This retrospective, single-center study was conducted at the General Surgery Department of Hitit University Erol Olçok Training and Research Hospital. Data from all patients aged 18 years or older who underwent surgery for colon cancer between January 1, 2016, and January 1, 2022, were reviewed using the hospital’s information system. Given the retrospective nature of the study and anonymized data collection, the requirement for informed consent was waived by the Clinical Research Ethics Committee (Protocol No: 2023-60, 24.05.2023), in accordance with institutional and international ethical standards. Exclusion criteria included patients with distant organ metastases at the time of diagnosis, those with known oncologic diseases other than newly diagnosed colon cancer, those who underwent palliative surgery, patients with incomplete data, and those under 18 years of age. Additionally, cases with rectal tumors were excluded due to differences in treatment protocols. However, tumors located in the rectosigmoid junction or distal sigmoid colon, which were managed using colon cancer protocols, were included in the analysis. These tumors may have required anterior or low anterior resection depending on their proximity to the rectum, which explains the proportion of such surgical procedures in the cohort. Although patients with stage 0 colon cancer do not typically have lymph node involvement, they were retained in the overall cohort for completeness. However, they were excluded from subgroup analyses where MLNR or nodal metastasis was the primary variable of interest. This approach preserved statistical power while minimizing bias. Ethical approval for this study was obtained from the Clinical Research Ethics Committee of Hitit University Faculty of Medicine (Protocol No: 2023-60) and the study protocol was conducted in accordance with the 1964 Helsinki Declaration.

A total of 122 patients who met the inclusion criteria were enrolled in the study. All patients included in the study were of Turkish ethnicity, as the data were collected from a single tertiary healthcare institution in Türkiye. This demographic context should be considered when interpreting the generalizability of the results. Data collected included demographic characteristics, tumor location, total and metastatic lymph node counts retrieved during surgery, disease stage, recurrence and mortality status during follow-up, length of hospital stay, chemotherapy treatments, postoperative complications, and follow-up durations.

MLNR was calculated as the number of metastatic lymph nodes divided by the total number of lymph nodes retrieved during surgery.

Disease-free survival (DFS) was defined as the time from surgery to the date of documented recurrence or last follow-up without recurrence. Overall survival (OS) was defined as the time from surgery to death from any cause or last follow-up.

All statistical analyses were performed using IBM SPSS Statistics for Windows, version 26 (IBM Corp., Armonk, NY, USA). Descriptive statistics were used to summarize categorical variables as frequencies and percentages, while numerical variables were expressed as mean ± standard deviation or median, depending on the distribution. The normality of data distribution was assessed using the Shapiro-Wilk test. Relationships between variables were analyzed using Pearson or Spearman correlation coefficients based on data distribution. Comparisons of numerical variables between study groups were conducted using the Mann-Whitney U test for all variables except age, which was analyzed using the Student’s t-test based on a Gaussian distribution. Categorical variables were compared using the Chi-square test.

Interactions between variables were examined through binomial logistic regression analysis to evaluate the independent predictive capacity of the MLNR for recurrence. Confounding variables were identified based on both clinical relevance and statistical significance in univariate analysis (p<0.10). These variables were entered into a multivariate logistic regression model to evaluate the independent effect of MLNR on recurrence, adjusting for potential confounders such as age, comorbidities, emergency surgery, operation type, and adjuvant therapy. The model demonstrated good fit (Nagelkerke R²=0.278, classification accuracy: 84.4%).

In calculating the MLNR, all retrieved lymph nodes from the surgical specimen were included in the denominator, regardless of their anatomical location. However, we acknowledge that this approach may introduce bias, especially in right hemicolectomy specimens, where lymph nodes from the midcolic area may be included alongside truly regional nodes such as the ileocolic group. Although such inclusion was done for consistency and feasibility in this retrospective study, we recognize the potential influence of anatomical variability on MLNR values.

Receiver Operating Characteristic (ROC) curve analysis was used to evaluate the discriminative ability of MLNR in predicting recurrence. Several cut-off values in the range of 0.10 to 0.15 were tested during ROC analysis. Among these, 0.125 was selected as the optimal threshold based on the highest Youden index value, providing the best trade-off between sensitivity (61.9%) and specificity (83.2%) for predicting recurrence. Diagnostic performance metrics, including sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), accuracy, and odds ratios were calculated for this cut-off.

Kaplan-Meier survival analysis was performed to evaluate disease-free survival, and statistical significance between groups was assessed using the Log-Rank test. A p-value of <0.05 was considered statistically significant.

