Prognostic impact of nodal classification systems in rectal adenocarcinoma: a retrospective cohort study Authors: Quach Van Kien, Nguyen Thi Thanh Tam, Vu Duc Thinh, Tran Minh Hieu, Than Van Sy Published: 2026 ABSTRACT AbstractBackgroundLymph node metastasis is a key determinant of prognosis and adjuvant treatment in rectal cancer and is currently staged according to the American Joint Committee on Cancer (AJCC) 8th edition pathological N (pN) classification, which is based solely on the number of metastatic lymph nodes. However, prognostic heterogeneity exists among patients within the same pN category, particularly when lymph node retrieval is inadequate. Alternative nodal metrics, including the lymph node ratio (LNR) and log odds of positive lymph nodes (LODDS), have been proposed to improve risk stratification.MethodsThis retrospective study included 170 patients with histologically confirmed rectal adenocarcinoma who underwent curative resection. Cancer-specific survival (CSS) was analyzed using the Kaplan-Meier method, and differences among groups were assessed with the log-rank test. Multivariate Cox proportional hazards models were employed to identify independent prognostic factors. Model performance was evaluated using the area under the curve (AUC), Akaike information criterion (AIC), and chi-square trend and likelihood ratio tests.ResultsAll three classification systems—pN, LNR, and LODDS—were significantly associated with CSS (log-rank P<0.001 for all). In multivariable analysis, each nodal classification system was independently associated with CSS after adjustment for covariates. Among them, LNR showed the slightly higher discriminative ability, with the highest AUC at 3 years (0.664), followed by pN (AUC =0.662) and LODDS (AUC =0.603). Prognostic performance improved notably in patients with ≥12 lymph nodes retrieved.ConclusionspN, LNR, and LODDS are independent prognostic indicators in rectal cancer. Among them, LNR demonstrates slightly higher prognostic performance and may serve as a more effective tool for risk stratification than the AJCC 8th edition pN classification or LODDS, particularly when lymph node retrieval is adequate.Keywords: Rectal cancer, lymph node ratio (LNR), log odds of positive lymph nodes (LODDS), nodal staging, cancer-specific survival (CSS) Highlight box Key findings • Lymph node ratio (LNR) showed slightly better prognostic performance than pathological N (pN) and log odds of positive lymph nodes (LODDS) in rectal adenocarcinoma, particularly in patients with ≥12 retrieved lymph nodes. What is known and what is new? • pN staging remains the standard nodal classification system in rectal cancer, but its prognostic accuracy may be limited when lymph node retrieval is inadequate. LNR and LODDS have been proposed as alternative prognostic tools. This study provides real-world evidence comparing these three systems and demonstrates that LNR showed slightly superior prognostic discrimination and model fit. What is the implication, and what should change now? • LNR may serve as a practical supplementary tool for risk stratification in rectal cancer. Further prospective multicenter studies are needed before routine clinical implementation. Introduction Colorectal cancer remains one of the most common malignancies and a leading cause of cancer-related mortality worldwide, with rectal cancer accounting for a significant proportion of these cases (1). Multimodal treatment is the cornerstone of rectal cancer management, in which curative surgery—typically involving total mesorectal excision (TME) and lymphadenectomy—plays a pivotal role in achieving local control and long-term survival (2). Accurate assessment of lymph node status is critical in rectal cancer staging and prognosis. According to the 8th edition of the American Joint Committee on Cancer (AJCC) tumor-node-metastasis (TNM) staging system, the presence of lymph node metastasis is associated with significantly worse survival outcomes. Five-year overall survival (OS) rates are approximately 99% for stage I, 68–83% for stage II, and only 45–65% for stage III rectal cancer (3). Consequently, nodal involvement is a key determinant in guiding