Stratifying prognosis in stage III colorectal cancer patients: the role of tumor deposits and lymph node ratio

Introduction

Globally, colorectal cancer (CRC) remains a highly prevalent and poorly prognostic malignant tumor (1-4). In spite of remarkable progress in surgical operations and treatment methods such as chemotherapy, radiotherapy and immunotherapy, the quick progression and wide-spread metastasis of this disease result in the 5-year overall survival (OS) rates staying at less than 20%. The Tumor, Node, Metastasis (TNM) staging system is regarded as the main standard for determining CRC treatment. Nevertheless, the TNM staging system’s precision is restricted by the variability found in stage I–III CRC. This is especially obvious in patients with stage III CRC. Among CRC patients at stage III, those who have undergone radical surgery still face a 30% recurrence risk (5). There is an urgent need to search for more accessible and inexpensive predictors for stage III CRC.

Hyperplastic tumor tissues known as tumor depositions (TDs) are often found in the peri-colorectal fat and regions of tumor lymphatic drainage. They do not have structures like lymph nodes and blood vessels (6). As a potential independent prognostic factor in CRC, TDs have drawn a great deal of attention (7). In the Union for International Cancer Control (UICC) TNM classification (8th edition), when there is no lymph node metastasis but positive TDs, it is classified as pN1c. For CRC patients, the prognostic value of TDs is still highly questioned when there is lymph node metastasis (8,9). Multiple research findings indicate that TD serves as an element contributing to the unfavorable prognosis of CRC patients (10), severely restricting the survival duration and prognosis of such patients (11).

The ratio of the number of lymph node metastases to the total number of retrieved lymph nodes is defined as positive lymph node ratio (LNR) (12), which is also a factor indicating poor prognosis for CRC patients (13). Surgeons typically concentrate on lymph node metastases after the operation; however, they frequently neglect the significance of the LNR. Indeed, LNR is a superior predictive factor of the prognosis compared to the number of lymph node metastases (14-16). Few studies combine analysis of TD and LNR on the prognosis of CRC patients (17,18), even TD is not considered as the predictive factor for the prognosis of patient (18). A retrospective cohort study was carried out in this study to explore the influence of TD and LNR on the long-term postoperative prognosis of stage III CRC patients. Moreover, this study provides clinicians with a more accurate approach to assess the prognosis of stage III CRC patients and proposes that more stringent treatment and follow-up measures should be carried out for high-risk CRC patients in order to improve their prognosis. We present this article in accordance with the STROBE reporting checklist (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1447/rc).

Methods

Design of the patient-related study

Two datasets were used in this retrospective cohort analysis. Data from the Surveillance, Epidemiology, and End Results (SEER) database between 2010–2015 were used to form the training set which consisted of 42,470 patients. There were 231 patients who had received surgery in the Department of Gastrointestinal Surgery, The First Affiliated Hospital, School of Medicine, Nanchang University, from January 2016 to December 2019, and they were included in the validation set. The criteria for inclusion were as follows: (I) patients were included if they had newly diagnosed stage III CRC and had received surgical resection; (II) no previous chemotherapy or radiation treatment before surgery; (III) main site ICD-O-3 codes in the SEER database: C18.0, C18.1, C18.2, C18.3 C18.4, C18.5, C18.6, C18.7, C18.8, C18.9, C19.9, C20.9; Histological ICD-O-3 codes: 8140-8147, 8210-8211, 82208221, 8260-8263, 8480-8481, 8490 (18); (IV) all patients included in this study underwent curative surgery and had histologically confirmed R0 resection margins. The criteria for exclusion were as follows: (I) medical records that were incomplete or inaccurate; (II) those who survived less than 1 month after surgery; (III) study participants were younger than 18 years old; (IV) patients diagnosed with previous or concurrent malignancies. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the ethical committee of the First Affiliated Hospital, School of Medicine, Nanchang University (IIT [2024] Clinical Ethics Review No. 253; data:2024.06.18). Informed consent was waived in this retrospective study.

