Prognostic Analysis of Lymphovascular Invasion in Stages I–III Colorectal Cancer: A Retrospective Study Based on Propensity Score Match : American Journal of Clinical Oncology

Secondary Logo

Journal Logo

Original Article

Prognostic Analysis of Lymphovascular Invasion in Stages I–III Colorectal Cancer

A Retrospective Study Based on Propensity Score Match

Lin, Zhuoqun MM; Zheng, Yitao MM; Yang, Jun MM; Jin, Wei MM; Wang, Junqi MM; Wang, Weichen MM; Li, Shaotang MD, PhD

Author Information
American Journal of Clinical Oncology ():10.1097/COC.0000000000001015, May 23, 2023. | DOI: 10.1097/COC.0000000000001015
  • Open
  • PAP


As of 2018, colorectal cancer (CRC) has become the third most common malignancy worldwide and the second leading cause of cancer deaths.1 According to statistics, metastases occur in about 50% to 60% of patients with CRC.2,3 Although an increasing number of patients are receiving surgery and adjuvant therapy, the prognosis of some patients is still inconsistent with the estimates.4 At present, the tumor-lymph node metastasis (TNM) staging system is routinely used to predict the prognosis of patients with clinical CRC and develop postoperative treatment plans. However, this system is usually limited, and the prognosis of patients with the same staging could vary greatly.5 Therefore, more microscopic factors have been identified as high-risk prognostic factors for clinical assistance and personalized prediction of patient outcomes.6,7

Lymphovascular invasion (LVI) refers to the invasion of tumor cells into the microvessels or lymphatics surrounding the lesion, and it is thought to increase the risk of metastasis and spread of tumor.7 In previous reports, the incidence of LVI has fluctuated significantly between 8% and 89.5%.8–10 In the 2020 version of the National Comprehensive Cancer Network guidelines, LVI is considered a high-risk factor for tumor recurrence, so as patients with stage II CRC accompanied by LVI-positive are recommended for postoperative adjuvant chemotherapy.11 However, no sufficient evidence is available to show that the poor prognosis of CRC is directly related to LVI. Some studies found that patients with positive LVI have a worse prognosis than those with negative LVI, but other studies did not find corroborative results.12–15 For example, a 2019 study found that intramural LVI was not a strong predictor of worse tumor outcomes in patients with stage II-III CRC, while another 2010 study rejected the effect of LVI on outcomes in patients with I-II CRC.16,17 We believe this may be due to the size of the population included, the diagnostic techniques of the center, inconsistencies in the classification of LVI, and the fact that these studies did not exclude relevant confounding factors.18

Propensity score matching (PSM) is a statistical method that can control multiple confounding factors. It can express the influence of multiple confounding factors with a comprehensive propensity score to reduce the number of independent variables and achieve a balance of patient characteristics after eliminating some data. Compared with traditional covariates, PSM is not dependent on the number of events and has the advantage of being easy to analyze, present and explain.19–21 At present, we find few studies using PSM to study the relationship between LVI and the prognosis of CRC. Therefore, the present study aimed to investigate the relationship between LVI and prognosis in patients with stages I–III CRC by using propensity score matching (PSM) and drawing a Nomogram as a supplement to TNM staging.


Source and Selection of Patients

A total of 1389 patients who underwent surgical operations for CRC in the Department of Colorectal Surgery at the First Affiliated Hospital of Wenzhou Medical University between March 2018 and February 2020 were selected. The inclusion criteria were as follows: (1) 18 to 80 years; (2) no metastasis (stage IV); and (3) complete clinical data. By using these criteria, 890 patients with stages I–III CRC were identified. The exclusion criteria were as follows: (1) postoperative pathology that did not mention LVI and perineural invasion; (2) previous history of malignant tumor; and (3) 2 or more primary tumors. Ultimately, a total of 610 eligible patients were included in the statistical analysis (Fig. 1).

Consort diagram showing patient selection process.