the use of adjuvant chemotherapy. While the pathological N (pN) staging system relies solely on the number of positive lymph nodes, it does not account for the total number of nodes examined. This limitation may lead to stage migration and inaccurate prognostication, particularly in cases where fewer than 12 lymph nodes are harvested (4-7). The current guideline-recommended threshold of ≥12 lymph nodes has been associated with more accurate staging and improved survival (8-11). To address the limitations of the pN system, alternative metrics such as the lymph node ratio (LNR)—defined as the ratio of positive to examined nodes—have been proposed. LNR integrates both the extent of nodal metastasis and the thoroughness of nodal dissection, and has demonstrated prognostic superiority in several studies (12,13). Another emerging metric, the log odds of positive lymph nodes (LODDS), which accounts for both positive and negative lymph nodes, may provide additional discriminatory power, especially in patients with no positive nodes or limited nodal retrieval (13). Despite growing interest, there remains no consensus on which lymph node classification system offers the most reliable prognostic stratification for rectal cancer patients. Therefore, this study aims to compare the prognostic performance of the pN stage, LNR, and LODDS systems in patients with resected rectal adenocarcinoma, and to identify the most suitable classification for predicting cancer-specific survival (CSS). In addition, this study specifically addresses a clinically relevant scenario that is often underrepresented in large datasets, namely patients with suboptimal lymph node retrieval. By focusing on subgroup analyses according to lymph node yield, our study aims to provide a more practical evaluation of nodal classification systems in real-world clinical settings. We present this article in accordance with the STROBE reporting checklist (available at https://tgh.amegroups.com/article/view/10.21037/tgh-2026-0004/rc). Methods This retrospective cohort study was conducted at a high-volume tertiary gastrointestinal surgical center between January 2015 and December 2023. A total of 170 patients with histologically confirmed rectal adenocarcinoma who underwent curative-intent surgery were included. Inclusion criteria were: (I) rectal adenocarcinoma confirmed by endoscopic biopsy; (II) treatment with segmental sigmoid-rectal resection and regional lymphadenectomy; and (III) availability of complete medical records and follow-up data. Patients with recurrent rectal cancer, non-adenocarcinoma histology, secondary rectal involvement from another primary malignancy, or death due to non-cancer-related causes or early postoperative complications were excluded. Curative-intent surgery was defined as R0 resection with no macroscopic or microscopic residual tumor. All consecutive eligible patients during the study period were included and followed until December 2024. Clinical and pathological data were retrospectively collected from hospital records, including patient demographics, neoadjuvant chemoradiotherapy status, tumor histopathology, nodal involvement, total number of lymph nodes examined, and survival outcomes. CSS was defined as the interval from surgery to death due to rectal cancer; deaths from other causes were censored. Patients who were alive at the last follow-up were censored. Survival analysis was performed using the Kaplan-Meier method. Survival rates represent overall CSS during the follow-up period. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Institutional Ethics Committee of Viet Duc University Hospital (approval No. 18/GCN-HĐĐĐ) and the requirement for informed consent was waived due to the retrospective design. Lymph node status was evaluated using three classification systems. First, the 8th edition of the AJCC TNM staging system was applied to classify nodal status into pN0 (no positive lymph nodes), pN1 (1–3 positive nodes), and pN2 (≥4 positive nodes) (14). Second, the LNR was calculated as the number of metastatic lymph nodes divided by the total number of examined nodes, and patients were categorized into four groups based on cutoff values