Prognostic divergence of hematogenous and lymphatic metastases in surgically treated patients with stage IV colorectal cancer Authors: Haoyue Ma, Liming Wang, Bolun Song, Yinggang Chen, Yasumitsu Hirano Published: 2026 ABSTRACT AbstractBackgroundIn patients with stage IV colorectal cancer (CRC), the prognostic impact of lymph node metastasis (LNM) remains controversial. A qualifying cohort served to explore its importance.MethodsA total of 493 eligible patients with stage IV CRC were included in this retrospective cohort study. All were surgically treated between April 2007 and December 2020 at a high-volume cancer center in Japan. Subjects were stratified by presence/absence of LNM (N+M1 vs N0M1), and propensity score matching (PSM) was done at 1:1 ratio. We then compared cancer-specific (CSS) and overall (OS) survival rates before and after PSM, using Cox regression to identify pertinent independent risk factors.ResultsPatients assigned to N+M1 and N0M1 groups totaled 384 (77.9%) and 109 (22.1%), respectively. Five-year OS proved superior for the N0M1 (vs N+M1) group, both before (41.7% vs 30.8%; p = 0.020) and after (43.7% vs 26.5%; p = 0.042) PSM. LNM also emerged as an independent prognosticator of OS in multivariate analyses conducted before (hazard ratio [HR] = 1.5, 95% confidence interval [CI]: 1.09–2.08; p = 0.014) and after (HR = 1.72, 95% CI: 1.121–2.64; p = 0.013) matching.The same was true of 5-year CSS, both before (N0M1, 46.1%; N+M1, 32.4%; p = 0.009) and after (N0M1, 48.8%; N+M1, 28.0%; p = 0.027) matching, again verifying LNM as an independent prognostic factor before (HR = 1.6, 95% CI: 1.13–2.25; p = 0.007) and after (HR = 1.86, 95% CI: 1.185–2.90; p = 0.007) PSM.ConclusionsIn patients with stage IV colorectal cancer (M1), those with distant hematogenous metastasis without regional lymph node involvement (N0M1) have a better prognosis than those with concurrent regional lymph node metastasis (N+M1). This apparent prognostic divergence must be reflected in refined tumor classification subsets.Keywords: Stage IV colorectal cancer, Lymph node metastasis, Propensity score matching Introduction Colorectal cancer (CRC) is one of the leading causes of cancer deaths worldwide [1]. Although early detection is achievable through screening, distant metastases are seen in a number of patients (20–25%) at time of discovery [2]. Historically, such patients have done poorly, but recent gains through surgical advances and novel chemotherapeutic agents [3] have fueled efforts to subdivide stage IV CRC on the basis of patient survival. The 8th edition of the American Joint Committee on Cancer/Union for International Cancer Control (AJCC/UICC) staging manual now stipulates three subsets of stage IV CRC (IVA, IVB, and IVC [M1a/M1b/M1C]), a move requiring commensurate survival stratification. However, spread to distant organs is chiefly addressed at present [4], largely ignoring lymph node metastasis (LNM) in the staging hierarchy. For stage III CRC, LNM status is the most important prognostic factor [5], yet it is unclear whether LNM is truly prognostic of stage IV CRC. There is controversy surrounding this issue, some sources confirming [6–8] and others denying [9] the ramifications thereof. No doubt, the reason for such controversy is the paucity of related research [8, 10, 11]. In addition to lymph node metastasis, tumor deposits have emerged as a promising prognostic factor for recurrence (especially liver recurrence) in CRC, and are classified as N+ status in current staging systems [12]. Recent studies have hinted at potential revisions to the AJCC classification and individualized