Predictors of lateral lymph node metastasis and prognostic factors in patients with low rectal cancer who underwent lateral lymph node dissection without preoperative treatment Authors: Riki Ohno, Haruka Oi, Soichiro Natsume, Kazuki Kawasaki, Yuichiro Yoshioka, Nao Kakizawa, Junko Kishikawa, Toshiya Nagasaki Published: 2025 ABSTRACT AbstractPurposeThis study aimed to identify predictors of lateral lymph node metastasis (LLNM) and assess prognostic factors in patients with locally advanced low rectal cancer (LALRC), with the goal of informing optimal treatment strategies for LALRC.MethodsWe retrospectively analyzed clinicopathological data from patients with LALRC who underwent lateral lymph node dissection without preoperative treatment between 2014 and 2023. The radiological criterion for LLNM was a short-axis diameter of ≥6 mm on magnetic resonance imaging (MRI).ResultsOf 163 patients, 27 (16.6%) had pathological LLNM (pLLNM). Among 130 patients preoperatively classified as LLNM-negative, 5 (3.8%) were found to have pLLNM. Univariate and multivariate analyses showed that meeting the radiological LLNM criterion independently predicted pLLNM (odds ratio, 53.000; P<0.001). The accuracy, sensitivity, specificity, positive predictive value, and negative predictive value of this criterion were 90.2%, 81.5%, 91.9%, 66.7%, and 96.2%, respectively. In multivariate analyses, pLLNM was an independent risk factor for 3‑year relapse‑free survival. MRI‑detected extramural vascular invasion (mrEMVI) was independently associated with 3‑year relapse‑free survival, local recurrence‑free survival, and distant recurrence‑free survival.ConclusionThese radiological criteria may help clinicians develop personalized treatment plans for patients with LALRC. The high negative predictive value and specificity of LLNM assessment can assist in avoiding overtreatment in appropriate patients. Further evaluation is needed to define optimal management for mrEMVI‑positive cases.Keywords: Rectal cancer, Both lateral lymph node dissection, Lateral lymph node metastasis INTRODUCTION The treatment strategies for patients with locally advanced low rectal cancer (LALRC) are quite different between Japan and other countries. Preoperative treatment for LALRC is not common in Japan. Japanese guidelines recommend total mesorectal excision (TME) and bilateral lateral lymph node dissection (LLND) without preoperative treatment, given the reported 15% to 25% incidence of LLN metastasis (LLNM) in LALRC and the poor prognosis of patients with pathological LLNM (pLLNM) [1–6]. In Western countries, preoperative treatment (radiotherapy, chemoradiotherapy [CRT], and total neoadjuvant therapy [TNT]) followed by TME is the standard of care for rectal cancer. LLND is not regarded as a routine surgical procedure because it has been reported to be associated with higher postoperative complication rates and not to improve overall survival or distant recurrence‑free survival (DRFS). However, a high rate of lateral local recurrence (LR) has been demonstrated in patients with enlarged LLNs after CRT [7–10]. In contrast to international practice, LLND is performed for patients with LALRC without preoperative treatment at our institution. Although we could not evaluate outcomes in patients with LALRC who underwent preoperative treatment, we were able to determine the presence or absence of pLLNM in these cases. The aim of this study was to investigate clinical predictors of pLLNM and other prognostic factors in patients with LALRC. METHODS Ethics statement This study was approved by the Institutional Review Board of Saitama Cancer Center (No. 1969). All patients provided written informed consent. Study design Our standard treatment strategy for LALRC was TME with bilateral LLND without preoperative therapy. LALRC was defined as pathological stage II (T3–4, N0) or stage III (any T, N1–2) rectal adenocarcinoma with the lower tumor margin located below the peritoneal reflection. Preoperative CRT or systemic chemotherapy was reserved for patients with circumferential resection margin (CRM) involvement. CRM involvement, including mesorectal fascia