All patients underwent computed tomography (CT) of the chest and abdomen for preoperative evaluation and pelvic magnetic resonance imaging for rectal cancer. Postoperative pathologic staging was performed in accordance with the American Joint Committee on Cancer Handbook of Cancer Staging (eighth edition), and postoperative chemoradiation was recommended for patients with stages II and III CRC with high-risk factors. However, considering the wishes of patients and their families and the economic situation of the families, some patients did not receive chemoradiation. Therefore, the item of postoperative chemoradiation was included in this study as a variable in the statistics to exclude corresponding confounding factors.

All patients were first reviewed in the outpatient clinic 2 weeks after surgery, quarterly for 2 years after surgery, and then every 6 months, usually in the outpatient clinic or by telephone. Most patients received serum carcinoembryonic antigen (CEA) levels at the follow-up to assess disease changes, and abdominal CT or ultrasound was used for imaging. Patients with suspected metastasis or recurrence were further evaluated by local enhanced CT or whole-body positron emission CT. All characteristic data of patients were collected from existing cases, including pathologic features such as LVI. Hematoxylin and eosin staining combined with immunohistochemistry is routinely used for diagnosis, undoubtedly improving the accuracy of LVI diagnosis in the center and further reducing confusion. Since it is a retrospective study, informed consent was not required, and this study was approved by the Ethics Committee of the First Affiliated Hospital of Wenzhou Medical University.

The study endpoints were disease-free survival (DFS) and overall survival (OS) in accordance with whether patients had pathologically confirmed LVI. OS denotes the time between surgery and death from any cause, and DFS refers to the time from surgical resection to tumor recurrence or death.

Statistical Analysis

All statistical analyses were performed on SPSS 26.0 and R 4.2.1. Mann-Whitney U test, Pearson χ2 test, and Yates’s corrected χ2 test were used to describe the baseline differences between groups. PSM was used to balance patient characteristics between groups to reduce the effect of potential confounding factors. Ten variables, including age, sex, CEA level, perineural invasion, tumor location, T stage, N stage, histologic grade, neoadjuvant chemoradiotherapy, and chemoradiation, were included in the analysis. CEA level refers to the nearest preoperative serum CEA index. LVI, perineural invasion, T stage, N stage, and histologic grade were determined by postoperative pathology. Kaplan-Meier method and log-rank test were used to calculate the survival rate before and after matching. Cox proportional hazards models were used to determine the prognostic factors of patients with CRC. Variables with P value less than 0.1 in univariate analysis were gradually backward selected into multivariate analysis, and a nomogram was drawn in accordance with the independent prognostic factors obtained from the multivariate analysis of patients before matching. The accuracy of the nomogram was evaluated by C-index, receiver operating characteristic curve, and calibration curve. A P value less than 0.05 was considered statistically significant.


Characteristics of Patients Grouped in Accordance With the Presence of LVI Before and After Matching

Among the 610 patients with stages I–III CRC, 150 (24.6%) were found to have LVI. First, univariate analysis of patients with LVI (n=150) and without LVI (n=460) showed that the former had higher CEA levels (P=0.001), higher T stage (P <0.001), higher N stage (P <0.001), higher CEA levels (P <0.001), and worse histologic grade (P <0.001) and were more likely to suffer from perineural invasion (P <0.001) than the latter. Moreover, patients with positive LVI were more likely to receive adjuvant chemotherapy after surgery than those with negative LVI (86.0% vs. 50.0%; P <0.001). However, no significant differences were found in terms of age, sex, tumor location, and neoadjuvant chemoradiotherapy.

Subsequently, a 1:1 PSM yielded 120 couples of patients with CRC whose characteristics were balanced, and no differences were observed in the univariate analyses of each variable (Table 1).