chemotherapy regimens based on N+ status, highlighting the need to refine prognostic stratification in metastatic CRC [13]. However, the prognostic impact of lymph node involvement in stage IV CRC remains controversial, and few studies have compared the outcomes of N0M1 and N+M1 subsets, which is the focus of the present investigation. To further improve outcomes of stage IV CRC, it is imperative that survival factors be established through patient stratification, identifying those individuals amenable to cure. Our intent during this study was to explore the impact of LNM on patient survival in the setting of stage IV CRC. Patients and methods This study was retrospective in nature, drawn from a 13-year body of data (April 2007 to December 2020) collected at Saitama Medical University International Medical Center in Japan. Our protocol received approval of the institutional Ethics Committee, with written informed consent granted by all subjects preoperatively. The study was approved by the Ethics Committee of Saitama Medical University International Medical Center (Approval No.:2019-006). A total of 626 patients underwent surgical treatment for stage IV CRC. Cases with simultaneous or metachronous resection of distant metastases were included in the study, while patients with recurrences of stage IV CRC or histories of preoperative chemotherapy were excluded. Additionally, due to inadequate evaluation of peritoneal metastasis severity and the diverse locations of distant LNM, those with only peritoneal metastases (n = 78) or only distant LNM (n = 55) were excluded. This left a final cohort of 493 eligible subjects, who were grouped as N+M1 (n = 384) or N0M1 (n = 109) (Fig. 1). The median follow-up duration was 68 months (interquartile range [IQR] 42–96 months) for the entire cohort, and 65 months (IQR 40–92 months) for the matched cohort. Schematic of patient allocation/study design All tumor (T) and lymph node (N) classifications adhered to guidelines of the Japanese Society for Cancer of the Colon and Rectum (JSCCR, 7th Edition). Mesenteric lymph nodes were thus separable into pericolic, intermediate, and apical nodes. Initially, we compared clinicopathologic characteristics and survival rates by group (i.e., extent of LNM). Perioperative outcomes were then reviewed, to include additional data on stage III CRC. Propensity score matching (PSM) served to offset differences in baseline variables, and multivariate regression models were used to identify potential independent risk factors. The following parameters were incorporated into a multivariate model for PSM: sex; age; family history of cancer; carcinoembryonic antigen (CEA) level; tumor location, size, histotype, differentiation, pathologic T/N classification, lymphatic invasion, perineural infiltration, venous invasion, and distribution of distant organ metastases. Propensity score matching was restricted to pre-treatment baseline characteristics to avoid potential bias from post-exposure variables. Postoperative complications, hospital stay, and adjuvant chemotherapy were included in multivariate Cox regression models as confounding factors rather than matching variables, as they may be influenced by treatment decisions and disease severity. Cox proportional hazards regression analysis was performed to identify independent prognostic factors for overall survival (OS) and cancer-specific survival (CSS). The proportional hazards assumption of the Cox regression model was verified using Schoenfeld residuals test, and no significant violations were observed (p > 0.05 for all variables). CSS was defined as the time from surgery to death