involvement and tumor invasion into adjacent structures, was assessed using magnetic resonance imaging (MRI). Patients who received preoperative treatment and those with multiple synchronous or metachronous cancers, familial adenomatous polyposis, or anal fistula cancer were excluded from analysis. The following variables were retrospectively analyzed: sex, age at surgery, tumor distance from the anal verge (AV), preoperative serum carcinoembryonic antigen (CEA) level, preoperative serum carbohydrate antigen 19‑9 (CA19‑9) level, clinical T category, clinical mesorectal lymph node metastasis (cMLNM), clinical LLNM (cLLNM), clinical stage, clinical CRM status, tumor size, histological type, MRI‑detected extramural vascular invasion (mrEMVI), pathological T category, pathological MLNM (pMLNM), pLLNM, pathological stage, pathological CRM status, and whether postoperative treatment was administered. The depth of tumor invasion and lymph node size were determined by MRI. In line with a previous report, the radiological criterion for cMLNM and cLLNM was a short‑axis diameter of ≥6 mm (Fig. 1A) [11]. mrEMVI was classified according to the MRI‑EMVI scoring system [12], with scores of 0–2 defined as mrEMVI‑negative and scores of 3–4 defined as mrEMVI‑positive (Fig. 1B). Predictors of pLLNM were evaluated by univariate and multivariate analyses. Preoperative magnetic resonance imaging findings. (A) Clinical lateral lymph node metastasis positive (arrow). (B) Magnetic resonance imaging–detected extramural vascular invasion positive (arrow). Postoperative surveillance included measurement of serum CEA and CA19‑9 levels every 3 months for the first 3 years and every 6 months thereafter. Computed tomography of the chest, abdomen, and pelvis was performed every 6 months, and total colonoscopy was performed annually. Long‑term outcomes were assessed as RFS, LR‑free survival (LRFS), and DRFS. RFS was defined as the interval from the date of operation to any relapse. LR was defined as recurrence within the pelvis, and distant recurrence (DR) as recurrence outside the pelvis. For LRFS, events included LR even when DR occurred simultaneously or earlier. Statistical analysis The proportion of patients with pLLNM was analyzed using the chi‑square or Fisher exact tests, as appropriate. Survival analyses for RFS, LRFS, and DRFS were performed using the Kaplan-Meier method and compared with the log‑rank test. A P-value of <0.05 was considered statistically significant. Variables with P<0.10 in univariate analyses were subsequently entered into multivariate models. Multivariate logistic regression was used to identify predictors of pLLNM. To assess predictors of long‑term survival, multivariate analyses using the Cox proportional hazards model were conducted, with results reported as hazard ratios (HRs) and 95% confidence intervals (CIs). All statistical analyses were performed using EZR (Saitama Medical Center, Jichi Medical University). RESULTS Between January 2014 and December 2023, a total of 233 patients with LALRC underwent TME plus bilateral LLND and were diagnosed with pathological stage II or III disease. Preoperative treatment was performed in 48 patients, who were excluded from this analysis. In addition, 18 patients with multiple primary cancers, 3 with anal fistula cancer, and 1 with familial adenomatous polyposis were excluded. Consequently, 163 patients with LALRC who underwent LLND without preoperative treatment were analyzed (Fig. 2). Patient demographics and tumor characteristics are summarized in Table 1. The mean tumor distance from the AV was 47.3 mm. Because of advanced age, severe comorbidity, or patient refusal, 31 patients with clinically positive CRM status did not receive preoperative treatment. pLLNM was confirmed in 27 patients (16.6%). Unilateral and bilateral pLLNM were identified in 19 and 8 patients, respectively. Nineteen patients had both pLLNM and pMLNM, whereas the remaining 8 did not have pMLNM. mrEMVI was observed in 48 patients (29.4%). A total of 93 patients had pathological stage III disease, of whom 71 (76.3%) received postoperative treatment; 2 of 70 patients with