Abstract
Background
Colorectal cancer (CRC) remains a leading cause of malignancy-related mortality, with prognosis heavily contingent upon early detection. Notably, a significant proportion of CRC patients present with normal preoperative Carcinoembryonic Antigen (CEA) levels, creating a diagnostic blind spot where malignant lesions may be indistinguishable from benign colorectal disease (BCD). This study aimed to develop and validate multidimensional nomograms to accurately diagnose and prognostically stratify patients with CEA-negative CRC.
Methods
We retrospectively analyzed data from 553 CEA-negative CRC patients and 200 BCD patients. Optimal biomarker thresholds were determined via ROC analysis. Diagnostic and prognostic predictors were identified using multivariable logistic and Cox regressions. Model performance was evaluated using C-indices, calibration plots, Decision Curve Analysis (DCA), and Clinical Impact Curves (CIC).
Results
Of 895 screened CRC patients, 61.8% were CEA-negative. An eight-variable diagnostic model (age, sex, CA50, CA125, CA19-9, WBC, PLR, FAR) demonstrated strong discriminative power, yielding C-indices of 0.886 (training) and 0.895 (validation). For overall survival, a prognostic signature incorporating age, tumor location, Lymph Node Ratio (LNR), Monocyte-to-Lymphocyte ratio (MLR), and Tumor-Stroma Ratio (TSP) achieved C-indices of 0.749 (training) and 0.829 (validation), showing favorable prognostic stratification compared to the AJCC 8th TNM staging system. DCA and CIC confirmed superior clinical net benefit for both models.
Conclusion
Integrating systemic inflammatory indices and stromal features provides a potentially useful adjunctive tool for cancer detection and risk assessment. These user-friendly nomograms may assist in personalized prognostication, potentially refining clinical decision-making and patient stratification for therapeutic trials. However, independent external validation is required before clinical implementation due to the single-center design and internally derived cut-offs.
Supplementary Information
The online version contains supplementary material available at 10.1186/s12957-026-04452-x.
Keywords: Colorectal cancer, Diagnostic prediction, Prognostic assessment, Tumor markers, Nomogram
Introduction
Colorectal cancer (CRC) remains a formidable global health challenge, ranking as the third most prevalent malignancy and the second leading cause of cancer-related mortality worldwide [1]. Current projections paint a concerning picture: by 2040, the global burden of CRC is expected to surge by nearly 60% compared to 2020 levels [2]. While therapeutic advancements have improved survival rates for early-stage disease, the prognosis for patients with advanced or metastatic presentations remains dismal [3]. Consequently, the cornerstone of improving long-term outcomes lies in the precise identification of high-risk individuals to facilitate timely, personalized interventions. However, a critical diagnostic gray zone persists for a substantial subset of patients who do not exhibit classical biochemical signals, leaving clinicians without a reliable compass for risk stratification and early detection.
The Carcinoembryonic Antigen (CEA) has long served as the standard-bearer for CRC surveillance. Historically, elevated CEA levels are strongly correlated with tumor burden, recurrence, and metastasis [4–8]. Yet, a significant clinical paradox exists: over 50% of CRC patients present with normal preoperative CEA levels [4, 9]. These marker-negative patients are not necessarily at lower risk; rather, they often harbor aggressive malignancies that bypass traditional secretory pathways, potentially leading to a false sense of security and delayed treatment [10, 11]. Furthermore, the low sensitivity of CEA renders it ineffective for general screening [12]. Therefore, relying exclusively on CEA monitoring creates a dangerous blind spot in the management of this large patient population.