The study included a total of 122 patients, of whom 44 (36.07%) were female and 78 (63.93%) were male. The median age was 68.46 ± 10.61 years. Among the patients, 73 (59.84%) underwent elective surgeries. The mean DFS duration was 36 months, ranging from 6 to 72 months. During the 60-month follow-up period, 25 patients (20.49%) died. Detailed demographic and clinical variables are summarized in Table 1 .

Patients were categorized into two groups based on survival status: alive (n=97) and deceased (n=25). While no statistically significant difference was observed in gender distribution (p=0.159), age showed a significant association with mortality. The median age of surviving patients was 66.48 ± 9.56 years, compared to 76.12 ± 11.2 years for those who had died (p<0.001; Table 1 ).

Emergency surgery and tumor localization were significantly associated with mortality in univariate analyses (p=0.023 and p=0.009, respectively). However, operation type, histopathology, disease stage, and length of hospital stay did not show significant associations with mortality (p=0.108, p=0.952, p=0.836, and p=0.822, respectively; Table 2 ). Treatment and follow-up factors were also evaluated. Local recurrence, distant metastasis, and overall recurrence were significantly associated with mortality (p<0.001, p=0.008, and p<0.001, respectively). However, adjuvant therapy, the number of malignant lymph nodes, and total lymph node count were not significant predictors (p=0.697, p=0.082, and p=0.802, respectively; Table 2 ). Complications during treatment were not significantly associated with mortality (p=0.507) ( Table 2 ). Contrary to our initial hypothesis, MLNR was not significantly different between survivors and deceased patients.

Patients were then stratified into recurrence (n=21) and non-recurrence (n=101) groups. Gender distribution did not differ significantly between the groups (p=0.199). Age was significantly associated with mortality, whereas no statistically significant association was found between age and recurrence (p=0.935). The presence of comorbidities (p=0.774), emergency surgery (p=0.444), tumor localization (p=0.075) and type of surgical intervention (p=0.170) showed no significant associations with recurrence.

Disease staging was significantly associated with recurrence (p=0.034). Although adjuvant therapy was more frequently administered in the recurrence group (p=0.015), this likely reflects clinical decisions informed by higher initial risk profiles or unfavorable prognostic factors, rather than implying a direct causal relationship with recurrence. Complications during treatment were not significantly associated with recurrence (p=0.135). Patients without recurrence had significantly longer follow-up durations compared to those with recurrence (p=0.011), likely due to higher mortality rates in the recurrence group (11.88% vs. 61.9%, p<0.001). The presence of malignant lymph nodes was significantly associated with recurrence (p<0.001). Although the total lymph node count was not a significant predictor (p=0.273), MLNR emerged as a highly significant factor in univariate analysis (p<0.001). A full comparison of clinical and pathological variables between recurrence groups is presented in Table 3 .

To determine whether MLNR independently predicted recurrence, multivariate logistic regression analysis was performed, incorporating variables identified as clinically relevant or statistically significant in univariate analysis. These included age, gender, comorbidities, surgical urgency (emergency vs. elective), type of surgery, postoperative complications, use of adjuvant chemotherapy, and MLNR (categorized by the cut-off value of 0.125). As shown in Table 4 , MLNR ≥ 0.125 emerged as an independent predictor of disease recurrence (odds ratio [OR]: 13.07; 95% confidence interval [CI]: 1.03-166.57; p=0.048). Conversely, receipt of adjuvant therapy was associated with a significant reduction in recurrence risk (OR: 0.234; 95% CI: 0.059-0.923; p=0.038). Other factors such as age, gender, and surgical urgency did not retain statistical significance in the final model.

The discriminative ability of MLNR in predicting recurrence was further evaluated using ROC curve analysis. The area under the ROC curve (AUC) was 0.752 (95% CI: 0.634-0.869; p < 0.001), indicating good prognostic performance. The optimal MLNR threshold for recurrence prediction was identified as 0.125 using the Youden Index. At this cut-off, MLNR had a sensitivity of 61.9% and a specificity of 83.2%. The positive predictive value (PPV) and negative predictive value (NPV) were 43.3% and 91.3%, respectively, yielding an overall diagnostic accuracy of 79.5%. The odds of recurrence in patients with MLNR ≥ 0.125 were significantly higher compared to those below this threshold (OR: 8.03; 95% CI: 2.89-22.34; p < 0.001). These diagnostic performance metrics are presented in Table 5 , and the ROC curve is illustrated in Figure 1 .