TABLE 1 - Characteristics of Patients With Stages I–III Colorectal Cancer in Accordance With Lymphovascular Invasion
Before PSM After PSM
Characteristic LVI(+)(n=150)* LVI(-)(n=460)* P-Value LVI(+)(n=120)* LVI(-)(n=120)* P-Value
Sex, n (%) 0.445 0.791
 Male 91 (60.7) 295 (64.1) 74 (61.7) 72 (60.0)
 Female 85 (39.3) 165 (35.9) 46 (38.3) 48 (40.0)
 Age (y) 65.0 (56.0,72.0) 64.5 (55.0,71.0) 0.552 66.0 (56.0,72.0) 65.0 (55.0,72.0) 0.858
 CEA level (ng/l) 3.90 (2.30,8.10) 2.90 (1.70,5.70) 0.001 3.90 (2.40,7.69) 4.40 (2.43,10.43) 0.561
Tumor location, n (%) 0.597 0.883
 Left colon 23 (15.3) 86 (18.7) 22 (18.3) 25 (20.8)
 Right colon 23 (15.3) 62 (13.5) 18 (15.0) 18 (15.0)
 Rectum 104 (69.3) 312 (67.8) 80 (66.7) 77 (64.2)
T stage, n (%) <0.001 0.233
 T1-T2 15 (10.0) 217 (47.2) 14 (11.7) 13 (10.8)
 T3 109 (72.7) 225 (48.9) 84 (70.0) 94 (78.3)
 T4 26 (17.3) 18 (3.9) 22 (18.3) 13 (10.8)
N stage, n (%) <0.001 0.182
 N0 26 (17.3) 353 (76.7) 26 (21.7) 35 (29.2)
 N1–N2 124 (82.7) 107 (23.3) 94 (78.3) 85 (70.8)
Histologic grade, n (%) <0.001 0.054
 WD/MD 95 (63.3) 396 (86.1) 74 (61.7) 88 (73.3)
 PD/MUC/SRC 55 (36.7) 64 (13.9) 46 (38.3) 32 (26.7)
Perineural invasion, n (%) <0.001 0.506
 No 45 (30.0) 320 (69.6) 43 (35.8) 48 (40.0)
 Yes 105 (70.0) 140 (30.4) 77 (64.2) 72 (60.0)
Neoadjuvant chemoradiotherapy, n (%) 0.926 0.499
 No 146 (97.3) 445 (96.7) 117 (97.5) 114 (95.0)
 Yes 4 (2.7) 15 (3.3) 3 (2.5) 6 (5.0)
Chemoradiation, n (%) <0.001 0.861
 No 21 (14.0) 230 (50.0) 20 (16.7) 19 (15.8)
 Yes 129 (86.0) 230 (50.0) 100 (83.3) 101 (84.2)
*n (%); Median (IQR).
Mann-Whitney U test; Pearson’s χ2 test; Yates’s corrected χ2 test.
CEA indicates carcinoembryonic antigen; LVI, lymphovascular invasion; MD, moderately differentiated; MUC, mucinous carcinoma; PD, poorly differentiated; PSM, propensity score matching; SRC, signet ring cell carcinoma; WD, well differentiated.

Relationship Between LVI and Survival

Among the 610 patients, 61 died (10.0%), and 138 had tumor recurrence or metastasis (22.6%) during follow-up, with a median follow-up of 162 weeks. Before matching, the 3-year OS was 75.4% (95% CI=67.7%–83.9%) for patients with positive LVI and 93.2% (95% CI=90.8%–95.7%) for patients with negative LVI (P <0.0001); the corresponding 3-year DFS rates were 61.2% (95% CI=53.3%–70.2%) and 83.5% (95% CI=80.1%–87.1%, P <0.0001), respectively. After matching, the 3-year OS was 72.3% (95% CI=63.5%–82.4%) for patients with positive LVI and 86.5% (95% CI=80.3%–93.1%) for patients with negative LVI (P=0.0091); the corresponding 3-year DFS rates were 57.9% (95% CI=49.1%–68.3%) and 73.2% (95% CI=65.4%–81.9%, P=0.011; Fig. 2), respectively.

Survival curves for patients in accordance with lymphovascular invasion. (A) 3-year OS before matching; (B) 3-year DFS before matching; (C) 3-year OS after matching; (D) 3-year DFS after matching. DFS indicates disease-free survival; OS, overall survival; WK, weeks.

The Cox proportional hazards model for matched patients showed that those with LVI had a 2.115 times higher risk of death than those without LVI (95% CI=1.122–3.988, P=0.021), and the risk of tumor recurrence or metastasis was 1.695 times that of negative patients (95% CI=1.087–2.641; P=0.020; Table 2).