Study indicators

The clinical characteristics were recorded by means of medical record review and phone calls. The items such as sex, the age when diagnosed, the clinical pathological stage (ranging from stage I to stage IV), the TNM stage (according to AJCC, version 8), the location of the tumor, the size of the tumor, the metastatic lymph nodes, the total examined lymph nodes (Examined N), the histology, the grade, the perineural invasion (PNI), the carcinoembryonic antigen (CEA), the adjuvant chemotherapy (using the first-line treatment regimen CAPOX), and the survival state (alive or died) were all included. The TDs and LNR were derived from the final histopathology report. The training set collected survival outcomes from the SEER database, and the same was done for both in other cases. In the validation set, follow-up initiated by telephone and the Chinese death registration system were used.

Patient monitoring

All postoperative patients were regularly followed up, specifically every 3 months in the first 2 years, every 6 months from the 3rd to the 5th year, and then annually after 5 years. Physical examination and blood laboratory tests, which include tests for serum CEA and Carbohydrate Antigen 19-9, were routinely part of the follow-up examinations. A chest, abdominal, and pelvic computed tomography scan is typically received by most patients within 3–6 months. At the physician’s judgment, other examinations like colonoscopy, pelvic magnetic resonance imaging or positron emission computed tomography are carried out. The follow-up of patients continued until May 30, 2024 or the patient’s demise.

The follow-up time was calculated from the date of surgery to the date of the last known contact or death. To provide a robust estimate that accounts for censored data, the median follow-up time for the entire cohort was calculated using the Reverse Kaplan-Meier method, where the event of interest was defined as the loss to follow-up (censoring), and patient deaths were treated as censored observations.

Pathological assessment

Sample handling and preservation

Following surgical resection, all CRC specimens were fixed in 10% neutral buffered formalin for a standardized duration of 24–48 hours. Subsequently, the specimens were systematically dissected by experienced pathologists to identify lymph nodes and suspicious tumor nodules.

Lymph node assessment

All identified lymph nodes were entirely submitted for processing, embedded in paraffin, sectioned, and stained with hematoxylin and eosin. The number of LNR was determined through microscopic examination by two pathologists, and the Examined N was recorded for each patient to calculate the LNR.

TD assessment

TDs were defined as discrete irregular nodules in the periodic fat, within the lymphatic drainage area of the primary tumor, without residual lymph node tissue. The assessment was conducted on the same hematoxylin and eosin-stained sections by the same pathologists, and the count of TDs was recorded for each patient. Any discrepant cases were reviewed jointly to reach a consensus.

Statistical analysis

The X-tile software (8,19) (version 3. was utilized.5.0; The Yale University in the USA was utilized to figure out the best cutoffs for the LNR and TD count in the positive TD subgroup which was sorted in the training set. This was done by using the maximum X-squared score from the Kaplan-Meier test.

Patients were divided into two groups based on the optimal cutoff value of LNR (0.31): a low LNR group (LNR-L) and a high LNR group (LNR-H). In addition, the TD-positive (TD-p) group was further divided into the TD-p1 group (1–3 TDs) and the TD-p2 group (4 or more TDs) using a cutoff value of 3. Subsequently, the Kaplan-Meier method and Cox regression analyses were employed to explore the correlations between TD, LNR and OS. OS was defined as the time interval from the date of primary surgery (or diagnosis) to the date of death from any cause. Patients who were still alive at the last follow-up were censored on that date.

SPSS 25.0 and GraphPad Prism 95.0 software were utilized for data analysis. For continuous variables that were approximately normally distributed, comparisons between groups were performed using the Independent Samples Student’s t-test. Data are presented as mean ± standard deviation. For categorical variables, comparisons between groups were performed using the Chi-squared (χ²) test or Fisher’s exact test (where cell counts were less than 5). Data are presented as numbers and percentages. Continuous variables with a skewed distribution, such as age, are presented as median with interquartile range (IQR). Univariate and multifactorial Cox regression analyses were employed by the study to analyze the independent risk factors of CRC patients. To ensure a clinically robust model, a priori included all pre-specified variables with established clinical relevance to CRC prognosis, rather than relying on a univariate screening process. The proportional hazards assumption for the final Cox model was verified using Schoenfeld residual plots and a statistical test based on the scaled Schoenfeld residuals. A difference of P<0.05 indicates statistical significance.

