All patients were identified as having CRC and their samples were provided anonymously for clinical study.

A total of 9,514 patients with CRC who underwent radical resection in the Sixth Affiliated Hospital, Sun Yat-sen University from December 2013 to August 2018 were screened. Patients meeting the following criteria were enrolled: (1) The patient was diagnosed with CRC by histopathology and underwent primary radical resection of CRC. (2) The patient developed synchronous or metachronous CRLM. (3) The number of lymph node biopsies was greater than 12. (4) The patient had no blood diseases, infectious diseases, immune diseases, or other diseases caused by poor nutritional status. (5) The patient had no intestinal perforation and/or severe ileus. (6) The patient did not receive drugs that may affect markers of inflammation. (7) Death occurred only from tumor-related causes. (8) The patient had complete clinical data of laboratory examination, imaging examination and follow-up.

Clinical data, including patient age at diagnosis, gender, location of primary tumor, histological differentiation, T stage, N stage, CRLM features and treatment regimens, were extracted from electronic medical records. The laboratory data, including neutrophils, lymphocytes, LDH, carcinoembryonic antigen (CEA), carbohydrate antigen 19–9 (CA19-9) and carbohydrate antigen-125 (CA-125), were collected from peripheral blood within 7 days before the surgical resection of their primary tumor. The pathological stage of CRC was classified according to the American Joint Committee on Cancer (AJCC) tumor-node-metastasis (TNM) staging system, eighth edition. All patients were followed up until August 2020 or until death. The primary endpoint was overall survival (OS).

OS was defined as the time interval from the date of diagnosis to death from any cause or to the last known date of life. Synchronous CRLM was defined as metastasis that occurred at the time of diagnosis of CRC or within 6 months after radical resection of the primary CRC. Metachronous liver metastasis was defined as liver metastasis occurring more than 6 months after radical resection of CRC (30).

Routine blood tests were performed to detect neutrophils and lymphocytes. The NLR was defined as the absolute count of neutrophils divided by the absolute count of lymphocytes. Markers of inflammation include neutrophil count, lymphocyte count, NLR, and LDH. Both CEA and CA 19–9 are tumor markers.

Variables were described by categorical variables, and the clinical and pathological characteristics of patients were summarized by descriptive analysis. The optimal cut-off value of the NLR for the OS analysis was calculated using a receiver operating characteristic (ROC) curve. The chi-square test of categorical variables was used to compare the differences between groups. Logistic regression analysis was used to analyze the risk factors for the time frame of CRLM. The probability of OS was estimated by the Kaplan–Meier method in the univariate survival analysis, and the log-rank test was used for comparison. A Cox proportional hazard regression model was used to obtain the survival risk hazard ratio (HR) and 95% confidence interval (CI). Significant predictors in the univariate analysis were further estimated by the forward stepwise (likelihood ratio) selection of multivariate analysis so that only significant variables had HR and 95% CI values within the threshold. Finally, area under the ROC curve (AUC) was used to evaluate the performance of the model. The above statistical analyses were conducted using SPSS version 25.0, R version 3.6.1 and GraphPad Prism9, and differences with a p < 0.05 were defined as statistically significant.

The optimal cut-off value for the NLR based on the time of liver metastasis was calculated by ROC curve analysis and was found to be 2.071 (AUC: 0.591, Figure 1). According to laboratory reference values, the normal ranges of CEA, CA19-9 and LDH are 0–5 ng/mL, 0–37 U/mL and 120–250 U/L, respectively. According to the NLR-LDH score, we categorized patients into three groups: high score [NLR ≤ 2.071 and LDH ≤ upper limit normal (ULN)], intermediate score (NLR > 2.071 or LDH > ULN) and low score (NLR > 2.071 and LDH > ULN).

A total of 9,514 CRC patients were assessed and included, and 997 patients (10.5%) developed LM. Patients without key data on inflammatory markers, patients with infection or blood disease, and patients lost to follow-up were excluded. Finally, 300 patients with CRLM were included in the study (Figure 2). There were 197 males (65.7%) and 103 females (34.3%), with a median age of 60 years (23–91 years) at the time of diagnosis. The median survival time was 30 months. Overall, in the study cohort, according to the AJCC-TNM staging, there were 14 patients (4.6%) in stage I, 58 patients (19.0%) in stage II, 93 patients (30.5%) in stage III, and 124 patients (43.9%) in stage IV. During the study period, 152 patients died from cancer-related causes. Additional information, including clinical, pathological, and therapeutic characteristics, is summarized in Table 1.

The chi-square test showed significant differences in tumor location, R0 resection, NLR, CEA, CA19-9, LDH, NLR-LDH values, KRAS mutation with CRLM stage (p < 0.05) (Table 2).