proposed by Qin et al. (15): LNR1 (0< LNR <0.11), LNR2 (0.11≤ LNR <0.39), LNR3 (0.39≤ LNR <0.68), and LNR4 (≥0.68). Third, the LODDS was calculated using the formula log10[(number of positive lymph nodes + 0.5)/(number of negative lymph nodes + 0.5)], where the value of 0.5 was added as a continuity correction to avoid division by zero. Patients were then classified into three groups according to the cut-off values proposed by Han et al. (16): LODDS1 (≤−2.51), LODDS2 (−2.51< LODDS ≤−1.68), and LODDS3 (>−1.68). All statistical analyses were performed using SPSS software (version 20.0; SPSS Inc., Chicago, IL, USA) and R software (version 4.5.1; R Foundation for Statistical Computing, Vienna, Austria). The Kaplan-Meier method was used to estimate CSS, and survival differences were compared using the log-rank test. Variables with statistical significance in univariate analysis were subsequently included in a multivariate Cox proportional hazards model to identify independent prognostic factors. The proportional hazards assumption was evaluated using Schoenfeld residuals, and no significant violations were observed. Time-dependent receiver operating characteristic (ROC) analysis was performed to assess discriminative ability for CSS. The area under the curve (AUC) was estimated at 3 years using methods appropriate for censored survival data, based on the approach proposed by Heagerty et al. (17) and implemented in the “timeROC” package in R. Pairwise comparisons of AUCs were conducted using the DeLong test. Results A total of 249 patients were screened for eligibility, of whom 170 met the inclusion criteria and were included in the final analysis. The study population comprised 103 males and 67 females, with a mean age of 60.9±12.7 years (range, 29–91 years). The mean number of retrieved lymph nodes was 12.7±6.1 (range, 3–42), while the mean number of metastatic lymph nodes was 1.2±2.5 (range, 0–16). Notably, 45.3% of patients (n=77) had fewer than 12 lymph nodes examined. The CSS rate for the entire cohort was 84.1%, with a mean CSS of 95.6±3.4 months (range, 10–115 months). The median CSS was not reached during the follow-up period. There was no statistically significant difference in CSS between patients with <12 and ≥12 examined lymph nodes (log-rank P=0.59), despite the apparent difference in mean survival estimates. Baseline characteristics of the study population are summarized in Table 1. Table 1Clinicopathological characteristics and cancer-specific survival analysis in 170 patients with resected rectal cancerVariableSubgroupPatientsSurvival rate (%)Cancer-specific survival time (months), mean ± SE†χ2 value (log-rank test)P valueAge (years)≤607783.192.1±4.6<0.0010.99>609384.997.0±4.4GenderMale10384.595.9±4.3<0.0010.99Female6783.687.3±4.2Tumor size≤4 cm9288.099.3±4.42.6470.10>4 cm7879.588.6±4.9Neoadjuvant chemoradiotherapyYes3177.466.8±7.73.2680.07No13985.698.4±3.4Histological differentiationWell or moderate10687.799.5±3.94.7060.03Poor or undifferentiated6478.187.5±5.6IMA ligation levelHigh ligation7780.592.3±5.10.8770.35Low ligation9387.189.0±3.8Lymphovascular invasionPresent10281.491.1±4.83.0230.08Absent6888.298.2±4.2Perineural invasionPresent4973.583.0±7.27.0140.008Absent12188.497.1±3.5Mesorectal fascia invasionPresent3969.279.5±8.212.459<0.001Absent13188.590.9±2.9T stage (AJCC 8th edition)Tis + T1 + T25096.088.4±3.07.2950.007T3 + T412079.290.5±4.3N stage (pN, AJCC 8th edition)N011389.499.0±3.321.348<0.001N13287.586.7±7.4N22556.063.3±10.1LNR stageLNR112989.999.4±3.119.219<0.001LNR22475.071.2±9.5LNR31258.367.4±14.4LNR4540.047.8±14.5LODDS stageLODDS110789.789.2±2.813.3230.001LODDS22387.095.9±7.6LODDS34067.575.6±8.4Number of lymph nodes retrieved<127783.197.1±4.50.2930.59≥129384.985.5±4.3†, mean survival estimates may be influenced by censoring and should be interpreted with caution. AJCC, American Joint Committee on Cancer; IMA, inferior mesenteric artery; LNR, lymph node ratio; LODDS, log odds of positive lymph nodes; N, node; pN, pathological N; SE, standard error; T, tumor. Clinicopathological characteristics and cancer-specific survival analysis in 170 patients with resected rectal cancer †, mean survival estimates may