Results

Patient characteristics

In the training and validation sets, the median follow-up time was 105.0 (IQR, 104.5–105.5) and 73.0 (IQR, 70.3–75.7) months, respectively. The median ages were 64 (IQR, 54–74) and 66 (IQR, 58–74) years, respectively. The 5-year OS rates for CRC patients within the two groups were 58.4% and 71.4%, respectively. As shown in Figure 1, In the training set, 8,797 patients (22.34%) had TDs, of whom 3,387 (8.60%) had pN1a/b, 1,704 (4.33%) had pN1c, and 3,706 (9.41%) had pN2 stage tumors; and 8,263 patients (19.97%) had LNR-H, of whom 958 patients (2.31%) had pN1 and 7,305 patients (17.66%) had pN2 (Table 1). In the validation group, 91 patients (39.39%) had TD, of whom 42 (18.18%) had pN1a/b, 27 (11.69%) had pN1c, and 22 (9.52%) had pN2; furthermore, 46 patients (20.08%) had LNR-H, of which 8 patients (3.49%) had pN1 and 38 patients (16.59%) had pN2 (Table 2).

Figure 1 Inclusion flow chart of stage III CRC patients. LNR-Low, the ratio of LNR is less than 0.31; LNR-High, the ratio of LNR is not less than 0.31; TD-p1, TD-positive (the number of tumor deposits is 1–3); TD-p2, TD-positive (the number of tumor deposits is not less than 4). CRC, colorectal cancer; LNR, lymph node ratio; TD, tumor deposit.

In the training set, 20,539 patients (67.2%) in the TD-negative (TD-n) group received adjuvant therapy, and 5,917 patients (67.3%) in the TD-p group received adjuvant therapy; 22,201 patients (67.0%) in the LNR-L group and 5,444 patients (65.9%) in the LNR-H group received adjuvant therapy. In the validation set, 100 patients (71.4%) in the TD-n group received adjuvant therapy, and 70 patients (76.9%) in the TD-p group received adjuvant therapy; 135 patients (73.8%) in the LNR-L group and 34 patients (73.9%) in the LNR-H group received adjuvant therapy (Tables 1,2).

In the training set, we found significant differences in sex, tumor size, tumor site, T-stage, N-stage, Examined N, PNI, CEA, and LNR (all P<0.05, Chi-squared test) between TD-p patients and TD-n patients, whereas in the validation set, we only found significant differences in Examined N and CEA (P<0.05, Chi-squared test). In the training set, patients with LNR-H had statistically significant sex differences, tumor size, tumor site, T-stage, N-stage, Examined N, CEA, PNI, Chemotherapy, and TDs (all P<0.05, Chi-squared test) compared with LNR-L; while in the validation set, patients with LNR-H only had statistically significant differences in N-stage, Examined N, and PNI (all P<0.05, Chi-squared test) compared to LNR-L.

Individual analysis of TD and LNR on the prognosis of CRC patients

Univariate and multivariate Cox proportional hazards regression models were further employed to assess the impact of various clinicopathological factors on OS. For stage III CRC patients in the training set, factors such as the location of the tumor, its size, certain pathological features, sex, the stage of the tumor in terms of T and N, the level of CEA, whether adjuvant chemotherapy was carried out, LNR, and TDs were all independent factors related to prognosis (Table 3 and Figure 2). For stage III CRC patients in the validation set, TDs as well as PNI served as independent predictive elements. Moreover, the sample size might be relatively insufficient, so that LNR could influence the prognosis of CRC patients without being an independent prognostic factor (Table 4).

Figure 2 Forest plot of multivariate Cox regression analysis in the training group. The P values were all calculated using the log-rank test. LNR-Low, the ratio of LNR is less than 0.31; LNR-High, the ratio of LNR is not less than 0.31. CEA, carcinoembryonic antigen; CI, confidence interval; Examined N, total examined lymph nodes; LNR, lymph node ratio; N, node; OR, odds ratio; PNI, perineural invasion; T, tumor; TD, tumor deposits.