pathological stage II disease received postoperative treatment. Patient inclusion flowchart. TME, total mesorectal excision; LLND, lateral lymph node dissection. Table 1.Patient demographics and tumor characteristics (n=163)CharacteristicValueSex Female45 (27.6) Male118 (72.4)Age (yr)64.8±10.7Tumor distance from AV (mm)47.3±18.2Preoperative serum CEA (ng/mL)15.6±37.3Preoperative serum CA19-9 (U/mL)27.4±82.2Clinical T category T23 (1.8) T3134 (82.2) T426 (16.0)cMLNM Negative79 (48.5) Positive84 (51.5)cLLNM Negative130 (79.8) Positive33 (20.2)Clinical stage II72 (44.2) III91 (55.8)Clinical CRM Negative128 (78.5) Positive35 (21.5)Tumor size (mm)52.8±19.9Histological type Well/moderately differentiated154 (94.5) Poorly differentiated, mucinous, signet ring cell9 (5.5)mrEMVI Negative115 (70.6) Positive48 (29.4)Pathological T category T1b, T211 (6.7) T3138 (84.7) T414 (8.6)pMLNM Negative80 (49.1) Positive83 (50.9)pLLNM Negative136 (83.4) Positive27 (16.6)Pathological stage II70 (42.9) III93 (57.1)Pathological CRM Negative149 (91.4) Positive14 (8.6)Postoperative chemotherapy Yes73 (44.8) No90 (55.2)Values are presented as number (%) or mean±standard deviation.AV, anal verge; CEA, carcinoembryonic antigen; CA19-9, carbohydrate antigen 19-9; cMLNM, clinical mesorectal lymph node metastasis; cLLNM, clinical lateral lymph node metastasis; CRM, circumferential resection margin; mrEMVI, magnetic resonance imaging–detected extramural vascular invasion; pMLNM, pathological mesorectal lymph node metastasis; pLLNM, pathological lateral lymph node metastasis. Patient demographics and tumor characteristics (n=163) Values are presented as number (%) or mean±standard deviation. AV, anal verge; CEA, carcinoembryonic antigen; CA19-9, carbohydrate antigen 19-9; cMLNM, clinical mesorectal lymph node metastasis; cLLNM, clinical lateral lymph node metastasis; CRM, circumferential resection margin; mrEMVI, magnetic resonance imaging–detected extramural vascular invasion; pMLNM, pathological mesorectal lymph node metastasis; pLLNM, pathological lateral lymph node metastasis. Table 2 shows the association between preoperative clinical factors and pLLNM. In univariate analysis, cMLNM‑positive and cLLNM‑positive status were significantly associated with pLLNM. Multivariate analysis revealed that cLLNM‑positive status was an independent predictor of pLLNM (odds ratio, 53.000; 95% CI, 15.100–185.00; P<0.001). Using our radiological criterion, 33 patients were classified as cLLNM‑positive, and of these patients, pLLNM was confirmed in 22 (66.7%). Among 130 patients classified as cLLNM‑negative, 5 (3.8%) were found to have pLLNM. The accuracy, sensitivity, specificity, positive predictive value, and negative predictive value (NPV) of the cLLNM criterion were 90.2%, 81.5%, 91.9%, 66.7%, and 96.2%, respectively (Fig. 3). Moreover, among 71 patients who were both cLLNM‑negative and cMLNM‑negative, pLLNM was confirmed in 2 (2.8%). Of 59 patients with cMLNM‑positive and cLLNM‑negative status, 3 (5.1%) were diagnosed with pLLNM. Table 2.The association between pLLNM and preoperative clinical factors (n=163)FactorpLLNMUnivariate analysisMultivariate analysisNegative (n=136)Positive (n=27)P-valueOR (95% CI)P-valueSex0.816 Female378 Male9919Age (yr)0.531 <656010 ≥657617Tumor distance from AV (mm)0.184 >50516 ≤508521Preoperative serum CEA (ng/mL)0.294 <56710 ≥56917Preoperative serum CA19–9 (U/mL)0.0840.257 <37124211 (Reference) ≥371262.618 (0.495–13.889)Clinical T category0.294 T221 T311420 T4206cMLNM0.0360.643 Negative7181 (Reference) Positive65191.371 (0.361–5.208)cLLNM<0.001<0.001 Negative12551 (Reference) Positive112253.000 (15.100–185.000)Clinical CRM0.305 Negative10919 Positive278mrEMVI0.0690.242 Negative100151 (Reference) Positive36122.170 (0.593–7.950)Histological type0.645 Well/moderately differentiated12925 Poorly differentiated, mucinous, signet ring cell72pLLNM, pathological lateral lymph node metastasis; OR, odds ratio; CI, confidence interval; AV, anal verge; CEA, carcinoembryonic antigen; CA19-9, carbohydrate antigen 19-9; cMLNM, clinical mesorectal lymph node metastasis; cLLNM, clinical lateral lymph node