TABLE 2 - Prognostic Factors in Matched Patients in Accordance with Lymphovascular Invasion
Overall survival Disease-free survival
Univariate Multivariate Univariate Multivariate
Factors P HR (95% CI) P P HR (95% CI) P
 Male versus female 0.827 0.167
 Age 0.024 1.038 (1.005–1.073) 0.022 0.554
 CEA level 0.002 1.012 (1.004–1.020) 0.002 0.031
Tumor location
 Left colon versus rectum 0.681 0.480
 Right colon versus rectum 0.986 0.828
T stage
 T1–T2 versus T3 0.219 2.415 (0.573–10.181) 0.230 0.123
 T1–T2 versus T4 0.020 6.299 (1.403–28.277) 0.016 0.027
N stage
 N0 versus N1–N2 0.051 <0.001 3.846 (1.852-7.985) <0.001
Histological grade
 WD/MD versus PD/MUC/SRC 0.293 0.753
Lymphovascular invasion
 Yes versus no 0.011 2.115 (1.122–3.988) 0.021 0.013 1.695 (1.087–2.641) 0.020
Perineural invasion
 Yes versus no 0.717 0.203
Neoadjuvant chemoradiotherapy
 Yes versus no 0.475 0.567
 Yes versus no 0.013 0.272
CEA indicates carcinoembryonic antigen; CI, confidence interval; HR, hazard ratio; MD, moderately differentiated; MUC, mucinous carcinoma; PD, poorly differentiated; SRC, signet ring cell carcinoma; WD, well differentiated.

The Cox proportional hazards model was used for statistical analysis of patients before matching. Univariate analysis showed that age (HR=1.050, 95%CI=1.021–1.080, P=0.001), CEA level (HR=1.014, 95%CI=1.008-1.020, P<0.001), T stage (T1–2 VS. T3: HR=2.681, 95%CI=1.370-5.247, P=0.004; T1-2 VS. T4: HR=7.988, 95%CI=3.518–18.137, P<0.001), N stage (HR=4.333, 95%CI=2.517–7.458, P<0.001), histologic grade (HR=2.344, 95%CI=1.373–4.004, P=0.002), LVI (HR=4.371, 95%CI=2.634–7.254, P<0.001), and perineural invasion (HR=2.068, 95%CI=1.249–3.426, P=0.005) were prognostic factors for 3-year OS in patients with stage I–III CRC. Multivariate analysis showed that age, CEA level, T stage, N stage, histologic grade, and LVI were independent prognostic factors for 3-year OS of stage I–III CRC patients, excluding the variable of perineural invasion. The risk of death in patients with positive LVI was 2.012 times higher than that in patients in those without LVI (95%CI=1.101–3.676; P=0.023; Table 3).

TABLE 3 - Univariate and Multivariate Cox Regression Analyses for Overall Survival (OS) of Pre-matched Patients
Overall survival
Univariate Multivariate
Factors HR (95% CI) P HR (95% CI) P
 Male versus female 0.867 (0.508–1.480) 0.600
 Age 1.050 (1.021–1.080) 0.001 1.050 (1.020–1.080) 0.001
 CEA level 1.014 (1.008–1.020) <0.001 1.011 (1.004–1.018) 0.002
Tumor location
 Left colon versus rectum 0.934 (0.482–1.809) 0.840
 Right colon versus rectum 0.653 (0.278–1.532) 0.327
T stage
 T1–T2 versus T3 2.681 (1.370–5.247) 0.004 1.299 (0.625–2.699) 0.484
 T1–T2 versus T4 7.988 (3.518–18.137) <0.001 3.618 (1.472–8.892) 0.005
N stage
 N0 versus N1–N2 4.333 (2.517–7.458) <0.001 2.124 (1.091–4.137) 0.027
Histologic grade
 WD/MD versus PD/MUC/SRC 2.344 (1.373–4.004) 0.002 1.924 (1.109–3.339) 0.020
Lymphovascular invasion
 Yes versus no 4.371 (2.634–7.254) <0.001 2.012 (1.101–3.676) 0.023
Perineural invasion
 Yes versus no 2.068 (1.249–3.426) 0.005
Neoadjuvant chemoradiotherapy
 Yes versus no 1.470 (0.460-4.692) 0.515
 Yes versus no 1.460 (0.861-2.477) 0.160
CEA indicates carcinoembryonic antigen; CI, confidence interval; HR, hazard ratio; MD, moderately differentiated; MUC, mucinous carcinoma; PD, poorly differentiated; SRC, signet ring cell carcinoma; WD, well differentiated.