To address this limitation, attention has shifted from tumor-derived markers to the host-tumor interface, specifically systemic inflammation and the tumor microenvironment (TME) [13]. Systemic inflammatory markers, including the Neutrophil-to-Lymphocyte ratio (NLR), Monocyte-to-Lymphocyte ratio (MLR), Platelet-to-Lymphocyte ratio (PLR), Fibrinogen-to-Albumin ratio (FAR), and C-reactive protein-to-Albumin ratio (CAR), have emerged as critical determinants of malignant progression and patient survival [14–17]. Furthermore, recent evidence highlights their predictive value in therapeutic settings; specifically, pre-treatment levels of CRP, NLR, PLR, and MLR have been strongly correlated with both progression-free survival (PFS) and overall survival (OS) in patients undergoing immune checkpoint inhibitor therapy [18–20]. However, most existing studies conflate CEA-positive and CEA-negative populations, a methodological limitation that risks diluting the specific predictive value of inflammatory markers for the CEA-negative subgroup, where they may serve as the primary indicators of disease progression [21].
Parallel to inflammatory indices, alternative serum tumor markers such as CA19-9, CA125, CA50, and CA724 are being explored as supplementary tools [8, 22–28]. While individual markers often lack sufficient sensitivity [24], panel-based approaches combining multiple antigens have shown promise in enhancing diagnostic accuracy [10, 25, 26, 29, 30]. However, a significant gap remains in the current literature: few studies have successfully integrated these serological markers with histological features of the tumor microenvironment, specifically the Tumor-Stroma Ratio (TSP), to create a comprehensive predictive model. TSP is a powerful indicator of desmoplasia and tumor invasiveness and has been validated as a robust prognosticator independent of TNM staging in our previous study [31]. The integration of serological biomarkers and histopathological features offers a promising strategy to mitigate the diagnostic challenges in CEA-negative patients, yet the clinical value of this multidimensional approach lacks sufficient validation.
To address this unmet clinical need, our study retrospectively evaluated a large cohort of CEA-negative CRC patients. We aimed to dissect the complex interplay between serum tumor markers, systemic inflammatory indicators, and clinicopathological characteristics. By integrating these diverse variables, we developed and validated novel nomogram models. These tools are designed to provide clinicians with a precise, user-friendly method for differentiating benign disease from CEA-negative CRC and for accurately stratifying overall survival, thereby potentially supporting more personalized clinical decision-making. However, external validation is required before these tools can be implemented clinically.
Materials and methods
Study design and patient population
This retrospective analysis utilized clinical data derived from a comprehensive, prospectively maintained institutional database at Shanghai General Hospital. The study population was composed of two distinct cohorts: a primary cohort of patients with confirmed CRC and a control group comprising patients with benign colorectal disease (BCD). Eligible participants were consecutively identified from the hospital’s electronic medical records system, ensuring a representative sample for both diagnostic and prognostic evaluation. The study protocol adhered to the ethical guidelines of the Declaration of Helsinki and was approved by the Clinical Research Ethics Committee of Shanghai General Hospital.
Inclusion and exclusion criteria
All CRC patients underwent radical surgical resection. The inclusion criteria were as follows: (i) No history of neoadjuvant chemotherapy or radiotherapy within three months prior to surgery; (ii) Histologically confirmed colorectal adenocarcinoma; and (iii) Availability of complete clinicopathological, laboratory, and follow-up data. Patients were excluded if they met any of the following conditions: lack of survival follow-up data (n = 976), missing key pathological indicators (n = 20), or absence of preoperative CEA measurements (n = 164). A detailed flowchart of the patient selection process is presented in Fig. 1.
Fig. 1.
Flowchart depicting the patient selection and enrollment process for the colorectal cancer (CRC) and benign colorectal disease (BCD) cohorts
The final study population comprised 895 CRC patients, of whom 342 exhibited elevated preoperative CEA levels and 553 presented with normal levels (defined as CEA-negative). Tumor staging was determined in accordance with the 8th edition of the American Joint Committee on Cancer (AJCC) TNM staging system. Pathological diagnoses were independently verified by two senior pathologists.
The BCD patients were consecutively enrolled during the same period as the CRC cohort to minimize temporal bias and included only those with confirmed non-malignant pathology (colorectal polyps, n = 121; benign tumors, n = 79). Patients with acute or active chronic inflammatory conditions were explicitly excluded to avoid confounding inflammatory indices. Serum tumor markers and inflammatory indices were collected from these patients to serve as the non-malignant control arm for the diagnostic nomogram.