The clinicopathological features grouped by NLR, LDH and NLR-LDH values are shown in Table 3. The values for NLR, LDH and NLR-LDH were significantly correlated with the LM time of CRC (p < 0.05). There were significantly more cases of synchronous LM in the high NLR group (p = 0.010), the high LDH group (p < 0.001) and the high NLR-LDH group (p < 0.001). NLR, LDH and NLR-LDH were also significantly correlated with age (p < 0.050).

Table 4 shows the univariate analysis of synchronous CRLM. In the analysis, a high NLR indicated a high likelihood of synchronous CRLM (HR: 1.878, 95% CI: 1.159–3.049, p = 0.011), Patients with elevated LDH had a much higher incidence of synchronous CRLM (HR: 11.875, 95% CI: 2.803–50.306, p = 0.001). In addition, the NLR-LDH value showed significant prognostic effects on synchronous CRLM, especially high NLR-LDH score (synchronous CRLM, p = 0.002, HR = 24.868, 95% CI: 3.269–186.417), although the intermediate group failed to have significant prognostic value. Tumors located in the rectum (HR = 2.362, 95% CI: 1.449–3.851, p = 0.001) and with CEA > 5 (HR = 2.583, 95% CI: 1.581–4.219, p < 0.001), CA19-9 > 37 (HR = 3.587, 95% CI: 1.958–6.573, p < 0.001) and mutant KRAS (HR: 1.991, 95% CI: 1.113–3.564, p = 0.020) were more likely to have synchronous LM than tumors with other features.

In terms of OS (Table 5), patients with a high NLR showed significantly worse OS (HR: 1.583, 95% CI: 1.136–2.205, p = 0.007) than those with a low NLR, and the same was true of LDH (HR: 2.932, 95 CI: 1.970–4.365, p < 0.001). In addition, the NLR-LDH value showed significant prognostic effects on OS, especially for patients with high NLR-LDH scores (high, HR = 3.672, 95% CI: 2.262–5.960, p < 0.05). Similar trend was also observed for patients with low or intermediate NLR-LDH scores (intermediate, HR = 1.507, 95% CI: 1.053–2.157, p < 0.05), Moreover, our results indicated that age (p = 0.005), histology (p = 0.037), liver metastases (p = 0.003), CRLM number (p = 0.027), R0 (p = 0.024), preoperative therapy (p = 0.019), CEA (p = 0.006),CA19-9 (p < 0.001), T classification (<0.001), N classification (<0.001), and KRAS status (p = 0.048) were statistically significant.

The results of Kaplan–Meier survival curves are shown in Figure 3. The combination of a high NLR and elevated LDH and NLR-LDH levels suggested a poor prognosis and a significantly shorter median survival time.

In terms of synchronous CRLM, KRAS status and CEA, LDH and NLR-LDH values were found to be statistically significant, with independent predictive value (Tables 6, 7).Our results indicated that T stage (p < 0.001), N stage (p < 0.001) and poor histology (p < 0.05) were independently associated with poor OS (forest plots, Figure 4). There was a significant difference in OS between the elevated and normal LDH groups (p < 0.05), although the limited number of patients with NLR > 2.071 affected the validity of the results. A high NLR-LDH was an independent predictor of OS (HR = 2.198, 95% Cl: 1.290–3.746, p = 0.004). T and N stage were significant influencing factors for survival and CRLM in clinical practice, so even though no statistical significance was found in the single-factor analysis, they were still included in the multi-factor analysis (forest plots, Figure 4).

The NLR-LDH score has better predictive ability than the NLR or LDH score. The ROC curve analysis showed that the predictive value of the NLR-LDH score for synchronous CRLM (p = 0.010) and OS (p = 0.001) was superior to that of the NLR or LDH score alone (Figure 5). In the NLR-LDH scoring model, the AUC of synchronous CRLM was predicted to be 0.623 ± 0.033, while the AUC of OS was predicted to be 0.614 ± 0.032 (Figure 6).

The liver is the most common metastatic site of CRC. CRLM occurs in 15–25% of patients with CRC at the first diagnosis, while CRLM occurs in another 50% of patients during the progression of CRC (33). The treatment and prognosis of CRC with CRLM are significantly different from that of CRC without CRLM. Some CRCs already have minor CRLM that cannot be detected by imaging. Therefore, it is very important to identify whether CRC has CRLM and conduct intervention in the early stage. In our study, we proposed for the first time the use of NLR, LDH, and NLR-LDH indicators to predict the timing of CRLM and conducted univariate and multivariate analyses of risk factors for synchronous CRLM. The results confirmed that increases in NLR, LDH, and NLR-LDH indexes were all risk factors for the concurrent CRLM, and the LDH and NLR-LDH indexes were independent risk factors.