DFS was further analyzed using Kaplan-Meier survival curves stratified by MLNR groups. Patients with MLNR < 0.125 demonstrated significantly longer DFS compared to those with MLNR≥0.125 (log-rank p < 0.001). The mean DFS duration was 42.13 ± 17.9 months in the MLNR <0.125 group, while it was 27.00 ± 15.66 months in the MLNR≥0.125 group. The estimated mean DFS for the low MLNR group was 66.99 months (95% CI: 63.66-70.32), in contrast to 39.70 months (95% CI: 31.32-48.08) for the high MLNR group. The overall cohort had an estimated mean DFS of 61.93 months (95% CI: 57.99-65.87), as summarized in Table 6 and visualized in Figure 2 . These findings underscore the prognostic value of MLNR in stratifying patients according to recurrence risk and disease-free survival. Although MLNR was significantly associated with DFS in Kaplan-Meier analysis, we did not perform Cox proportional hazards modeling to compute hazard ratios, as our primary survival outcome analysis was categorical (recurrence vs. no recurrence) and based on logistic regression.

MLNR emerges as a more sensitive prognostic indicator compared to the metastatic lymph node count in the TNM staging system. A meta-analysis has emphasized the prognostic value of MLNR in colon cancer, noting that patients with a higher MLNR tend to have worse survival outcomes (22). The current literature supports the use of lymph node ratio for colon cancer prognosis. In a systematic review and meta-analysis, increased lymph node ratio was associated with reduced OS (HR: 2.36) and DFS (HR: 3.71) (25). Similarly, another meta-analysis found significant associations between MLNR and both OS (HR: 1.91) and DFS (HR: 2.75) (24). Despite consistent findings across studies, there is no universally accepted cut-off value for MLNR. eta-analyses by Karjol et al. (2020) and Zhang et al. (2016) reported a wide range of thresholds for MLNR, varying from 0.05 to 0.6, with some studies categorizing patients into multiple risk groups. Frequently used single cut-off values include 0.1, 0.125, 0.2, and 0.3. Interestingly, the greatest consistency across studies has been observed when a threshold below 0.2 was applied (26, 27).

In our study, an MLNR cut-off of 0.125 was identified using ROC curve analysis and the Youden Index, providing an optimal balance between sensitivity and specificity for predicting recurrence. This threshold aligns with previously reported values and may offer more precise risk stratification within our cohort. Unlike studies that applied thresholds such as 0.1 or 0.2, our ROC-based cut-off of 0.125 offers a comparable prognostic power, though direct statistical comparisons were beyond the scope of our analysis. We found that patients with an MLNR≥0.125 had significantly higher odds of recurrence, with more than a 13-fold increase compared to those with lower MLNRs, as confirmed by adjusted multivariate analysis (OR: 13.07, 95% CI: 1.03- 166.57, p=0.048). Additionally, patients with lower MLNRs demonstrated significantly longer disease-free survival (p<0.001). These findings support the potential role of MLNR in recurrence risk assessment. Recent studies have reinforced the prognostic importance of MLNR in colon cancer. Ichhpuniani et al. and Harman Kamalı et al. demonstrated that a high MLNR is associated with reduced survival outcomes, supporting its integration into staging and treatment planning. Additionally, Wang et al. emphasized MLNR’s role in guiding personalized therapeutic decisions, while recent updates in the field have highlighted the relevance of MLNR-based risk stratification when considered alongside lifestyle-related factors (28–30). However, its clinical implementation requires external validation in larger, prospective cohorts. Until then, MLNR may serve as a useful adjunct in triaging patients for closer surveillance or adjuvant treatment planning.

The TNM classification system considers the number of metastatic lymph nodes as one of the main factors determining prognosis. Guidelines recommend examining at least 12 lymph nodes in colon cancer surgery. However, this approach may overlook variations in the total number of lymph nodes examined, which can be influenced by surgeon and pathologist expertise, tumor location, and the urgency of surgery (31). Zhang and colleagues reviewed 33 studies analyzing different MLNR cut-off values and concluded that higher MLNR independently predicted survival in colon cancer patients, suggesting its inclusion in future staging systems (26). This finding is supported by numerous studies showing that MLNR is a better prognostic factor than the N stage (32–34). However, some studies have found that MLNR performs similarly to or less effectively than the N stage (35–37). Some researchers have proposed hybrid staging systems integrating MLNR with the TNM classification, which have shown superior outcomes (22, 38, 39). For example, Wang and colleagues demonstrated that patients classified as stage IIIB by TNM but with MLNR >30% had survival rates worse than those with MLNR ≤30%, more closely resembling stage IIIC patients. These findings suggest that MLNR may complement TNM staging in refining prognostic assessments, though further validation is warranted before recommending integration into standard staging algorithms.