Nomogram Based on Cox Proportional Hazards Model

The Cox proportional hazards model for pre-matched patients showed that age, CEA level, T stage, N stage, histologic grade, and LVI were independent prognostic factors for 3-year OS in patients with stages I–III CRC. Proportional hazards assumption and Schoenfeld residual diagram were used to judge the reliability of the model (Fig. 3). On the basis of the above 6 prognostic factors, a nomogram was constructed, and C-index was calculated. The C-index was 0.787 (95% CI=0.728–0.845). In the 3-year receiver operating characteristic curve, the area under the curve was 0.796. The calibration curve drawn subsequently showed that the nomogram had a good calibration result, and no significant difference was found between the predicted probability and the actual probability (Figs. 4–6).

Schoenfeld residual diagram of prognostic factors in patients with stage I–III colorectal cancer. CEA indicates carcinoembryonic antigen; LVI, lymphovascular invasion.
Nomogram for predicting 3-year overall survival of patients with stages I–III colorectal cancer. When using the nomogram, a line is drawn from the prognostic factors to the point axis for the number of points, the points from all the factors are added, and a line from the total points axis to the bottom line of the nomogram is drawn to determine the probability of conversion. CEA indicates carcinoembryonic antigen; OS, overall survival.
3-year receiver operating characteristic curve for patients with stages I–III colorectal cancer based the Cox proportional hazards model. AUC indicates area under the curve.
Calibration plot of the nomogram for predicting 3-year overall survival. In the calibration plot, the x and y axes represent the nomogram-predicted and actual probabilities, respectively. OS indicates overall survival.


CRC is the most common gastrointestinal malignancy. According to the statistics of the American Cancer Society in 2020, the annual age-standardized incidence of CRC in the United States was 38.7 cases per 100,000 people, and the mortality rate was 13.9 cases per 100,000 people. Meanwhile, the 5-year relative survival rates of patients with CRC without metastasis, with local metastasis, and with distant metastasis were 90%, 71%, and 14%, respectively.22 Therefore, how to improve the prognosis of patients with CRC has become an important research object of concern to researchers.

Postoperative adjuvant chemotherapy has been proven to improve outcomes for patients with stage III CRC, but defining which patients with resected stage II disease would benefit from adjuvant chemotherapy is controversial.23,24 The 2020 version of the National Comprehensive Cancer Network guidelines relapse in patients with high-risk factors for CRC: poorly differentiated/undifferentiated, LVI, intestinal obstruction, examination of lymph node <12, peripheral nerve invasion, local perforation or margin close, ambiguous, or positive, such as stage II CRC more than there is a risk factor for recurrence is recommended in patients with postoperative adjuvant chemotherapy.11 However, this does not indicate a direct adverse outcome of LVI in patients with CRC.

The incidence of LVI in the cohort was 24.6%, consistent with the previous result that ranged from 8% to 89.5%.8–10 This fluctuation in incidence could be due to technical differences in section detection between different centers.25,26 In addition, poor knowledge of the different types of LVI may also contribute to this phenomenon. LVI can be divided into vascular invasion and lymphatic invasion according to the different tubes invaded by tumor cells, but it is very difficult to tell them apart under the microscope.27 Currently, there is increasing interest in LVI as a prognostic factor, especially the presence of extramural vascular invasion (EMVI).28,29 For example, in 2015, McClelland et al prospectively screened pathologic report data from 2405 patients undergoing radical resection for colorectal cancer, of whom 670 (27.9%) reported EMVI. The survival analysis showed that EMVI decreased survival over time.28 For patients with rectal cancer, EMVI reported before magnetic resonance imaging also has predictive significance for tumor recurrence and metastasis.30 However, the prognostic significance of intramural vascular invasion and lymphatic invasion is controversial.31–33