metastasis; CRM, circumferential resection margin; mrEMVI, magnetic resonance imaging–detected extramural vascular invasion. The association between pLLNM and preoperative clinical factors (n=163) pLLNM, pathological lateral lymph node metastasis; OR, odds ratio; CI, confidence interval; AV, anal verge; CEA, carcinoembryonic antigen; CA19-9, carbohydrate antigen 19-9; cMLNM, clinical mesorectal lymph node metastasis; cLLNM, clinical lateral lymph node metastasis; CRM, circumferential resection margin; mrEMVI, magnetic resonance imaging–detected extramural vascular invasion. Prediction of lateral lymph node metastasis based on lymph node size. cLLNM, clinical lateral lymph node metastasis; pLLNM, pathological lateral lymph node metastasis; TP, true positive; FP, false positive; FN, false negative; TN, true negative. With a median follow‑up of 46.6 months (range, 0.5–124.9 months), 36 patients developed DR and 12 developed LR. Of the 12 patients with LR, 7 recurred in the lateral pelvic region, 3 in the presacral area, and 2 at the anastomotic site. Among the 27 patients with pLLNM, 3 had LR in the lateral pelvic area and 1 in the presacral area. Among the 5 patients who were cLLNM‑negative but had pLLNM, none developed LR and 4 developed DR. The 3‑year RFS, LRFS, and DRFS were 72.7%, 91.2%, and 77.6%, respectively. Table 3 shows correlations between clinicopathological factors and 3‑year outcomes. In univariate analyses, sex and age were not associated with prognosis. Tumor distance from the AV, pathological T category, pMLNM, pLLNM, pathological CRM‑positivity, and histological type were significantly associated with 3‑year RFS and DRFS. Preoperative serum CEA level was significantly associated with RFS and LRFS. mrEMVI was significantly associated with 3‑year RFS, LRFS, and DRFS. Table 4 presents the multivariate analyses of prognostic factors. Tumor distance from the AV was an independent predictor of 3‑year RFS and DRFS. pLLNM was independently associated with RFS. mrEMVI was an independent predictor of RFS, LRFS, and DRFS. Histological type was independently associated with DRFS. Table 3.The correlation between clinicopathological factors and 3-year prognosis (n=163)FactorNo. of patientsRFS (%)P-valueLRFS (%)P-valueDRFS (%)P-valueSex0.5470.6380.640 Female4571.591.476.5 Male11868.788.875.9Age (yr)0.1070.8380.095 <657076.489.782.6 ≥659363.991.870.5Tumor distance from AV (mm)<0.0010.149<0.001 >505793.295.293.5 ≤5010657.188.467.2Preoperative serum CEA (ng/mL)0.0240.0360.202 <57780.696.581.4 ≥58660.985.871.7Preoperative serum CA19–9 (U/mL)0.0540.4240.141 <3714571.091.177.0 ≥371856.786.766.6Pathological T category0.0380.1870.009 T1b–314971.691.378.5 T41443.171.843.1pMLNM0.0170.3320.004 Negative8079.688.487.0 Positive8361.292.461.7pLLNM<0.0010.056<0.001 Negative13675.892.481.1 Positive2728.280.749.0Pathological CRM<0.0010.348<0.001 Negative14972.590.079.4 Positive1430.510032.5mrEMVI<0.001<0.001<0.001 Negative11583.095.885.3 Positive4829.175.850.8Histological type<0.0010.484<0.001 Well/moderately differentiated15471.991.078.8 Poorly differentiated, mucinous, signet ring cell933.380.033.3Postoperative treatment0.8520.9730.625 No9070.291.575.2 Yes7369.290.177.6RFS, recurrence-free survival; LRFS, local recurrence-free survival; DRFS, distant recurrence-free survival; AV, anal verge; CEA, carcinoembryonic antigen; CA19-9, carbohydrate antigen 19-9; pMLNM, pathological mesorectal lymph node metastasis; pLLNM, pathological lateral lymph node metastasis; CRM, circumferential resection margin; mrEMVI, magnetic resonance imaging–detected extramural vascular invasion. The correlation between clinicopathological factors and 3-year prognosis (n=163) RFS, recurrence-free survival; LRFS, local recurrence-free survival; DRFS, distant recurrence-free survival; AV, anal verge; CEA, carcinoembryonic antigen; CA19-9, carbohydrate antigen 19-9; pMLNM, pathological mesorectal lymph node metastasis; pLLNM, pathological lateral lymph node metastasis; CRM, circumferential resection margin; mrEMVI, magnetic resonance imaging–detected extramural vascular invasion. Table 4.Multivariate analyses of the predictive factors