be influenced by censoring and should be interpreted with caution. AJCC, American Joint Committee on Cancer; IMA, inferior mesenteric artery; LNR, lymph node ratio; LODDS, log odds of positive lymph nodes; N, node; pN, pathological N; SE, standard error; T, tumor. Kaplan-Meier survival curves stratified by the three lymph node classification systems (pN, LNR, and LODDS) are presented in Figure 1, with numbers at risk displayed below each curve. Kaplan-Meier analysis of CSS in patients with rectal cancer stratified by different lymph node classification systems: (A) CSS curves according to the pN; (B) CSS curves according to the LNR; (C) CSS curves according to the LODDS. All three classification systems demonstrated statistically significant differences in survival outcomes among subgroups (log-rank test, P<0.05). CSS, cancer-specific survival; LNR, lymph node ratio; LODDS, log odds of positive lymph nodes; N, node; pN, pathological N. Survival analysis demonstrated significant differences in CSS according to the pN stage, LNR, and LODDS classification systems (Figure 1A-1C; all P<0.05). In univariate analysis, tumor size, histological differentiation, perineural invasion, mesorectal fascia involvement, T stage, pN stage, LNR, and LODDS categories were significantly associated with CSS (Table 2). Table 2Univariate analysis of CSS in rectal cancerParametersχ2 valueHazard ratio95% CIP valueAge0.2541.0080.978–1.0380.61Gender<0.0010.9970.462–2.1520.99Tumor size3.9481.2601.003–1.5830.047Histological differentiation4.6830.03 Well or moderate (reference)–1.000–– Poor or undifferentiated4.442.2661.059–4.8510.04 Lymphovascular invasion2.8852.0490.895–4.6870.09Perineural invasion6.4522.6621.251–5.6660.01Mesorectal fascia invasion10.8633.6171.684–7.7690.001T stage (AJCC 8th edition)7.2620.007 Tis + T1 + T2 (reference)–1.000–– T3 + T45.6675.7601.363–24.3480.02Number of lymph nodes retrieved1.3521.0340.977–1.0930.28pN (AJCC 8th edition)17.153<0.001 N0 (reference)–1.000–– N10.2341.3220.426–4.1080.63 N216.2025.4112.378–12.311<0.001LNR stage15.6210.001 LNR1 (reference)–1.000–– LNR26.0013.3721.275–8.9150.01 LNR38.3884.6231.640–12.0270.004 LNR48.9276.8001.934–23.9120.003LODDS stage11.5350.003 LODDS1 (reference)–1.000–– LODDS20.0171.0890.304–3.9040.90 LODDS310.4773.7871.691–8.4810.001AJCC, American Joint Committee on Cancer; CI, confidence interval; CSS, cancer-specific survival; LNR, lymph node ratio; LODDS, log odds of positive lymph nodes; N, node; pN, pathological N; T, tumor. Univariate analysis of CSS in rectal cancer AJCC, American Joint Committee on Cancer; CI, confidence interval; CSS, cancer-specific survival; LNR, lymph node ratio; LODDS, log odds of positive lymph nodes; N, node; pN, pathological N; T, tumor. Given the limited number of events (n=27), a parsimonious approach was used to avoid overfitting. Mesorectal fascia invasion showed the strongest and most consistent association with CSS and was therefore selected as the sole covariate across multivariable models to ensure stability and enable fair comparison of pN, LNR, and LODDS. Multivariable Cox regression analyses were then performed to identify independent prognostic factors (Table 3). Table 3Multivariate analysis of cancer-specific survival in rectal cancer for the different lymph node staging systemsParametersMultivariate analysis 1Multivariate analysis 2Multivariate analysis 3HR (95% CI)P valueHR (95% CI)P valueHR (95% CI)P valueMesorectal fascia invasion2.615 (1.165–5.871)0.022.973 (1.280–6.907)0.012.996 (1.374–6.532)0.006pN AJCC 8th0.005 N0 (reference) N11.424 (0.457–4.440)0.54 N24.161 (1.765–9.807)0.001LNR0.01 LNR1 (reference) LNR23.754 (1.409–9.997)0.008 LNR33.307 (1.128–9.692)0.03 LNR43.840 (1.017–14.494)0.047LODDS0.02 LODDS1 (reference) LODDS20.975 (0.271–3.510)0.97 LODDS33.113 (1.368–7.086)0.007AJCC, American Joint Committee on Cancer; CI, confidence interval; HR, hazard ratio; LNR, lymph node ratio; LODDS, log odds of positive lymph nodes; N, node; pN, pathological N. Multivariate analysis of cancer-specific survival in rectal cancer for the different lymph node staging systems AJCC, American Joint Committee on Cancer; CI, confidence interval; HR, hazard ratio; LNR, lymph