Prior to constructing the Cox regression models, the proportional hazards assumption was tested using Schoenfeld residuals. The results indicated that the global test for the model was not statistically significant (P>0.05, log-rank test), and no individual variable showed a significant correlation between its Schoenfeld residuals and time (P>0.05 for all, log-rank test). This confirms that the proportional hazards assumption was satisfied for the Cox models in this study, and the results are reliable.

Figure 2 shows that the 5-year OS of TD-n, TD-p, LNR-L, LNR-H were 62.2%, 47.0%, 63.3%, and 40.4% in the training set, and 76.4%, 63.7%, 74.9%, and 56.5% in the validation set, respectively (In the training sets, TD-p Group: 4,662 deaths among 8,797 patients; TD-n Group: 11,559 deaths among 30,581 patients; LNR-L Group: 12,152 deaths among 33,112 patients; LNR-H Group: 4,924 deaths among 8,263 patients. In the validation sets, TD-p Group: 33 deaths among 91 patients; TD-n Group: 33 deaths among 140 patients; LNR-L Group: 46 deaths among 183 patients; LNR-H Group: 20 deaths among 46 patients). In two sets of postoperative stage III CRC patients, the 5-year OS rate was remarkably poorer in the TD-p or LNR-H groups than in the TD-n or LNR-L groups (Figure 3A-3D, all P<0.05, log-rank test).

Figure 3 Kaplan-Meier plots of OS among different study indicator groups in the training group and the validation group. (A) The Kaplan-Meier plot of OS for the TD-p group and the TD-n group in the training group. (B) The Kaplan-Meier plot of OS for the TD-p group and the TD-n group in the validation group. (C) The Kaplan-Meier plot of OS for the LNR-L group and LNR-H group in the training group. (D) The Kaplan-Meier plot of OS for the LNR-L group and LNR-H group in the validation group. (E) The Kaplan-Meier plot of OS for the different number ranges of TDs groups in the training group. (F) The Kaplan-Meier plot of OS for the different number ranges of TDs groups in the validation group. The P values were all calculated using the log-rank test. LNR-L, the ratio of LNR is less than 0.31; LNR-H, the ratio of LNR is not less than 0.31; TD-p1, TD-positive (the number of tumor deposits is 1–3); TD-p2, TD-positive (the number of tumor deposits is not less than 4). LNR, lymph node ratio; OS, overall survival; TD, tumor deposit.

Compared with TD-n patients, TD-p patients exhibited a 1.5-fold incremental risk of death in the training set (HR, 1.518; 95% CI: 1.471–1.565, P<0.001, log-rank test) and a 1.9-fold incremental risk in the validation set (HR, 1.920; 95% CI: 1.208–3.052, P=0.006, log-rank test). Compared with LNR-L patients, LNR-H patients had a 1.9-fold risk of death in the training set (HR, 1.902; 95% CI: 1.845–1.960, P<0.001, log-rank test) and a 2.0-fold risk of death in the validation set (HR, 1.989; 95% CI: 1.196–3.310, P=0.008, log-rank test).

TD-positive patients were classified into two subgroups: TD-p1 [1–3] and TD-p2 [>3], respectively. TD-p1 and TD-p2 had a 5-year OS of 48.3%and 37.6% in the training set and 65.4%and 53.8% in the validation set, respectively (In the training sets, TD-p1 Group: 4,007 deaths among 7,751 patients; TD-p2 Group: 652 deaths among 1,046 patients. In the validation sets, TD-p1 Group: 27 deaths among 78 patients; TD-p2 Group: 6 deaths among 13 patients). Our results show that the 5-year OS rate was significantly lower in the TD-p2 group than in the TD-p1 group in the training and validation sets (Figure 3E,3F; P<0.05, log-rank test). In comparison with those having TD-p1, patients in the TD-p2 group had a 1.3-fold risk of death (HR, 1.290; 95% CI: 1.193–1.395; P<0.001, log-rank test) in the training and 1.7-fold risk of death (HR, 1.651; 95% CI: 0.726–3.758, P=0.23, log-rank test) in the validation set. These results show that patients with TD-p2 (>3) had worse cancer survival and high death risk.