Data collection and variable definitions
We extracted a comprehensive set of variables, including demographic characteristics (age, sex), pathological features (TNM stage, vascular/perineural invasion, tumor location/size), and the TSP. To eliminate observer bias and maximize reproducibility, the TSP was quantitatively evaluated using our previously validated automated MATLAB-based algorithm [31], which objectively calculates stromal percentage from whole-slide images with higher accuracy than manual estimation. A standard cut-off of 50% was applied to categorize patients into high (> 50%) versus low (≤ 50%) stromal groups. The Lymph Node Ratio (LNR) was also calculated, with a cut-off value of 0.
Hematological parameters included white blood cell count (WBC), neutrophils (NE), lymphocytes (LY), monocytes (MO), platelets (PLT), fibrinogen (FIB), C-reactive protein (CRP), albumin (Alb), and total bilirubin (TBil). Additionally, a panel of serum tumor markers (CEA, CA19-9, CA50, CA125, CA724, CA242) and molecular markers (CDX2, MLH1, MSH2, KI-67, ERCC1) were assessed.
Determination of optimal cut-off values
Patients with preoperative serum CEA levels ≤ 5.0 ng/mL were classified as the CEA-negative cohort. To maximize the diagnostic sensitivity of the remaining biomarkers specifically for the CEA-negative population, we recalculated optimal cut-off values using Receiver Operating Characteristic (ROC) curve analysis by determining the maximum Youden index (Sensitivity+Specificity-1). The determining thresholds were: CA19-9 (12.51 U/mL), CEA (1.77 ng/mL), CA125 (9.43 U/mL), CA72-4 (1.63 U/mL), CA50 (7.44 U/mL), and CA242 (13.64 U/mL). Similarly, optimal thresholds for inflammatory indices were established as follows: NLR (1.432), PLR (180.156), MLR (0.228), FAR (0.075), and CAR (0.043). For the prognostic nomogram, to facilitate the practical application of the scoring system, all continuous variables (including tumor markers and inflammatory indices) were dichotomized according to their optimal cut-off values prior to survival analysis.
Follow-up protocol and endpoint definition
Postoperative surveillance was conducted quarterly for the first two years and annually from the third to the fifth year. Follow-up assessments included serum tumor marker profiling, blood chemistry, and imaging (chest/abdominal CT or pelvic MRI), with annual colonoscopy. OS was defined as the interval from the date of surgery to the date of death from any cause. For patients who remained alive at the time of analysis, data were censored at the date of their last clinical contact.
Model construction and validation
To facilitate robust model development and internal validation, the 553 CEA-negative CRC patients were randomized into a training cohort (n = 369) and a validation cohort (n = 184) using a stratified sampling method (R function “createFolds”). This approach ensured balanced outcome distribution across cohorts. For the diagnostic nomogram (differentiating BCD from CEA-negative CRC), optimal cut-offs derived from ROC analysis were applied. Independent predictors were identified via univariate and multivariable binary logistic regression analyses. For the prognostic nomogram (predicting OS), 27 potential clinicopathological variables were screened using Kaplan-Meier survival curves and log-rank tests. Variables demonstrating statistical significance were subsequently entered into univariate and multivariable Cox proportional hazards regression models to identify independent prognostic factors.
Statistical analysis and model evaluation
Categorical variables were compared using the Chi-square (χ2) test. The performance of the nomograms was quantified using the concordance index (C-index) and visualized via calibration plots comparing predicted versus observed probabilities. The clinical utility and net benefit of the models were assessed using Decision Curve Analysis (DCA) and Clinical Impact Curves (CIC). For the prognostic model, DCA was specifically employed to benchmark the nomogram against the AJCC 8th edition TNM staging system.
To ensure data integrity and avoid imputation bias, missing values in this retrospective cohort were handled according to the proportion of missingness. Specifically, variables with a substantial amount of missing data (e.g., CA242) were excluded from multivariable modeling to prevent significant attrition of the study population. For the remaining incorporated variables with negligible missing values, complete case analysis (listwise deletion) was employed during all multivariable logistic and Cox regression analyses.