node ratio; LODDS, log odds of positive lymph nodes; N, node; pN, pathological N. The prognostic performance of the three lymph node classification systems was compared using the linear trend χ2 test, likelihood ratio χ2 test, Akaike information criterion (AIC), and time-dependent ROC analysis. The results are summarized in Table 4, with corresponding ROC curves shown in Figure 2. Table 4Comparison of prognostic ability between lymph node classification systems in rectal cancerClassification systemLinear trend χ2 scoreLikelihood ratio χ2 testAIC valueAUC at 3 yearsLNR17.863 (P<0.001)14.490 (P=0.002)138.333 (R2=0.140)0.664Eighth AJCC pN13.554 (P<0.001)12.062 (P=0.001)140.761 (R2=0.117)0.662LODDS9.860 (P=0.002)9.189 (P=0.002)143.634 (R2=0.090)0.603AIC, Akaike information criterion; AJCC, American Joint Committee on Cancer; AUC, area under the curve; LNR, lymph node ratio; LODDS, log odds of positive lymph nodes; pN, pathological N. Comparison of prognostic ability between lymph node classification systems in rectal cancer AIC, Akaike information criterion; AJCC, American Joint Committee on Cancer; AUC, area under the curve; LNR, lymph node ratio; LODDS, log odds of positive lymph nodes; pN, pathological N. Time-dependent ROC curves of pN, LNR, and LODDS for predicting CSS in rectal cancer at 3 years. AUC, area under the curve; CSS, cancer-specific survival; LNR, lymph node ratio; LODDS, log odds of positive lymph nodes; pN, pathological N; ROC, receiver operating characteristic. Time-dependent ROC analysis showed that LNR and pN had comparable discriminative ability at 3 years, with slightly higher AUC values than LODDS (Figure 2). Pairwise comparisons using the DeLong test revealed no significant differences between LNR and pN at 3 years (P=0.73). Similarly, comparisons between LNR and LODDS (P=0.07) and between pN and LODDS (P=0.10) were not statistically significant at 3 years. To assess the relationship between the number of examined lymph nodes and the three nodal staging systems, Pearson correlation analysis was performed. A weak inverse correlation was observed between the number of lymph nodes retrieved and the LODDS value (r=−0.251, P=0.001). In contrast, no significant correlation was found between the number of lymph nodes retrieved and either the number of metastatic nodes (r=0.141, P=0.07) or the LNR (r=−0.016, P=0.84). A strong linear correlation was observed between LNR and both pN (r=0.917, P<0.001) and LODDS (r=0.921, P<0.001), while pN and LODDS were also closely correlated (r=0.849, P<0.001). Given the retrospective design, potential sources of bias, including selection bias and information bias, may exist. To minimize these biases, uniform inclusion criteria were applied, and data were extracted from standardized medical records. To further evaluate the prognostic performance of the three classification systems in the context of varying lymph node yields, patients were stratified into two subgroups based on the number of lymph nodes examined: <12 nodes (n=77) and ≥12 nodes (n=93). As shown in Table 5, all three systems—pN, LNR, and LODDS—retained their ability to stratify survival outcomes across subgroups. Table 5Evaluation of prognostic value of different lymph node classification systems at varying levels of retrieved lymph nodesVariableSubgroupPatientsSurvival rate (%)χ2 valueP valueLymph node dissection (<12)7783.1 Eighth AJCC pNN05887.99.3540.009N11181.8N2850.0 LNRLNR16387.39.5460.02LNR2683.3LNR3650.0LNR4250.0 LODDSLODDS14887.56.3220.042LODDS21485.7LODDS31566.7Lymph node dissection (≥12)9384.9 Eighth AJCC pNN05590.010.9480.004N12190.5N21758.8 LNRLNR16692.410.7300.01LNR21872.2LNR3666.7LNR4333.3 LODDSLODDS15991.56.6570.04LODDS2988.9LODDS32568.0AJCC, American Joint Committee on Cancer; LNR, lymph node ratio; LODDS, log odds of positive lymph nodes; N, node; pN, pathological N. Evaluation of prognostic value of different lymph node classification systems at varying levels of retrieved lymph nodes AJCC, American Joint Committee on Cancer; LNR, lymph node ratio; LODDS, log odds of positive lymph nodes; N, node; pN, pathological N. Additionally, Table 6 presents a comparative analysis of prognostic metrics (including AUC, AIC, and χ2 scores) for each classification system in both subgroups, revealing that prognostic performance improved in patients with ≥12 nodes retrieved. Table 6Comparison of prognostic value of different lymph node classification systems at varying levels of retrieved lymph nodes for rectal cancerLymph node dissectionClassification systemLinear trend χ2 score (P)Likelihood ratio χ2 test (P)AIC value (R2)AUC at 3 years<12 (n=77)pN (AJCC 8th)6.211 (0.01)6.211 (0.01)6.211 (0.013)0.672LNR6.141 (0.01)6.141 (0.01)6.141 (0.013)0.674LODDS3.009 (0.08)2.833 (0.09)71.088 (0.061)0.608≥12 (n=93)pN (AJCC 8th)8.165 (0.004)7.565 (0.006)75.232 (0.137)0.639LNR12.246 (<0.001)10.656 (0.01)72.141 (0.189)0.640LODDS7.124 (0.008)6.691 (0.01)76.106 (0.121)0.595AIC, Akaike information criterion; AJCC, American Joint Committee on Cancer; AUC, area under the curve; LNR, lymph node ratio; LODDS, log odds of positive lymph nodes; pN, pathological N. Comparison of prognostic value of different lymph node classification systems at varying levels of retrieved lymph nodes for rectal cancer AIC, Akaike information criterion; AJCC, American Joint Committee on Cancer; AUC, area under the curve; LNR, lymph node ratio; LODDS, log odds of positive lymph nodes; pN, pathological N. At 3 years, DeLong test comparisons showed no statistically significant differences in AUC among pN, LNR, and LODDS in either subgroup. In patients with <12 examined lymph nodes, P values were 0.798 (pN vs. LNR), 0.186 (pN vs. LODDS), and 0.160 (LNR vs. LODDS), while in those with ≥12 nodes, the corresponding p-values were 0.976, 0.197, and 0.144, respectively. Subgroup time-dependent ROC analyses according to lymph node yield further confirmed these findings (Figure 3). Time-dependent ROC curves of pN, LNR, and LODDS according to the number of retrieved lymph nodes in rectal cancer: (A) <12 nodes; (B) ≥12 nodes. AUC, area under the curve; LNR, lymph node ratio; LODDS, log odds of positive lymph nodes; pN, pathological N; ROC, receiver operating characteristic. Discussion Postoperative survival prognosis In this retrospective real-world cohort study, we evaluated the prognostic value of three lymph node classification systems—pN, LNR, and LODDS—in patients with curatively resected rectal adenocarcinoma, with particular emphasis on subgroup analyses according to lymph node yield. This approach addresses a clinically relevant gap, as inadequate lymph node retrieval remains common in routine practice and may influence the performance of nodal classification systems. The mean CSS of the cohort was 95.6±3.4 months, with a range of 10 to 115 months. Our analysis identified histological grade, perineural invasion, mesorectal fascia involvement, tumor stage, and nodal status (as defined by all three systems) as significant predictors of CSS. Consistent with previous studies, the pN classification demonstrated a clear association with survival: patients with pN0 had the highest CSS (89.4%), followed by pN1 (87.5%) and pN2 (56.0%), reflecting a stepwise decline in survival with increasing nodal burden. These findings are in line with those reported by Scarinci et al., who found 3-year survival rates of 87.0%, 72.2%, and 61.7% for pN0, pN1, and pN2, respectively (18), and Pei et al., who reported 5-year survival rates ranging from 87.2% (N0) to 44.1% (N2b) in a large cohort of over 56,000 patients (19). The LNR-based classification also showed a strong correlation with survival. CSS rates for LNR1 to LNR4 declined progressively from 89.9% to 40.0%, with increasing LNR values corresponding to poorer survival (P<0.001). This pattern aligns with previous studies, including those by Rosenberg et al. (20) and Karjol et al. (21), who demonstrated that LNR is a reliable and independent prognostic indicator in colorectal cancer, especially in stage III disease. These findings are consistent with previous reports demonstrating a strong association between nodal burden and the risk of distant metastasis. In particular, a recent study by Diefenhardt et al. (22) reported a significant correlation between ypN stage and distant metastatic spread, further supporting the biological relevance of nodal