for 3-year prognosisFactorRecurrence-free survivalLocal recurrence-free survivalDistant recurrence-free survivalHR (95% CI)P-valueHR (95% CI)P-valueHR (95% CI)P-valueTumor distance from AV (mm)0.0090.037 >501 (Reference)1 (Reference) ≤504.196 (1.437–12.250)3.172 (1.071–9.395)pLLNM0.0470.1200.224 Negative1 (Reference)1 (Reference)1 (Reference) Positive2.086 (1.010–4.308)2.614 (0.779–8.772)1.626 (0.743–3.562)mrEMVI<0.0010.0070.012 Negative1 (Reference)1 (Reference)1 (Reference) Positive3.838 (1.939–7.600)5.225 (1.561–17.490)2.601 (1.230–5.502)Histological type0.039 Well/moderately differentiated1 (Reference) Poorly differentiated, mucinous, signet ring cell2.604 (1.048–6.494)The table includes significant risk factors alone, apart from pLLNM.HR, hazard ratio; CI, confidence interval; AV, anal verge; pLLNM, pathological lateral lymph node metastasis; mrEMVI, magnetic resonance imaging–detected extramural vascular invasion. Multivariate analyses of the predictive factors for 3-year prognosis The table includes significant risk factors alone, apart from pLLNM. HR, hazard ratio; CI, confidence interval; AV, anal verge; pLLNM, pathological lateral lymph node metastasis; mrEMVI, magnetic resonance imaging–detected extramural vascular invasion. DISCUSSION Our data indicated that an LLN short‑axis diameter ≥6 mm on MRI was strongly associated with pLLNM and yielded high NPV and specificity. Patients with LALRC who had pLLNM experienced poorer prognosis than those without pLLNM. Therefore, elucidating predictors of pLLNM is important. A strength of this study is that the accuracy of cLLNM was verified histopathologically in all cases. In our cohort, pLLNM was confirmed in 16.6% of patients with LALRC who underwent TME plus LLND without preoperative treatment, and this incidence is comparable to previous reports [1–5]. Multiple studies have shown that LLN short‑axis diameter is a valuable predictor of LLNM [13]. A recent Japanese multicenter prospective study identified ≥6 mm as the optimal cutoff by receiver operating characteristic analysis and confirmed it as an independent risk factor for LLNM [11]. Our results corroborate the diagnostic performance of this cutoff. Notably, the NPV (96.2%) and specificity (91.9%) were very high. Thus, the current radiological criterion for cLLNM may help prevent overtreatment in patients unlikely to harbor LLNM. However, LLN size alone may be insufficient for maximal accuracy. Morphologic MRI features, such as irregular borders and mixed signal intensity, can improve diagnostic accuracy [14]. Some studies have also shown that combining mrEMVI status with LLN size enhances performance [15, 16], and a validated nomogram has been developed incorporating LLN and perirectal lymph node short‑axis diameters, EMVI status, and other variables [17]. As noted in our study, pLLNM occurred in 2.8% of patients with both negative cLLNM and cMLNM, suggesting that the absence of enlarged MLNs may aid pLLNM prediction. These findings are useful for refining preoperative risk stratification. Treatment strategies for LALRC should be individualized according to LLN status, since LLND is associated with longer operative time, greater blood loss, male sexual dysfunction, and urinary dysfunction compared with TME alone [18, 19]. LLND is not commonly performed outside Japan. Worldwide, the standard of care for rectal cancer is neoadjuvant CRT followed by TME without LLND. Patients with LALRC and enlarged LLNs who undergo neoadjuvant CRT followed by TME alone have been reported to experience significantly higher lateral LR rates [7–10], likely because pLLNM is identified in 24.6% to 65.8% of patients with enlarged LLNs after CRT [20–22]. LLN metastases may not be completely eradicated by preoperative CRT. Our results showed that 3‑year RFS and DRFS were significantly worse in patients with pLLNM than in those without pLLNM despite LLND. In this study, LR also occurred in regions outside the LLN area. Moreover, 11.1% of patients with pLLNM had LR in the lateral pelvic area, a rate higher than in patients without pLLNM. These findings suggest that LLND alone is insufficient to prevent