due to colorectal cancer; deaths from other causes were censored as competing risks. For the purpose of comparing 5-year survival status between groups, a binary logistic regression analysis was additionally conducted (with 5-year survival as the dependent variable: 1 = survived, 0 = deceased). To compare categorical variables, chi-square and Fisher’s exact tests were applied, and Kaplan-Meier plots were generated for CSS estimates. All statistical computations relied on standard software (SPSS v22; IBM Corp, Tokyo, Japan), setting significance at p < 0.05. Results CRC survival rates (stage III vs stage IV) with and without LNM As a related exercise, we grouped 1060 patients with stage III CRC according to LNM locations and counts, using pericolic, intermediate, and apical compartments as measures of remoteness from primary tumors. The resultant 5-year OS rates were 83.3%, 77.7%, and 63.2%, respectively (Fig. 2A). We then analyzed our cohort (n = 493) with stage IV CRC. Of the 493 included patients, 326 (66.1%) underwent curative-intent resection of metastases (R0 resection), and 167 (33.9%) received palliative resection. The surgical strategy was determined based on tumor resectability: 245 (49.7%) patients underwent primary tumor-first resection, and 181 (36.7%) received metastasis-first resection (mostly liver-first for resectable liver metastases). Multivariate Cox proportional hazards analysis of overall survival (OS) before and after propensity score matching (PSM) Five-year OS in the N0M1 group was 41.7%, whereas N1M1 and N2M1 subsets of the N+M1 group showed 5-year OS rates of 35.0% and 26.8%, respectively (Fig. 2A). Five-year CSS rates across these six subsets similarly reflected worsening prognosis as the scope of primary CRC and nodal involvement advanced, recorded as 86.1%, 80.3%, 63.8%, 46.1%, 37.1%, and 27.6%, respectively (Fig. 2B). Clinicopathologic features of stage IV CRC cohort Once all qualifying patients (n = 493) with stage IV CRC had been grouped by LNM status (N0M1: 109/493, 22.1%; N+M1: 384/493, 77.9%) (Table 1), our analysis indicated greater propensity for lymphatic (60.9% vs 33.0%; p < 0.001) and venous (89.8% vs 78.0%; p = 0.008) invasion in the presence (N+M1) vs absence (N0M1) of LNM. The N0M1 (vs N+M1) group registered a lengthier mean hospital stay (15.1 ± 9.0 vs 10.3 ± 8.0; p = 0.03), but the two groups did not differ significantly in terms of family history of cancer, carcinoembryonic antigen (CEA) level, tumor characteristics (location, size, histotype, differentiation), pathologic T/N classifications, postoperative complications, or adjuvant chemotherapy (Table 1). Table 1Clinical characteristics and surgical outcomes of patient population before and after propensity score matching (PSM)ParametersBefore PSMAfter PSMN0M1N+M1P-valueN0M1N+M1P-valueSex (n)1093848787 Female36 (33.0%)130 (33.9%)NS31 (35.6%)30 (34.5%)NS Male73 (67.0%)254 (66.1%)56 (64.4%)57 (65.5%)Age (years) < 7056 (51.4%)214 (55.7%)NS44 (50.6%)51 (58.6%)NS ≥ 7053 (48.6%)170 (44.3%)43 (49.4%)36 (41.4%)Family history of cancer No42 (38.5%)163 (42.3%)NS34 (39.1%)37 (42.5%)NS Yes67 (61.5%)221 (57.6%)53 (60.9%)50 (57.5%)CEA (ng/ml) < 526 (23.9%)104 (27.1%)NS20 (23.0%)25 (28.7%)NS ≥ 583 (76.1%)280 (72.9%)67 (77.0%)62 (71.3%)Tumor location Colon67 (61.5%)222 (58.7%)NS56 (64.4%)49 (56.3%)NS Rectum42 (38.5%)162 (42.2%)31 (35.6%)38 (43.7%)Tumor size (cm) < 555 (50.5%)215 (56.0%)NS41 (47.1%)47 (54.0%)NS ≥ 554 (49.5%)169 (44.0%)46 (52.9%)40 (46.0%)Histotype Adenocarcinoma97 (89.0%)341 (88.8%)76 (87.4%)80 (92.0%) Sig, Muc12 (11.0%)43 (11.2%)NS11 (12.6%)7 (8.0%)NSDifferentiation Well, Mod94 (86.2%)298 (77.6%)73 (83.9%)75 (86.2%) Poor, Sig, Muc15 (13.8%)86 (22.4%)NS14 (16.1%)12 (13.8%)NST classification T1-T26 (5.5%)12 (3.1%)3 (3.4%)2 (2.3%) T3-T4103 (94.5%)372 (96.9%)NS84 (96.6%)85 (97.7%)NSN classification N0-N1114 (59.4%)205 (52.7%)83 (59.3%)79 (76.4%) N278 (40.6%)184 (47.3%)NS57 (40.7%)61 (43.6%)NSLymphatic invasion No73 (67.0%)150 (39.1%) < 0.00156 (64.4%)56 (64.4%)NS Yes36 (33.0%)234 (60.9%)31 (35.6%)931 (35.6%)Venous invasion No22 (20.2%)39 (10.2%)0.00813 (14.9%)13 (14.9%)NS Yes87 (78.0%)345 (89.8%)74 (85.1%)74 (85.1%)Liver metastasis No24 (22.0%)88 (22.9%)NS17 (19.5%)20 (23.0%)NS Yes85 (78.0%)296 (77.1%)70 (80.5%)67 (77.0%)Lung metastasis No76 (69.7%)262 (68.2%)NS65 (74.7%)59 (67.8%)NS Yes33 (30.3%)122 (31.8%)22 (25.3%)28 (32.2%)Bone metastasis No107 (98.2%)375 (97.7%)NS85 (97.7%)85 (97.7%)NS Yes2 (1.8%)9 (2.3%)2 (2.3%)2 (2.3%)Hospital stay (days) Mean ± SD15.1 ± 9.010.3 ± 8.00.0310.8 ± 8.09.83 ± 7.0NSPostoperative complications No74 (67.9%)295 (76.8%)0.07667 (77.0%)64 (73.6%)NS Yes35 (32.1%)89 (23.2%)20 (23.0%)23 (26.4%)Adjuvant chemotherapy No/unknown68 (62.4%)241 (62.8%)NS58 (66.7%)53 (60.9%)NS Yes41 (37.6%)143 (37.2%)29 (33.3%)34 (39.1%)CEA carcinoembryonic antigen, Well well differentiated, Mod moderately differentiated, Poor poorly differentiated, Muc mucinous carcinoma, Sig signet ring cell carcinoma, SD standard deviation, NS not significant Clinical characteristics and surgical outcomes of patient population before and after propensity score matching (PSM) Hospital stay (days) Mean ± SD CEA carcinoembryonic antigen, Well well differentiated, Mod moderately differentiated, Poor poorly differentiated, Muc mucinous carcinoma, Sig signet ring cell carcinoma, SD standard deviation, NS not significant Outcomes of LNM analysis before matching In analyzing long-term outcomes of LNM subsets, the N0M1 (vs N+M1) group demonstrated significantly better 5-year OS (41.7% vs 30.8%; p = 0.020) (Fig. 3A1) and CSS (46.1% vs 32.4%; p = 0.009) (Fig. 3A2). Multivariate Cox proportional hazards analysis of cancer-specific survival (CSS) before and after propensity score matching (PSM): (A1) OS before matching; (A2) CSS before matching; (B1) OS after matching; (B2) CSS after matching Cox regression analysis was done before PSM to assess effects of various parameters on patient prognosis. LNM subsequently emerged as a significant independent predictor of OS (hazard ratio [HR] = 1.5, 95% CI: 1.09–2.08; p = 0.014), as did CEA level (HR = 1.37, 95% CI: 1.03–1.84; p = 0.033), tumor histotype (HR = 0.54, 95% CI: 0.32–0.94; p = 0.03), tumor differentiation (HR = 0.45, 95% CI: 0.30–0.67; p < 0.001), liver metastasis (HR = 1.54, 95% CI: 1.02–2.32; p = 0.041), lung metastasis (HR = 1.63, 95% CI: 1.16–2.29; p = 0.005), and chemotherapy (HR = 0.53, 95% CI: 0.40–0.71; p < 0.001) (Table 2). Table 2Multivariate Cox proportional hazards analysis of overall survival (OS) before and after propensity score matching (PSM)VariableOS before PSMOS after PSMHR95% ClP-valueHR95% ClP-valueLNM+ vs LNM-1.5(1.09–2.08)0.0141.72(1.121–2.64)0.013Colon vs rectum0.81(0.62–1.06)NS0.64(0.404–1.03)NSAge (years) < 70 vs ≥ 701.26(0.98–1.64)NS1.05(0.677–1.61)NSTumor family history, yes vs no0.93(0.71–1.20)NS0.83(0.539–1.28)NSPostoperative complications, yes vs no1.3(0.98–1.73)NS1.13(0.704–1.83)NSCEA (ng/ml), < 5 vs ≥ 51.37(1.03–1.84)0.0331.29(0.756–2.19)NSTumor size (cm), < 5 vs ≥ 51.03(0.79–1.33)NS0.99(0.648–1.51)NST classification, 1–2 vs 3–41.37(0.63–2.94)NS4.16(0.553–31.37)NSHistotype, AdenoCA vs Muc, Sig0.54(0.32–0.94)0.030.31(0.099–0.95)0.04Liver metastasis, yes vs no1.54(1.02–2.32)0.0410.95(0.429–2.11)NSLung metastasis, yes