All statistical analyses were performed using SPSS version 26.0 (IBM, Armonk, NY, USA) and R software version 4.4.0 (The R Foundation for Statistical Computing). A complete list of R packages utilized is provided in Supplementary Table S1. All statistical tests were two-sided, and a p-value < 0.050 was considered statistically significant.
Results
Demographic and clinicopathological characteristics
A total of 895 patients with CRC undergoing radical resection were included in the study. Based on preoperative serum CEA levels, the cohort was stratified into the CEA-negative group (n = 553, 61.8%) and the CEA-elevated group (n = 342, 38.2%). To ensure statistical robustness, Stage III and IV patients were grouped together, which included 199 (36.0%) Stage III patients and a minimal 6 (1.1%) Stage IV patients in the CEA-negative cohort (n = 553). The detailed baseline characteristics are summarized in Table 1. Comparative analysis revealed significant disparities between the two groups, specifically, variables including gender, TNM stage, perineural/vascular invasion, tumor size, TSP, and expression levels of CA19-9, CA50, CA724, CA242, and ERCC1 differed statistically (P < 0.050).
Table 1.
Demographic and clinical characteristics of CEA-negative and CEA-positive CRC patients
| Characteristics | Total | CEA negative | CEA positive | P value | |
|---|---|---|---|---|---|
| 895 | 553 (61.8%) | 342 (38.2%) | |||
| Age | ≤ 60 years | 279 | 185 (33.5%) | 94 (27.5%) | 0.061 |
| > 60 years | 616 | 368 (66.5%) | 248 (72.5%) | ||
| Gender | Male | 511 | 290 (52.4%) | 221 (64.6%) | 0.000 |
| Female | 384 | 263 (47.6%) | 121 (35.4%) | ||
| TNM stage | I–II | 500 | 348 (62.9%) | 152 (44.4%) | 0.000 |
| III | 388 | 199 (36.0%) | 189 (55.3%) | ||
| IV | 7 | 6 (1.1%) | 1 (0.3%) | ||
| Histological grade | Low | 139 | 82 (14.8%) | 57 (16.7%) | 0.461 |
| Middle-High | 756 | 471 (85.2%) | 285 (83.3%) | ||
| Nerve invasion | yes | 326 | 123 (22.2%) | 203 (59.4%) | 0.000 |
| no | 569 | 430 (77.8%) | 139 (40.6%) | ||
| Vessel invasion | yes | 358 | 165 (29.8%) | 193 (56.4%) | 0.000 |
| no | 537 | 388 (70.2%) | 149 (43.6%) | ||
| Location | Right colon | 278 | 175 (31.6%) | 103 (30.1%) | 0.622 |
| Left colon | 235 | 139 (25.1%) | 96 (28.1%) | ||
| Rectum | 382 | 239 (43.2%) | 143 (41.8%) | ||
| Size | > 5 cm | 246 | 136 (24.6%) | 110 (32.2%) | 0.014 |
| ≤ 5 cm | 649 | 417 (75.4%) | 232 (67.8%) | ||
| TSP | ≤ 50% | 627 | 411 (74.3%) | 216 (63.2%) | 0.011 |
| > 50% | 203 | 113 (20.4%) | 90 (26.3%) | ||
| NA | 65 | 29 (5.2%) | 36 (10.5%) | ||
| CA19-9 | Negative(≤ 37 ) | 773 | 523 (94.6%) | 250 (73.1%) | 0.000 |
| Positive(> 37 ) | 122 | 30 (5.4%) | 92 (26.9%) | ||
| CA50 | Negative(≤ 30 ) | 560 | 370 (66.9%) | 190 (55.6%) | 0.000 |
| Positive(> 30 ) | 51 | 7 (1.3%) | 44 (12.9%) | ||
| NA | 284 | 176 (31.8%) | 108 (31.6%) | ||
| CA125 | Negative(≤ 30.2 ) | 566 | 355 (64.2%) | 211 (61.7%) | 0.161 |
| Positive(> 30.2) | 84 | 46 (8.3%) | 38 (11.1%) | ||
| NA | 245 | 152 (27.5%) | 93 (27.2%) | ||
| CA72-4 | Negative(≤ 6.9) | 724 | 465 (84.1%) | 259 (75.7%) | 0.002 |