Combined analysis of TD and LNR on the prognosis of CRC patients

TD and LNR are each an independent prognostic factor for patients with stage III cancer. Four subgroups were created for the combined TD and LNR variables, namely TD-n/LNR-L, TD-p/LNR-L, TD-n/LNR-H and TD-p/LNR-H. The combination of variables had an effect on the stratification of prognosis, with 5-year OS of 65.1%, 52.9%, 43.9%, and 32.1% for the training set and 78.44%, 69.57%, 66.67%, and 45.45% for the validation set, respectively (Figure 4A,4B) (in the training sets, TD-n/LNR-L Group: 9,687 deaths among 27,758 patients; TD-p/LNR-L Group: 2,968 deaths among 6,303 patients; TD-n/LNR-H Group: 3,317 deaths among 5,914 patients; TD-p/LNR-H Group: 1,693 deaths among 2,494 patients. In the validation sets, TD-n/LNR-L Group: 25 deaths among 116 patients; TD-p/LNR-L Group: 21 deaths among 69 patients; TD-n/LNR-H Group: 8 deaths among 24 patients; TD-p/LNR-H Group: 12 deaths among 22 patients). Among the four risk groups, the survival rate of cancer patients with TD-p/LNR-H was the lowest. In comparison to TD-n/LNR-L patients, TD-p/LNR-H patients had a 2.5-fold risk of death in the training set (HR, 2.538; 95% CI: 2.418–2.663; P<0.001, log-rank test) and a 3.9-fold risk of death in the validation set (HR, 3.897; 95% CI: 1.997–7.607; P<0.001, log-rank test). These results indicated that the combination of TDs and LNR is able to effectively classify the prognosis and mortality risk among CRC patients.

Figure 4 Kaplan-Meier plots of OS among the combined study indicator groups in the training group and the validation group. (A) The Kaplan-Meier plots of OS by the combined variable of TD and LNR in the training set. (B) The Kaplan-Meier plots of OS by the combined variable of TD and LNR in the validation group. (C) The Kaplan-Meier plot of OS by the combined variables for LNR and different number ranges of TDs in the training group. (D) The Kaplan-Meier plot of OS by the combined variables for LNR and different number ranges of TDs in the validation set. The P values were all calculated using the log-rank test. LNR-L, the ratio of LNR is less than 0.31; LNR-H, the ratio of LNR is not less than 0.31; TD-p1, TD-positive (the number of tumor deposits is 1–3); TD-p2, TD-positive (the number of tumor deposits is not less than 4). LNR, lymph node ratio; OS, overall survival; TD, tumor deposit.

The combined variable of LNR and different number ranges of TD was classified into six subgroups: TD-n/LNR-L, TD-p1/LNR-L, TD-p2/LNR-L, TD-n/LNR-H, TD-p1/LNR-H, and TD-p2/LNR-H (Figure 4C,4D). In the training set, patients with TD-p1/LNR-L, TD-p2/LNR-L, TD-p1/LNR-H, and TD-p2/LNR-H had 5-year OS rates of 53.95%, 43.69%, 32.94%, and 27.53%, respectively (P<0.001, log-rank test) (In the training sets, TD-p1/LNR-L Group: 2,603 deaths among 5,653 patients; TD-p2/LNR-L Group: 366 deaths among 650 patients; TD-p1/LNR-H Group: 1,406 deaths among 2,098 patients; TD-p2/LNR-H Group: 286 deaths among 396 patients). We also observe similar results in the validation set. The risk of death in patients with TD-p1/LNR-H and TD-p2/LNR-H CRC in the training set was 2.5 and 2.8 times greater than that in patients with TD-n/LNR-L (HR, 2.496; 95% CI: 2.364–2.625, P<0.001, log-rank test; HR, 2.801; 95% CI: 2.504–3.133, P<0.001, log-rank test). In the validation set a similar result was observed (Figure 4C,4D). However, we observed no difference in risk of death between TD-p1/LNR-H and TD-p2/LNR-H groups in the training set (P=0.12, log-rank test) and in the validation set (P=0.72, log-rank test). These results suggest that the combined variable of LNR and TDs is not dependent on the amounts of TDs.