vs no1.63(1.16–2.29)0.0051.23(0.624–2.43)0.54Differentiation, Well/Mod vs Poor/Muc/Sig0.45(0.30–0.67) < 0.0010.19(0.076–0.47) < 0.001Chemotherapy, yes vs no0.53(0.40–0.71) < 0.0010.49(0.306–0.79)0.004LNM lymph node metastasis, CEA carcinoembryonic antigen, Well well differentiated, Mod moderately differentiated, Poor poorly differentiated, AdenoCA adenocarcinoma, Muc mucinous carcinoma, Sig signet ring cell carcinoma, NS not significant Multivariate Cox proportional hazards analysis of overall survival (OS) before and after propensity score matching (PSM) LNM lymph node metastasis, CEA carcinoembryonic antigen, Well well differentiated, Mod moderately differentiated, Poor poorly differentiated, AdenoCA adenocarcinoma, Muc mucinous carcinoma, Sig signet ring cell carcinoma, NS not significant Findings were similar for 5-year CSS, LNM again constituting a significant independent predictor (HR = 1.6, 95% CI: 1.13–2.25; p = 0.007), along with CEA level (HR = 1.40, 95% CI: 1.04–1.90; p = 0.028), tumor histotype (HR = 0.54, 95% CI: 0.31–0.94; p = 0.029), tumor differentiation (HR = 0.43, 95% CI: 0.29–0.64; p < 0.001), liver metastasis (HR = 1.62, 95% CI: 1.06–2.46; p = 0.024), lung metastasis (HR = 1.78, 95% CI: 1.26–2.52; p = 0.001), and chemotherapy (HR = 0.57, 95% CI: 0.43–0.77; p < 0.001) (Table 3). Table 3Multivariate Cox proportional hazards analysis of cancer-specific survival (CSS) before and after propensity score matching (PSM)VariableCSS before PSMCSS after PSMHR95% ClP-valueHR95% ClP-valueLNM+ vs LNM-1.6(1.13–2.25)0.0071.86(1.185–2.90)0.007Colon vs rectum0.78(0.59–1.03)NS0.61(0.374–1.00)NSAge (years) < 70 vs ≥ 701.19(0.91–1.56)NS1.04(0.569–1.63)NSFamily history of tumor, yes vs no0.94(0.72–1.23)NS0.86(0.546–1.34)NSPostoperative complications, yes vs no1.16(0.86–1.57)NS1.13(0.690–1.86)NSCEA (ng/ml), < 5 vs ≥ 51.40(1.04–1.90)0.0281.32(0.756–2.32)NSTumor size (cm), < 5 vs ≥ 50.97(0.75–1.27)NS0.91(0.587–1.42)NST classification, 1–2 vs 3–41.24(0.57–2.66)NS3.88(0.514–29.35)NSHistotype, AdenoCA vs Muc, Sig0.54(0.31–0.94)0.0290.24(0.074–0.79)0.018Liver metastasis, yes vs no1.62(1.06–2.46)0.0240.97(0.43–2.17)NSLung metastasis, yes vs no1.78(1.26–2.52)0.0011.30(0.655–2.59)NSDifferentiation, Well/Mod vs Poor/Muc/Sig0.43(0.29–0.64) < 0.0010.16(0.063–0.40) < 0.001Chemotherapy, yes vs no0.57(0.43–0.77) < 0.0010.53(0.327–0.87)0.013LNM lymph node metastasis, CEA carcinoembryonic antigen, Well well differentiated, Mod moderately differentiated, Poor poorly differentiated, AdenoCA adenocarcinoma, Muc mucinous carcinoma, Sig signet ring cell carcinoma, NS not significant Multivariate Cox proportional hazards analysis of cancer-specific survival (CSS) before and after propensity score matching (PSM) LNM lymph node metastasis, CEA carcinoembryonic antigen, Well well differentiated, Mod moderately differentiated, Poor poorly differentiated, AdenoCA adenocarcinoma, Muc mucinous carcinoma, Sig signet ring cell carcinoma, NS not significant Outcomes of LNM analysis after matching After 1:1 PSM (N0M1, 87; N+M1, 87), we compared clinical characteristics and pathologic variables of LNM subsets. Univariate analysis confirmed that the two groups were comparable (Table 1), and OS was still significantly better in the N0M1 (vs N+M1) group (43.7% vs 26.5%; p = 0.042) (Fig. 3B1). LNM (HR = 1.72, 95% CI: 1.121–2.64; p = 0.013) also proved to be independently predictive of OS, in addition to tumor histotype (HR = 0.31, 95% CI: 0.099–0.95; p = 0.04), tumor differentiation (HR = 0.19, 95% CI: 0.076–0.47; p < 0.001), and chemotherapy (HR = 0.49, 95% CI: 0.306–0.79; p = 0.004) (Table 2). Likewise, the N0M1 group