status as a surrogate marker of tumor dissemination. Regarding LODDS, we observed a similar trend: patients classified as LODDS1 and LODDS2 had higher CSS rates (89.7% and 87.0%, respectively), while those in LODDS3 had a lower survival rate (67.5%). This inverse relationship between LODDS and OS is consistent with the findings of Wang et al. (23) and Persiani et al. (24), who reported the prognostic utility of LODDS in large rectal cancer cohorts. However, in our study, LODDS appeared to have slightly lower discriminative power compared to LNR, as reflected by lower AUC and χ2 values. Comparison of prognostic accuracy between pN, LNR, and LODDS classification systems When directly comparing the three systems, all demonstrated statistically significant prognostic discrimination. The LNR system showed the highest linear trend χ2 score, likelihood ratio χ2, and Nagelkerke R2 (0.140), as well as the highest AUC at 3 years (0.664), suggesting slightly better discriminatory ability and model fit compared with pN and LODDS. Our findings are in agreement with those of Qin et al., who analyzed over 70,000 patients and concluded that LNR may be a more practical and slightly better-performing prognostic tool but also more feasible for clinical application than LODDS due to its simpler calculation (15). Nevertheless, a lack of consensus on LNR cut-off points across studies remains a limitation for its widespread adoption in clinical practice (12,16,21,25). Although these cut-offs were derived from cohorts with different clinical characteristics, including advanced-stage disease, their use may provide more stable estimates compared with data-driven thresholds in relatively small samples. In our study, exploratory analyses using cohort-specific cut-offs yielded consistent trends, supporting the robustness of our findings. Impact of the number of retrieved lymph nodes on survival prognosis The number of lymph nodes retrieved is a critical factor influencing staging accuracy. Studies have recommended examining at least 12 lymph nodes to ensure reliable prognostication (8,26). In our cohort, 45.3% of patients had fewer than 12 nodes examined, which may have compromised staging accuracy using the pN system. Similar proportions of inadequate nodal retrieval were reported by Pei et al. (29.4%) (19). The prognostic performance of the three nodal classification systems was further evaluated according to the number of retrieved lymph nodes. In patients with <12 examined lymph nodes (n=77), pN and LNR demonstrated comparable prognostic performance, with similar linear trend χ2 and likelihood ratio χ2 values, while LODDS showed lower discriminatory ability (AUC at 3 years =0.608). In patients with ≥12 lymph nodes (n=93), all three systems showed higher prognostic performance. LNR demonstrated the highest linear trend χ2, likelihood ratio χ2, and Nagelkerke R2, as well as the highest AUC at 3 years (0.640), suggesting slightly better discriminatory ability in this subgroup. However, DeLong test comparisons at 3 years showed no statistically significant differences in AUC among pN, LNR, and LODDS in either subgroup. In patients with <12 examined lymph nodes, the P values were 0.798 (pN vs. LNR), 0.186 (pN vs. LODDS), and 0.160 (LNR vs. LODDS). In patients with ≥12 lymph nodes, the corresponding p-values were 0.976, 0.197, and 0.144, respectively. These findings suggest differences in prognostic discrimination rather than causal effects. Additionally, although the mean CSS appeared higher in the <12 lymph node subgroup, this may reflect censoring effects and differences in event distribution rather than a true prognostic advantage. The Pearson correlation analysis in our study demonstrated that LODDS was weakly associated with the number of examined lymph nodes (r=−0.251, P=0.001), while no significant correlation was observed for pN or LNR. These findings are consistent with the results of Vigorita et al. (27), who also reported that LNR and LODDS maintained prognostic utility independent of nodal yield, particularly when lymph node retrieval was suboptimal. Beyond pathological nodal classification systems, imaging-based nodal assessment may further