In our study, 10 variables, including age, sex, CEA level, perineural invasion, tumor location, T stage, N stage, histologic grade, neoadjuvant chemoradiotherapy, and chemoradiation were initially included. Among these, CEA, T stage, N stage, histologic grade, perineural invasion, and chemoradiation were considered to be closely related to LVI (P<0.05), but the correlation between these variables and LVI was eliminated in the process of PSM. In the multivariate analysis based on the Cox proportional hazards model of matched patients, LVI was considered to be an independent risk factor for 3-year OS and 3-year DFS, and some other variables were also considered independent prognostic factors. However, considering that PSM would lead to the lack of randomization of other variables except for LVI, we still draw the nomogram based on the patients before matching.20

Six variables, including age, CEA, LVI, histologic grade, T stage, and N stage, were included in the drawing of the nomogram, all of which were previously reported to be correlated with the prognosis of CRC.34,35 As can be seen from the nomogram, the axis length of the N stage is similar to that of LVI. That means, if a patient with postoperative tumor staging TXN0M0 has LVI, then his survival prognosis is similar to that of a patient with postoperative tumor staging TXN1-2M0 without LVI. This suggests that simple TNM staging lacks a certain accuracy in predicting the prognosis of intermediate colorectal tumors, and some studies have also proposed the same argument.36–38 For example, A 2020 study evaluated the effect of preoperative tumor markers (TM) combined with LVI (TM-LVI) levels on relapse-free survival at different TNM stages. The c-index and Akaike’s information criterion were used to select the best-matched subtypes for predicting relapse-free survival. The results showed that the TM-LVI subtype, which consisted of 3 categories by TM elevation status and severity of LVI status, was an independent prognostic factor for relapse-free survival in different TNM stages.31

In recent years, many researchers have studied how tumor cells spread outward through the lymphatic vascular network. Some studies have indicated that tumors interact with the lymphatic system in a variety of ways, including vessel co-option, chemotactic migration and invasion into lymphatic vessels, and induction of lymphangiogenesis. Tumor-induced lymphangiogenesis both locally and in regional lymph nodes has been correlatively and functionally associated with metastasis formation and poor prognosis. Endothelial growth factor plays an important role in this process. These growth factors can be produced by tumor cells themselves, stromal cells, tumor-associated macrophages, or damaged platelets, among which vascular endothelial growth factor (VEGF) family members and their homologous receptors VEGFR-1, VEGFR-2, VEGFR-3 have been studied more widely. Some studies have shown that inhibiting the formation of these pathways can inhibit the formation of blood vessels and lymphatics in tumor models, thus inhibiting the spread of tumors, which provides ideas for more anti-tumor-targeted drug research in the future.39,40

To the best of our knowledge, this is one of the first studies that attempt to use PSM to investigate the effect of LVI on the prognosis in patients with stages I–III CRC. Although the conclusion is encouraging, this study still needs to be explored under the following limitations. (1) The statistical method of PSM was used, which may lack randomization to a certain extent because of the artificial exclusion of some data. (2) The short follow-up time may not reflect the complete prognostic outcome of patients. (3) This study was a single-center retrospective study, and the conclusion could have been more convincing if multiple centers participated and a prospective large sample cohort had been established.

In conclusion, this study aimed to investigate the effect of LVI on the prognosis of patients with stages I–III CRC and establish a relevant prediction model as a supplement to the simple TNM staging. The calculated C-index was 0.787 (95% CI=0.728–0.845), and the area under the ROC curve was 0.796, suggesting that the model had a good ability to predict the 3-year OS of patients with stage I–III CRC. The calibration curve then suggested that the predicted probability of the model was in good agreement with the actual probability. It may provide individualized recommendations and predictions for patients with stage I–III CRC, thereby improving treatment decisions and treatment plans.