Discussion

Stage III CRC patients from training set and validation set had 5-year OS rates of 58.4% and 71.4%, respectively. In both data sets, TD positivity and high LNR (values ≥0.31) were related to the decline in OS. Data analyses demonstrated that TD and LNR were independent factors predicting a poor prognosis in stage III CRC. Combining TD and LNR, regardless of the TD amount, can more precisely predict the death risk of CRC patients.

The 5-year OS rate for stage III CRC patients in our center exceeded 70%, much higher than that in the SEER database, suggesting that the treatment outcomes of CRC in China are comparable to those in Japan and Korea. In the SEER database, TD-p and TD-n patients differ in characteristics such as sex, tumor size, tumor location, T-stage, N-stage, number of examinations, CEA and PNI. Nevertheless, remarkable differences in Examined N and CEA were noticed at our center. Naturally, the same can be seen in LNR as well. It is considered that the main reason is the insufficiency of sample size.

Currently, according to the AJCC 8th edition TNM staging system, if there is TD without lymph node metastasis, no matter how many peripheral TD there are, it is defined as stage pN1c (6). CRC patients at stage pN1c had a poorer prognosis in comparison with those at stage pN1a or pN1b (17), yet they had a greater OS duration when compared to patients at stage pN2 (20). So, it is believed that TD can have an adverse effect on the prognosis of CRC patients and might even be a prognostic factor for CRC patients.

After utilizing the “all variables included” model, we observed a phenomenon that contradicts established clinical knowledge. In the validation cohort, the HR for TD, LNR, and PNI were all less than 1, which suggested they were “protective factors”. This finding clearly contradicts biological and clinical logic.

We conducted an in-depth investigation into this anomalous result and concluded that multicollinearity and small sample size were the primary causes (21). Specifically, within our variable set, N staging, Examined N, and LNR serve as indicators of lymph node burden from different perspectives. These variables exhibit a strong intrinsic correlation. When such highly correlated variables are included simultaneously in a regression model, the model struggles to accurately estimate the independent contributions of each variable. This can lead to counterintuitive shifts in the HR, even causing a reversal in direction (from HR >1 to HR <1). This phenomenon does not reflect a true biological effect but is instead a statistical distortion.

In CRC, TD serves as a crucial biomarker, having a close association with adjacent structures like blood vessels, lymphatic vessels and peripheral nerves. However, TDs cannot be considered as metastatic lymph nodes in gastric cancer (22). TD is an independent factor for prognosis in CRC patients (10). It has been demonstrated in our research that TD is also verified as an independent factor for prognosis in stage III CRC patients within both sets. In the latest TNM staging system version, when lymph node metastases exist, TDs are not included in the final N stage. This rule has led to a great deal of disputes. In the current AJCC TNM staging system, TD was taken into account in 4.33% (1,704/39,378) of stage III patients with CRC (pN1c patients). However, TD could be used to offer valuable prognostic information in 22.34% (8,797/39,378) of stage III patients with CRC. In the past few years, multiple studies have investigated the viability of regarding TDs as positive lymph nodes (pLNs) for the purpose of amending N staging. Counting TDs as pLNs for inclusion in the pLN count has been shown to be viable (23). A clinical trial (17) showed that the prognosis of stage III patients who were first classified as N1 but then restaged as N2 was poorer than those initially classified as N1, yet similar to those initially classified as N2. This approach considers the existence of TDs and the significance of variable amounts, yet it overlooks the precise number of TDs and lymph nodes, particularly when both pathological characteristics are present (18). In our study, we found that patients with positive TDs [>3] have worse prognosis than patients with positive TDs [1-3]. For III stage CRC patients, the quantity of TDs that is comparable to pLN can be recognized as a predictor, and might even be a significant factor influencing CRC staging.

In the past, the number of lymph nodes served as an N-staging index for CRC patients. Currently, in patients with progressive CRC, LNR has a better prognostic value compared to the number of lymph node metastases (24). This study also revealed that patients with a high LNR are at a greater risk of death compared to those with a low LNR. There is an indication that the combination of TDs and LNR can enhance the precision of prognosis prediction in stage III colon cancer (18). This study’s findings indicated that the combined prognostic signature effectively stratified stage III CRC patients into distinct risk groups with significantly different survival outcomes in both the training and validation cohorts. This suggests that combining TD with LNR effectively utilizes their respective values and can stratify the prognosis.