pelvic recurrence and may not fully eliminate LLN recurrence in patients with pLLNM. A previous report showed that LR rates were similar between Japanese patients who underwent TME plus LLND and Western patients who underwent preoperative CRT followed by TME without LLND [23]. Although our data do not include outcomes for patients who received preoperative CRT, long‑term results, including overall survival, RFS, LRFS, and lateral LR, reportedly improve in pLLNM when neoadjuvant CRT is followed by TME plus LLND [22, 24, 25]. Collectively, these reports indicate that combining neoadjuvant CRT with LLND could be a promising strategy for LALRC with enlarged LLNs. The necessity of LLND for LALRC without enlarged LLNs remains unclear. A Japanese multicenter randomized controlled trial (JCOG0212) tested the noninferiority of TME alone versus TME with LLND in patients without enlarged LLNs. Although noninferiority was not proven, 5‑year overall survival, RFS, and LRFS were similar between the groups [26, 27]. In our series, using the present radiological criterion, 5 of 130 cLLNM‑negative patients (3.8%) had pLLNM. Given the adverse short‑term effects of LLND, the procedure can be omitted in cLLNM‑negative LALRC. Our study also showed that mrEMVI was an independent prognostic factor in multivariate analysis. By Kaplan-Meier estimates, 3‑year RFS, LRFS, and DRFS were significantly worse in mrEMVI‑positive than in mrEMVI‑negative patients (29.1% vs. 83.0%, 75.8% vs. 95.8%, and 50.8% vs. 85.3%, respectively). The presence of EMVI is mechanistically linked to a higher risk of DR due to hematogenous spread [11, 28], and mrEMVI has been associated with increased CRM involvement and pelvic sidewall dissemination [29,30]. A previous report demonstrated that regression of mrEMVI after CRT is associated with improved outcomes [31]. Therefore, even in cLLNM‑negative patients, multimodal therapy should be considered for those who are mrEMVI‑positive. In this study, 71 of 93 patients (76.3%) with pathological stage III disease received postoperative chemotherapy after bilateral LLND, and postoperative chemotherapy was not associated with improved prognosis. Recently, TNT has been reported to increase chemotherapy completion and improve outcomes in rectal cancer [32–35]. Further research is needed to define optimal strategies for LALRC with this prognostic risk factor. In addition to mrEMVI and pLLNM, tumor distance from the AV (within 50 mm) and poor histological type were independently associated with poor prognosis in multivariate analyses. In the total cohort, 41 patients had none of these risk factors and experienced neither LR nor DR. These patients might not require preoperative therapy or LLND. Limitations This study has limitations. First, it was a retrospective, single‑center analysis with a limited sample size; thus, selection bias is possible. Second, 14 patients (8.6%) had less than 1 year of follow‑up; longer observation is needed to fully evaluate the prognostic factors identified here. Third, detailed morphologic MRI findings of LLNs and pathological confirmation of EMVI were not evaluated. We focused on LLN size and EMVI findings on MRI, and these simple features were clearly associated with prognosis. Fourth, because our treatment strategy differs from international standards, the results should be interpreted with caution. Despite these limitations, we believe our findings may assist with treatment selection for patients with LALRC. Conclusions Our radiological criterion for cLLNM was independently associated with pLLNM, with very high NPV and specificity. Accurate pretreatment diagnosis is essential to determine appropriate treatment strategies for patients with LALRC. Footnotes Conflict of interest No potential conflict of interest relevant to this article was reported. Funding None. Author contributions Conceptualization: all authors; Data curation: all authors; Formal analysis: all authors; Investigation: all authors; Methodology: all authors; Writing–original draft: all authors; Writing–review & editing: all authors. All authors read and approved the final manuscript. REFERENCES