surpassed the N+M1 group in CSS (48.8% vs 28.0%; p = 0.027) (Fig. 3B2), and LNM was identified as a significant independent predictor of CSS (HR = 1.86, 95% CI: 1.185–2.90; p = 0.007). Other factors predictive of CSS included tumor histotype (HR = 0.24, 95% CI: 0.074–0.79 p = 0.018), tumor differentiation (HR = 0.16, 95% CI: 0.063–0.40; p < 0.001), and chemotherapy (HR = 0.53, 95% CI: 0.327–0.87; p = 0.013). Both Cox regression (for time-to-event outcomes) and logistic regression (for 5-year survival status) confirmed that lymph node metastasis (LNM) was an independent adverse prognostic factor for stage IV CRC (Table 2 and 3). Discussion Above findings clearly demonstrate that in patients with stage IV CRC, hematogenous organ metastasis carries a significantly better prognosis than does regional LNM. Our analysis also confirmed regional LNM as an independent risk factor for long-term prognosis. This is not surprising, given the presumption by Dukes long ago that depth of tumor invasion (through colorectal wall) and LNM are strongly implicated in CRC prognosis [14]. LNM status has since become a critical factor in the treatment of CRC [15, 16]. The Japanese classification of CRC takes into account distant reaches of positive lymph nodes, using pericolic, intermediate, and apical regions to gauge their relative remoteness from primary tumors [5]. In our examination of patients with stage III CRC, it was apparent that the further positive nodes were situated from sites of primary lesions, the worse the prognosis seemed to be. LNM status thus proved highly prognostic of stage III CRC. For stage III CRC, the prognosis worsens with increasing distance of lymph node metastasis from the primary tumor. However, the biological behavior of stage IV CRC may not follow a sequential dissemination model. Our data showed that N0M1 patients had a better prognosis than N+M1 patients, which may be related to the distinct metastatic pathways (hematogenous vs lymphatic-hematogenous combined spread) and tumor biological characteristics. However, distant implants of CRC likely result from hematogenous spread, becoming more ominous when coupled with LNM (N+M1). Indeed, the prognosis worsens as lymph node positivity climbs, lending support to a link between outcomes of stage IV CRC and LNM. The natural process of CRC metastasis is generally one in which tumor first breaks through intestinal wall, then invades nearby lymph nodes, and finally arrives at distant organs. In our cohort with stage IV CRC, regional LNM was lacking for ~ 22.1%, with the progression of nodal spread (N0M1 → N1M1 → N2M1) met by a worsening of prognosis. Such events raise the question of whether LNM and hematogenous metastasis occur together or independently. In some of our patients, hematogenous metastases actually preceded nodal metastases. Common sites of metastasis from CRC are liver, lung, peritoneum, and systemic lymph nodes, each variably ascribed to hematogenous, lymphatic, or peritoneal routes of spread [17]. Because venous drainage of the intestine is through the portal system, hematogenous metastases to liver are the first to occur, followed by tumor dissemination to lungs, bone, brain, and other sites. Cancers of distal rectum may instead first metastasize directly to the lungs, traveling from the inferior rectal vein to the inferior vena cava (IVC). Prior to PSM, our patient groups showed no significant differences in tumor locations or in metastases to liver or lung, but the mechanisms for metastasis of stage IV CRC must be researched more extensively. Since 1977, the AJCC has assumed publication of the TNM classification system