The NLR in peripheral blood has been widely studied as an important indicator of systemic inflammation in recent years (14, 34). Its elevation indicates a relative increase in neutrophils and a decrease in lymphocytes, indicating a high degree of systemic inflammation. The persistence of the systemic inflammatory response can regulate the tumor microenvironment and promote tumor progression and metastasis. As shown in our study, an elevated NLR in CRC patients indicates the presence of a systemic inflammatory response, and a higher NLR (NLR > 2.071) is associated with a higher risk of concurrent CRLM.

According to previous studies, elevated LDH subtypes are more common in malignant tumors than in normal cells (35). In malignant tumors, tumor cells are more likely to obtain energy through glycolysis in the presence of oxygen, and LDH is a key enzyme in glycolysis. The glycolytic metabolite lactic acid can regulate cellular immune metabolism in the tumor microenvironment, promote immunosuppression in the tumor region, and facilitate tumor progression and metastasis (36). Our study confirmed that LDH is an independent risk factor of synchronous CRLM and an important indicator for the diagnosis of CRLM.

We found that the preoperative peripheral blood NLR-LDH score had a higher AUC for the diagnosis of concurrent CRLM than either of the independent diagnostic indexes (NLR + LDH_AUC = 0.623, NLR_AUC = 0.591, LDH_AUC = 0.565). The reason may be that this value can reflect the inflammatory response and immune function of patients with CRC. The combination of NLR and LDH can make up for the limitations of a single-indicator diagnosis and increase the diagnostic accuracy to some extent. In clinical practice, rational use of the preoperative peripheral blood NLR-LDH index combined with traditional imaging examination can more accurately predict whether patients with CRC have concurrent CRLM, providing justification for necessary further treatment.

Our study also found that colon cancer is more likely to develop concurrent CRLM than rectal cancer, possibly because colon cancer tumor cells are more likely to return to the portal vein through the inferior mesenteric vein, thus promoting the occurrence of the concurrent CRLM. KRAS mutation in CRC indicates a poor prognosis (37). In our study, it was found that CRC patients with KRAS mutations are more likely to have concurrent CRLM than patients without KRAS mutations and that KRAS mutation status is as an independent risk factor. Moreover, increased preoperative CEA and CA19-9 are also risk factors for concurrent CRLM, which is consistent with the findings of Zheng et al. (38).

In recent years, NLR and LDH have been identified to be associated with the survival and prognosis of patients with a variety of malignant tumors, including lung cancer (39), esophageal cancer (40), melanoma (41), and uterine sarcoma (42). As an indicator reflective of neutrophil and lymphocyte counts, the NLR may reflect systemic inflammatory responses (43). Neutrophils can be categorized as N1 and N2 phenotype by TGF-beta signaling in the tumor microenvironment (44). N2 neutrophils can produce vascular endothelial growth factor (VEGF) and metalloproteinase, promoting tumor angiogenesis, invasion and metastasis (45), while lymphocytes play an important role in antitumor immunity (46). Consistent with our findings, an elevated NLR is a risk factor influencing prognosis in patients with CRLM. As shown in the survival curve (Figure 3), the OS of patients with CRLM and an elevated NLR was lower than that of those with a decreased NLR (median OS: 35 vs. 49 months).

Increased LDH promotes tumor progression by regulating tumor metabolism and the microenvironment and serves as an indicator of poor prognosis in cancer patients (35, 47). Our study also found that LDH is an independent risk factor for the prognosis of CRLM patients. As shown in the survival curve, the OS of CRLM patients and elevated LDH was lower than that of those with decreased LDH (median OS: 21 vs. 49 months).

More importantly, when the NLR and LDH indexes were combined to evaluate the prognosis of CRLM, patients in the high NLR-LDH group had significantly lower OS than those in either the intermediate group or the low group (median OS: 21 vs. 37 vs. 42 months), and this index was an independent risk factor. A simultaneous increase in NLR and LDH means that the body is in a state of systemic inflammatory response and immunosuppression, which makes the tumor microenvironment more suitable the growth and metastasis of tumor cells, leading to a poor prognosis. The preoperative detection of both NLR and LDH provides a great reference for the assessment of the prognosis of CRLM patients.

In our study, independent risk factors for CRLM included age, pathological characteristics, T and N stage, and KRAS mutation status. As patient age over 60 years old, the OS decreased. When the pathological type of CRC was unfavorable and the stage of the primary tumor was late, the tumor was more malignant and the OS was shorter.