LVI is an independent risk factor for poor prognosis of patients with stages I–III CRC. The corresponding nomogram was used to establish a prediction model, which helped prove the significance of LVI as a supplement to TNM staging system and could provide an individualized prognostic prediction for clinical practice.


1. Bray F, Ferlay J, Soerjomataram I, et al. Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2018;68:394–424.
2. Lee WS, Yun SH, Chun HK, et al. Pulmonary resection for metastases from colorectal cancer: prognostic factors and survival. Int J Colorectal Dis. 2007;22:699–704.
3. Van Cutsem E, Nordlinger B, Adam R, et al. Towards a pan-European consensus on the treatment of patients with colorectal liver metastases. Eur J Cancer. 2006;42:2212–2221.
4. Lugli A, Zlobec I. The battle for prognosis at the invasive front of colorectal cancer. EBioMedicine. 2020;58:102918.
5. Zhang C, Mei Z, Pei J, et al. A modified tumor-node-metastasis classification for primary operable colorectal cancer. JNCI Cancer Spectr. 2021;5:pkaa093.
6. Shin JK, Park YA, Huh JW, et al. Tumor budding as a prognostic marker in rectal cancer patients on propensity score analysis. Ann Surg Oncol. 2021;28:8813–8822.
7. Minsky B, Mies C. The clinical significance of vascular invasion in colorectal cancer. Dis Colon Rectum. 1989;32:794–803.
8. Messenger DE, Driman DK, Kirsch R. Developments in the assessment of venous invasion in colorectal cancer: implications for future practice and patient outcome. Hum Pathol. 2012;43:965–973.
9. Lim SB, Yu CS, Jang SJ, et al. Prognostic significance of lymphovascular invasion in sporadic colorectal cancer. Dis Colon Rectum. 2010;53:377–384.
10. Sternberg A, Amar M, Alfici R, et al. Conclusions from a study of venous invasion in stage IV colorectal adenocarcinoma. J Clin Pathol. 2002;55:17–21.
11. National Comprehensive Cancer Network.NCCN Clinical Practice Guidelines in Oncology (NCCN Guidelines®) Colon Cancer,version 2.2020.[DB/OL].
12. Wang X, Cao Y, Ding M, et al. Oncological and prognostic impact of lymphovascular invasion in Colorectal Cancer patients. Int J Med Sci. 2021;18:1721–1729.
13. Lupinacci RM, Mello ES, Pinheiro RS, et al. Intrahepatic lymphatic invasion but not vascular invasion is a major prognostic factor after resection of colorectal cancer liver metastases. World J Surg. 2014;38:2089–2096.
14. Artac M, Turhal NS, Kocer M, et al. Do high-risk features support the use of adjuvant chemotherapy in stage II colon cancer? A Turkish Oncology Group study. Tumori. 2014;100:143–148.
15. Chok KS, Law WL. Prognostic factors affecting survival and recurrence of patients with pT1 and pT2 colorectal cancer. World J Surg. 2007;31:1485–1490.
16. Leijssen LGJ, Dinaux AM, Amri R, et al. Impact of intramural and extramural vascular invasion on stage II-III colon cancer outcomes. J Surg Oncol. 2019;119:749–757.
17. Baumhoer D, Thiesler T, Maurer CA, et al. Impact of using elastic stains for detection of venous invasion in the prognosis of patients with lymph node negative colorectal cancer. Int J Colorectal Dis. 2010;25:741–746.
18. Kataoka M, Hirano Y, Ishii T, et al. Impact of lymphovascular invasion in patients with stage ii colorectal cancer: a propensity score-matched study. In Vivo. 2021;35:525–531.
19. Benedetto U, Head SJ, Angelini GD, et al. Statistical primer: propensity score matching and its alternatives. Eur J Cardiothorac Surg. 2018;53:1112–1117.
20. Elze MC, Gregson J, Baber U, et al. Comparison of propensity score methods and covariate adjustment. J Am Coll Cardiol. 2017;69:345–357.
21. Austin PC, Xin Yu AY, Vyas MV, et al. Applying propensity score methods in clinical research in neurology. Neurology. 2021;97:856–863.