enhance risk stratification in rectal cancer. The recently proposed Node Reporting and Data System (Node-RADS) provides a standardized framework for evaluating nodal malignancy using computed tomography (CT) and magnetic resonance imaging (MRI). Previous studies have demonstrated good inter-reader agreement and a moderate correlation with histopathologic nodal involvement, supporting its potential as a reproducible imaging adjunct. Integrating imaging-based nodal assessment with pathological nodal burden metrics, such as pN, LNR, and LODDS, may further improve prognostic accuracy and guide individualized treatment strategies. Limitations of the study Despite the strengths of this study, several limitations should be acknowledged. First, the retrospective design may introduce inherent selection and information biases due to variability and incompleteness in medical records. Second, the exclusion of patients with missing data on key variables may have resulted in additional bias. Third, the single-center design and convenience sampling may limit the generalisability of the findings. Furthermore, treatment-related factors were not fully accounted for, including incomplete data on adjuvant therapy, and residual confounding cannot be excluded. Neoadjuvant chemoradiotherapy was not included in the final multivariable model, as it was not significantly associated with CSS in univariate analysis. However, given its potential influence on lymph node yield and survival outcomes, residual confounding related to treatment cannot be fully excluded. In addition, detailed data on adjuvant therapy were not available for all patients and could not be incorporated into the analysis, which may further contribute to residual confounding. Finally, the follow-up duration was confined to the study period, which may not fully capture long-term survival outcomes. Therefore, caution is warranted when extrapolating these results to other populations or clinical settings. Generalisability Although this was a single-center retrospective study, the patient population and treatment strategies reflect routine clinical practice, particularly in the context of variable lymph node yield—an important factor that is often underrepresented in large population-based studies. Therefore, our findings provide additional real-world evidence on the prognostic utility of nodal classification systems in rectal cancer and may be applicable to patients with similar clinicopathological characteristics treated in comparable healthcare settings. Conclusions This study demonstrates that all three nodal classification systems—pN, LNR, and LODDS—serve as independent prognostic indicators in rectal cancer. Among them, LNR showed the highest prognostic performance, offering slightly higher discrimination and model fit. While LNR presents as a promising and practical tool for risk stratification, particularly in patients with adequate lymph node retrieval, it still has limitations that preclude its replacement of the AJCC pN system at present. Future prospective, multicenter studies are warranted to validate standardized LNR thresholds and determine its optimal role in routine clinical practice. Supplementary The article’s supplementary files as Acknowledgments None. 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 Institutional Ethics Committee of Viet Duc University Hospital (approval No. 18/GCN-HĐĐĐ) and the requirement for informed consent was waived due to the retrospective design. Footnotes Reporting Checklist: The authors have completed the STROBE reporting checklist. Available at https://tgh.amegroups.com/article/view/10.21037/tgh-2026-0004/rc Funding: None. Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tgh.amegroups.com/article/view/10.21037/tgh-2026-0004/coif). The authors have no conflicts of interest to declare. Data Sharing Statement Available at https://tgh.amegroups.com/article/view/10.21037/tgh-2026-0004/dss References Available at https://tgh.amegroups.com/article/view/10.21037/tgh-2026-0004/prf Supplementary Materials The article’s supplementary files as Data Availability Statement Available at https://tgh.amegroups.com/article/view/10.21037/tgh-2026-0004/dss