originally devised by Pierre Denoix [18], relegating stage IV CRC to a consistently poor prognosis throughout the years. Although little attention has been paid to its risk factors, recent advances in preoperative imaging, improved survival rates after hepatectomy, and the availability of new adjuvant chemotherapeutic agents [19] have now provided the impetus for refinement of outcome projections. During the present investigation, we found no significant difference between groups (N0M1 vs N+M1) with respect to postoperative adjuvant chemotherapy, whereas survival proved significantly better for members of the N0M1 (vs N+M1) group. This discovery underscores the potential curability of otherwise systemic metastatic disease, in the absence of LNM. Unfortunately, a consensus is lacking on the order of needed tumor resection (primary vs metastasis). In Europe and in the USA, “liver first” is the rule, removing metastatic and primary tumors sequentially. Primary lesions are generally resected first in Japan, owing to lower rates of postsurgical complications [20, 21]. A primary lesion left intact at time of hepatectomy may continue to seed the liver or lungs, risking recurrence or metastatic spread [16]. Despite observed preferences in treating stage IV CRC, the shared objective is still radical removal of both primary and secondary tumors, hoping to prolong patient survival [19, 22]. At present, stage IV CRC is defined by the status of organ metastasis, designating solitary organ involvement as IVA, dual involvement as IVB, and combined peritoneal spread as IVC. Alternatively, our data have implicated LNM as an independent prognostic factor for stage IV disease. As with stage III CRC, it is quite feasible that precise counts of positive nodes may more accurately reflect the true prognosis of stage IV CRC, helping to ensure proper clinical treatment. There are some study limitations that are worthy of mention. Our single-center, retrospective design and the relatively small population examined may have influenced these results. Various treatment methods for distant metastases, such as resection (in any order) or chemoradiotherapy alone, are inclined to impact patient prognosis as well. A broader patient study, mandating synchronous removal of distant tumors, is warranted to corroborate the findings herein. Conclusion In patients with stage IV colorectal cancer (M1), those with distant hematogenous metastasis without regional lymph node involvement (N0M1) have a better prognosis than those with concurrent regional lymph node metastasis (N+M1). This apparent prognostic divergence must be reflected in refined tumor classification subsets. Author contribution Liming Wang and Haoyue Ma provided original pictures and drafted the manuscript, which was reviewed by Bolun Song, Yinggang Chen, and Yasumitsu Hirano. All authors have approved the final submission. Funding This study was supported by the Shenzhen High-Level Hospital Construction Fund. Data Availability Not applicable Declarations Ethics approval and consent to participate All study subjects granted informed consent. Our investigatory protocol was approved by the Ethics Committee of Saitama Medical University International Medical Center. Consent for publication Each participant agreed to the use of relevant personal medical data for research and publication. Conflict of interest The authors declare no competing interests. Footnotes Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Haoyue Ma and Liming Wang are co-first authors with equal contributions to this paper. References Associated Data Data Availability Statement Not applicable