H, Dong Q, Zheng B, Hu X, Xu JB, Tu S. Lymphovascular invasion is a high risk factor for stage I/II colorectal cancer: a systematic review and meta-analysis. Oncotarget. 2017;8:46565-46579.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Full Text (PDF)]  [Cited by in Crossref: 26]  [Cited by in RCA: 48]  [Article Influence: 6.0]  [Reference Citation Analysis (0)]
13.  Gramkow MH, Mosgaard CS, Schou JV, Nordvig EH, Dolin TG, Lykke J, Nielsen DL, Pfeiffer P, Qvortrup C, Yilmaz MK, Larsen O, Bojesen SE, Jensen BV, Johansen JS. The prognostic role of circulating CA19-9 and CEA in patients with colorectal cancer. Cancer Treat Res Commun. 2025;43:100907.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Cited by in Crossref: 4]  [Cited by in RCA: 6]  [Article Influence: 6.0]  [Reference Citation Analysis (3)]
14.  Li J, Ma X, Chakravarti D, Shalapour S, DePinho RA. Genetic and biological hallmarks of colorectal cancer. Genes Dev. 2021;35:787-820.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Full Text (PDF)]  [Cited by in Crossref: 461]  [Cited by in RCA: 449]  [Article Influence: 89.8]  [Reference Citation Analysis (19)]
15.  Tauriello DV, Calon A, Lonardo E, Batlle E. Determinants of metastatic competency in colorectal cancer. Mol Oncol. 2017;11:97-119.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Full Text (PDF)]  [Cited by in Crossref: 207]  [Cited by in RCA: 207]  [Article Influence: 23.0]  [Reference Citation Analysis (8)]
16.  Honda M, Oba K, Akiyoshi T, Maeda H, Kashiwabara K, Kanda M, Mayanagi S, Aoyama T, Hamada C, Sadahiro S, Fukunaga Y, Ueno M, Sakamoto J, Saji S, Yoshikawa T. Development and validation of a prognostic nomogram for colorectal cancer after radical resection based on individual patient data from three large-scale phase III trials. Oncotarget. 2017;8:99150-99160.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Full Text (PDF)]  [Cited by in Crossref: 4]  [Cited by in RCA: 8]  [Article Influence: 0.9]  [Reference Citation Analysis (0)]
17.  Hong T, Cai D, Jin L, Zhang Y, Lu T, Hua D, Wu X. Development and validation of a nomogram to predict survival after curative resection of nonmetastatic colorectal cancer. Cancer Med. 2020;9:4126-4136.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Full Text (PDF)]  [Cited by in Crossref: 14]  [Cited by in RCA: 16]  [Article Influence: 2.7]  [Reference Citation Analysis (0)]
18.  Tibshirani R. The lasso method for variable selection in the Cox model. Stat Med. 1997;16:385-395.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Cited by in RCA: 92]  [Reference Citation Analysis (1)]
19.  Blanche P, Kattan MW, Gerds TA. The c-index is not proper for the evaluation of t-year predicted risks. Biostatistics. 2019;20:347-357.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Cited by in Crossref: 61]  [Cited by in RCA: 114]  [Article Influence: 16.3]  [Reference Citation Analysis (0)]
20.  Austin PC, Harrell FE Jr, van Klaveren D. Graphical calibration curves and the integrated calibration index (ICI) for survival models. Stat Med. 2020;39:2714-2742.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Full Text (PDF)]  [Cited by in Crossref: 87]  [Cited by in RCA: 169]  [Article Influence: 28.2]  [Reference Citation Analysis (0)]
21.  Van Calster B, Nieboer D, Vergouwe Y, De Cock B, Pencina MJ, Steyerberg EW. A calibration hierarchy for risk models was defined: from utopia to empirical data. J Clin Epidemiol. 2016;74:167-176.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Cited by in Crossref: 310]  [Cited by in RCA: 670]  [Article Influence: 67.0]  [Reference Citation Analysis (0)]
22.  Vickers AJ, Van Calster B, Steyerberg EW. Net benefit approaches to the evaluation of prediction models, molecular markers, and diagnostic tests. BMJ. 2016;352:i6.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Full Text (PDF)]  [Cited by in Crossref: 812]  [Cited by in RCA: 841]  [Article Influence: 84.1]  [Reference Citation Analysis (6)]
23.  Kumamoto T, Yamaguchi S, Nakagawa R, Nagashima Y, Maeda F, Tani K, Kondo H, Koshino K, Kaneko Y, Bamba Y, Ogawa S, Inoue Y, Itabashi M. Prognostic risk factors for pT4 colon cancer: A retrospective cohort study. Oncol Lett. 2023;25:29.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Cited by in RCA: 2]  [Reference Citation Analysis (0)]
24.  Sun J, Zhong X, Yin X, Wu H, Li L, Yang R. Construction and validation of a nomogram for predicting disease-free survival after radical resection of rectal cancer using perioperative inflammatory indicators. J Gastrointest Oncol. 2024;15:668-680.  [RCA]  [PubMed]  [DOI]  [Full Text]  [Full Text (PDF)]  [Cited by in Crossref: 2]  [Cited by in RCA: 3]  [Article Influence: 1.5]  [Reference Citation Analysis (1)]
Footnotes

Peer review: Externally peer reviewed.

Peer-review model: Single blind

Specialty type: Oncology

Country of origin: China

Peer-review report’s classification

Scientific quality: Grade B, Grade D

Novelty: Grade B, Grade D

Creativity or innovation: Grade B, Grade D

Scientific significance: Grade B, Grade D

P-Reviewer: Huo Q, Associate Chief Physician, MD, PhD, Postdoc, China; Liu C, Lecturer, China S-Editor: Qu XL L-Editor: A P-Editor: Lei YY

Write to the Help Desk