22. Siegel RL, Miller KD, Goding Sauer A, et al. Colorectal cancer statistics, 2020. CA Cancer J Clin. 2020;70:145–164.
23. Tournigand C, de Gramont A. Chemotherapy: is adjuvant chemotherapy an option for stage II colon cancer? Nat Rev Clin Oncol. 2011;8:574–576.
24. Group QC. Adjuvant chemotherapy versus observation in patients with colorectal cancer: a randomised study. Lancet. 2007;370:2020–2029.
25. Kojima M, Shimazaki H, Iwaya K, et al. Pathological diagnostic criterion of blood and lymphatic vessel invasion in colorectal cancer: a framework for developing an objective pathological diagnostic system using the Delphi method, from the Pathology Working Group of the Japanese Society for Cancer of the Colon and Rectum. J Clin Pathol. 2013;66:551–558.
26. Kingston EF, Goulding H, Bateman AC. Vascular invasion is underrecognized in colorectal cancer using conventional hematoxylin and eosin staining. Dis Colon Rectum. 2007;50:1867–1872.
27. Roxburgh CS, McMillan DC, Anderson JH, et al. Elastica staining for venous invasion results in superior prediction of cancer-specific survival in colorectal cancer. Ann Surg. 2010;252:989–997.
28. McClelland D, Murray GI. A comprehensive study of extramural venous invasion in colorectal cancer. PLoS One. 2015;10:e0144987.
29. van Eeghen EE, Flens MJ, Mulder MMR, et al. Extramural venous invasion as prognostic factor of recurrence in stage 1 and 2 colon cancer. Gastroenterol Res Pract. 2017;2017:1598670.
30. Lehtonen TM, Koskenvuo LE, Seppälä TT, et al. The prognostic value of extramural venous invasion in preoperative MRI of rectal cancer patients. Colorectal Dis. 2022;24:737–746.
31. Knijn N, van Exsel UEM, de Noo ME, et al. The value of intramural vascular invasion in colorectal cancer - a systematic review and meta-analysis. Histopathology. 2018;72:721–728.
32. Cho SS, Park JW, Kang GH, et al. Prognostic impact of extramural lymphatic, vascular, and perineural invasion in stage II colon cancer: a comparison with intramural invasion. Dis Colon Rectum. 2023;66:366–373.
33. Desolneux G, Burtin P, Lermite E, et al. Prognostic factors in node-negative colorectal cancer: a retrospective study from a prospective database. Int J Colorectal Dis. 2010;25:829–834.
34. Xu W, He Y, Wang Y, et al. Risk factors and risk prediction models for colorectal cancer metastasis and recurrence: an umbrella review of systematic reviews and meta-analyses of observational studies. BMC Med. 2020;18:172.
35. Liu Z, Xu Y, Xu G, et al. Nomogram for predicting overall survival in colorectal cancer with distant metastasis. BMC Gastroenterol. 2021;21:103.
36. Darrell CM, Montironi R, Paner GP. Potential biomarkers and risk assessment models to enhance the tumor-node-metastasis (TNM) staging classification of urologic cancers. Expert Rev Mol Diagn. 2020;20:921–932.
37. Lu J, Dai Y, Xie J-W, et al. Combination of lymphovascular invasion and the AJCC TNM staging system improves prediction of prognosis in N0 stage gastric cancer: results from a high-volume institution. BMC Cancer. 2019;19:216.
38. Yamano T, Yamauchi S, Igeta M, et al. Combination of preoperative tumour markers and lymphovascular invasion with TNM staging as a cost and labour efficient subtyping of colorectal cancer. Sci Rep. 2020;10:10238.
39. Sleeman JP, Thiele W. Tumor metastasis and the lymphatic vasculature. Int J Cancer. 2009;125:2747–2756.
40. Chan DLH, Segelov E, Wong RS, et al. Epidermal growth factor receptor (EGFR) inhibitors for metastatic colorectal cancer. Cochrane Database Syst Rev. 2017;6:CD007047.

colorectal cancer; lymphovascular invasion; nomogram; prognostic analysis; propensity score matching

Copyright © 2023 The Author(s). Published by Wolters Kluwer Health, Inc.