The prognostic significance of different quantities of TDs in patients was further investigated. This study showed that the 5-year OS of CRC patients in the TD-p2 [>3] group was lower than that of patients in the TD-p1 [1–3] group. Furthermore, CRC patients in the TD-p2 group had a greater death risk compared to those in the TD-p1 group. So, when evaluated as an individual indicator, the quantity of TDs is also a crucial predictor for stage III CRC patients. It is interesting that when LNR is combined with TD-p1 group or TD-p2 group, the death risk remains the same. This implies that such combined analysis does not depend on the quantity of TDs. So, the combination of LNR and TDs, irrespective of how many TDs there are, might be utilized to recognize the high risk of CRC deaths.

Several limitations of our study should be considered. Firstly, as a retrospective study, the sample size was not predetermined by a formal power calculation, which might affect the generalizability of our findings and the stability of the multivariate model. Future prospective studies with pre-specified sample sizes are warranted to validate our results. Secondly, the training and validation cohorts were derived from patients treated in different time periods (2010–2015 vs. 2016–2019). While this temporal difference could introduce bias due to evolving clinical practices, it also strengthens the generalizability of our model by demonstrating its robust performance across a temporal validation cohort. Thirdly, the quantity of patients in the research on postoperative prognosis is insufficient, and certain conclusions reached might be affected by selection bias. Hence, it is essential to recruit more CRC patients in future research.

Conclusions

Our study demonstrated that both TD and LNR are independent prognostic factors for stage III CRC. The combined variable of LNR and TDs, regardless of the count of TDs, accurately predicted the risk of death in CRC patients.

Acknowledgments

We thank all the patients whose data were used for this study. Our greatest acknowledgement goes to the authors who made detailed data available for this study and to all our colleagues in this study for their hard work.

Footnote

Reporting Checklist: The authors have completed the STROBE reporting checklist. Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1447/rc

Data Sharing Statement: Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1447/dss

Peer Review File: Available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1447/prf