In our study, we found that emergency surgery and tumor localization significantly influenced mortality (p=0.023 and p=0.009, respectively). However, the type of surgery, histopathology, disease stage, and duration of hospitalization were not significantly associated with survival. Although the total lymph node count did not correlate with survival (p=0.273), MLNR was found to be a significant prognostic factor (p<0.001). This suggests that MLNR should be included as a complementary metric in the staging systems alongside TNM.

Adjuvant chemotherapy is a key component of colon cancer treatment and significantly impacts prognosis. The lymph node ratio (LNR) is an important determinant of both disease progression and adjuvant treatment planning. The relationship between MLNR and survival in colon cancer was first raised by Berger et al. (40). In another study examining 24,477 patients, MLNR was found to be a more accurate prognostic factor than the N stage. When the cut-off for MLNR was set at 0.2, patients with MLNR below this value had a survival rate of 81.1%, while those above it had a rate of 46.6%. Multivariate analysis showed that both thresholds were significant in predicting survival (41). Patients who received adjuvant therapy had a significantly lower risk of disease recurrence compared to those who did not (p=0.015). This indicates that adjuvant chemotherapy plays a critical role in colon cancer treatment and is effective in reducing the risk of recurrence. MLNR is a prognostic marker that can guide adjuvant therapy decisions. In multivariate analysis, patients with MLNR ≥ 0.125 had over 13-fold increased odds of recurrence (OR: 13.07, 95% CI: 1.03-166.57, p=0.048), whereas adjuvant therapy was associated with a 77% reduction in recurrence risk (OR: 0.23, 95% CI: 0.06-0.92, p=0.038). his finding suggests that adjuvant chemotherapy planning should be tailored more precisely for patients with elevated MLNR. Adjuvant chemotherapy decisions can be individualized by considering the MLNR cut-off value (0.125). Patients with MLNR <0.125 demonstrated better survival outcomes and lower recurrence rates. This suggests that MLNR can help identify the patient groups most likely to benefit from adjuvant chemotherapy. In multivariate analysis, adjuvant therapy and MLNR were identified as independent prognostic factors. The reduced recurrence risk among patients receiving adjuvant therapy (Exp(B): 0.234, p=0.038) underscores the efficacy of this treatment. Our results, supported by the literature, suggest that MLNR can guide adjuvant chemotherapy decisions, enabling more precise and individualized treatment planning.

Age was found to be a significant factor for both mortality and recurrence in our study (p<0.001). Older patients are often burdened with more comorbidities and have limited surgical options, which may explain this association. However, in multivariate analysis, age was not an independent risk factor, suggesting potential collinearity with comorbidities and other factors. This observation is consistent with previous reports indicating that older patients with colorectal cancer tend to present with more advanced disease and experience poorer survival outcomes due to comorbidities and limited treatment options (42). The impact of emergency surgery on mortality has been previously reported in the literature. In a study by Hogan et al., it was noted that patients undergoing emergency surgery had worse prognosis, with limitations in adjuvant treatment planning (31). Recent evidence highlights that emergency colorectal resections are associated with significantly worse survival outcomes due to a combination of inadequate surgical conditions and elevated postoperative complication rates (43). Similarly, our study found that patients who underwent emergency surgery had lower survival rates (p=0.023).

This study has several limitations. Its retrospective, single-center design limits the generalizability of the findings and introduces potential selection bias. One of the limitations of our study is the lack of clinical data regarding lifestyle-related risk factors such as obesity, alcohol consumption, and smoking, which prevented their inclusion in the prognostic analysis. The moderate sample size may restrict subgroup analyses and reduce the precision of effect estimates, especially in rare subgroups. The absence of molecular and histopathological data, including microsatellite instability, KRAS/BRAF mutations, and tumor grade, limits the comprehensiveness of recurrence risk modeling and may contribute to residual confounding. A high rate of emergency surgeries (~40%) may have compromised the completeness of lymph node dissection. This could have influenced MLNR calculation, as urgent procedures often deviate from standard oncologic protocols. Additionally, including all lymph nodes regardless of their anatomical relevance might have overestimated total lymph node counts and underestimated MLNR. Future multicenter, prospective studies that incorporate detailed anatomical and tumor-specific lymph node categorization are necessary to validate MLNR as a reliable prognostic tool.