Funding: This study was supported by grants from the National Natural Science Foundation of China (No. 82100665), Beijing Science Technology Innovation Medical Development Foundation (No. KC2023-JX-0186-FQ029), Jiangxi Provincial Health Commission Technology Plan Project (No. 202510222), “the Talent 555 project of Jiangxi Province”, China, and Jiangxi Provincial Natural Science Foundation (No. 20242BAB20352).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tcr.amegroups.com/article/view/10.21037/tcr-2025-1447/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the ethical committee of the First Affiliated Hospital, School of Medicine, Nanchang University (IIT [2024] Clinical Ethics Review No. 253; data:2024.06.18). Informed consent was waived in this retrospective study.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Morgan E, Arnold M, Gini A, et al. Global burden of colorectal cancer in 2020 and 2040: incidence and mortality estimates from GLOBOCAN. Gut 2023;72:338-44. [Crossref] [PubMed]
2. Sung H, Ferlay J, Siegel RL, et al. Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries. CA Cancer J Clin 2021;71:209-49. [Crossref] [PubMed]
3. Dekker E, Tanis PJ, Vleugels JLA, et al. Colorectal cancer. Lancet 2019;394:1467-80. [Crossref] [PubMed]
4. Bray F, Laversanne M, Sung H, et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin 2024;74:229-63. [Crossref] [PubMed]
5. van der Stok EP, Spaander MCW, Grünhagen DJ, et al. Surveillance after curative treatment for colorectal cancer. Nat Rev Clin Oncol 2017;14:297-315. [Crossref] [PubMed]
6. Ueno H, Nagtegaal ID, Quirke P, et al. Tumor deposits in colorectal cancer: Refining their definition in the TNM system. Ann Gastroenterol Surg 2023;7:225-35. [Crossref] [PubMed]
7. Delattre JF, Selcen Oguz Erdogan A, Cohen R, et al. A comprehensive overview of tumour deposits in colorectal cancer: Towards a next TNM classification. Cancer Treat Rev 2022;103:102325. [Crossref] [PubMed]
8. Zheng X, Cen W, Zhu J, et al. Prognostic Value of Tumor Deposits in Stage III Colorectal Cancer Patients with Different N Stages: A Population-Based, Retrospective, Cohort Study. Ann Surg Oncol 2023;30:8067-73. [Crossref] [PubMed]
9. Pei JP, Zhang CD, Liang Y, et al. A Modified Pathological N Stage Including Status of Tumor Deposits in Colorectal Cancer With Nodal Metastasis. Front Oncol 2020;10:548692. [Crossref] [PubMed]
10. Basnet S, Lou QF, Liu N, et al. Tumor deposit is an independent prognostic indicator in patients who underwent radical resection for colorectal cancer. J Cancer 2018;9:3979-85. [Crossref] [PubMed]
11. Nagayoshi K, Ueki T, Nishioka Y, et al. Tumor deposit is a poor prognostic indicator for patients who have stage II and III colorectal cancer with fewer than 4 lymph node metastases but not for those with 4 or more. Dis Colon Rectum 2014;57:467-74. [Crossref] [PubMed]
12. Chen SL, Steele SR, Eberhardt J, et al. Lymph node ratio as a quality and prognostic indicator in stage III colon cancer. Ann Surg 2011;253:82-7. [Crossref] [PubMed]
13. Naidu K, Chapuis PH, Connell L, et al. Lymph node ratio prognosticates overall survival in patients with stage IV colorectal cancer. Tech Coloproctol 2024;28:115. [Crossref] [PubMed]
14. Macedo F, Sequeira H, Ladeira K, et al. Metastatic lymph node ratio as a better prognostic tool than the TNM system in colorectal cancer. Future Oncol 2021;17:1519-32. [Crossref] [PubMed]
15. Hong KD, Lee SI, Moon HY. Lymph node ratio as determined by the 7th edition of the American Joint Committee on Cancer staging system predicts survival in stage III colon cancer. J Surg Oncol 2011;103:406-10.
16. Qin ZF, Gao XY, Zhu YZ, et al. Evaluation of LNR and modified N stage systems for prognostic stratification of metastatic lymph nodes in stage III colorectal Cancer. Sci Rep 2025;15:13735. [Crossref] [PubMed]
17. Cohen R, Shi Q, Meyers J, et al. Combining tumor deposits with the number of lymph node metastases to improve the prognostic accuracy in stage III colon cancer: a post hoc analysis of the CALGB/SWOG 80702 phase III study (Alliance) Ann Oncol 2021;32:1267-75. [Crossref] [PubMed]
18. Liu L, Ji J, Ge X, et al. Prognostic value of tumor deposits and positive lymph node ratio in stage III colorectal cancer: a retrospective cohort study. Int J Surg 2024;110:3470-9. [Crossref] [PubMed]
19. Zheng K, Zheng N, Xin C, et al. The Prognostic Significance of Tumor Deposit Count for Colorectal Cancer Patients after Radical Surgery. Gastroenterol Res Pract 2020;2020:2052561. [PubMed]
20. Jin M, Roth R, Rock JB, et al. The impact of tumor deposits on colonic adenocarcinoma AJCC TNM staging and outcome. Am J Surg Pathol 2015;39:109-15. [Crossref] [PubMed]
21. Bayman EO, Dexter F. Multicollinearity in Logistic Regression Models. Anesth Analg 2021;133:362-5. [Crossref] [PubMed]
22. Xu R, Li Y, Zhao H, et al. Tumor deposits in gastric cancer cannot be regarded as metastatic lymph nodes: A single-center retrospective study. Eur J Surg Oncol 2025;51:109719. [Crossref] [PubMed]
23. Ueno H, Mochizuki H, Shirouzu K, et al. Actual status of distribution and prognostic impact of extramural discontinuous cancer spread in colorectal cancer. J Clin Oncol 2011;29:2550-6. [Crossref] [PubMed]
24. Heng Y, Huang M, Xu J, et al. Prognostic value of tumor deposits and positive lymph nodes in colorectal cancer surgery: improved staging for long-term prognosis. BMC Gastroenterol 2025;25:154. [Crossref] [PubMed]