| Positive(> 6.9) | 149 | 76 (13.7%) | 73 (21.3%) | ||
| NA | 22 | 12 (2.2%) | 10 (2.9%) | ||
| CA242 | Negative(≤ 25) | 440 | 320 (57.9%) | 120 (35.1%) | 0.000 |
| Positive(> 25) | 137 | 32 (5.8%) | 105 (30.7%) | ||
| NA | 318 | 201 (36.3%) | 117 (34.2%) | ||
| CDX2 | Positive( + ) | 795 | 487 (88.1%) | 308 (90.1%) | 0.186 |
| Negative(- ) | 42 | 30 (5.4%) | 12 (3.5%) | ||
| NA | 58 | 36 (6.5%) | 22 (6.4%) | ||
| MLH1 | Positive( + ) | 712 | 445 (80.5%) | 267 (78.1%) | 0.941 |
| Negative(- ) | 95 | 59 (10.7%) | 36 (10.5%) | ||
| NA | 88 | 49 (8.9%) | 39 (11.4%) | ||
| MSH2 | Positive( + ) | 744 | 462 (83.5%) | 282 (82.5%) | 0.251 |
| Negative(- ) | 69 | 38 (6.9%) | 31 (9.1%) | ||
| NA | 82 | 53 (9.6%) | 29 (8.5%) | ||
| KI-67 | Negative(≤ 50%) | 174 | 99 (17.9%) | 75 (21.9%) | 0.149 |
| Positive(> 50%) | 615 | 387 (70.0%) | 228 (66.7%) | ||
| NA | 106 | 67 (12.1%) | 39 (11.4%) | ||
| ERCC1 | Positive( + ) | 605 | 376 (68.0%) | 229 (67.0%) | 0.044 |
| Negative(- ) | 105 | 76 (13.7%) | 29 (8.5%) | ||
| NA | 185 | 101 (18.3%) | 84 (24.6%) |
Bold indicates statistically significant (P<0.05)
Diagnostic performance of individual biomarkers
To identify potential candidates for the diagnostic model, we analyzed serum tumor markers and inflammatory indices in 553 CEA-negative CRC patients compared to 200 patients with BCD. Utilizing optimized cut-off values, we found that CEA retained diagnostic utility even within the normal range, achieving a sensitivity of 67.6% (Figure S1, Table S2). Among other tumor markers, CA50 and CA125 emerged as the most sensitive indicators (60.2% and 58.6%, respectively), exhibiting superior diagnostic performance with Area Under the Curve (AUC) values of 0.734 and 0.771, respectively (Figure S2A, Table S2). Notably, the combination of CA50 and CA125 significantly enhanced diagnostic efficiency, yielding an AUC of 0.840 (Figure S2B, Table S2).
Regarding inflammatory indices, the FAR and CAR demonstrated the highest individual discriminative ability, with AUCs of 0.739 and 0.700, respectively (Figure S2C, Table S3). However, unlike tumor markers, combining inflammatory indicators (e.g., MLR + CAR) did not yield a significant improvement in diagnostic accuracy (Figure S2D, Table S3). Consequently, serum tumor markers generally outperformed inflammatory indices in diagnostic precision for this specific population.
Development and validation of diagnostic nomograms
To systematically evaluate and optimize diagnostic accuracy, we constructed three distinct nomograms to differentiate CEA-negative CRC from BCD: Model 1 (based solely on tumor markers), Model 2 (based solely on inflammatory indices), and Model 3 (an integrated approach). Initially, multivariable logistic regression analysis was performed for the single-category models. For Model 1, eight tumor markers (including CA19-9, CA50, and CA125) were identified as independent risk factors (Table S4), yielding a robust C-index of 0.887 in the validation cohort (Table S7). In parallel, Model 2 incorporated inflammatory parameters, identifying indices such as NLR, PLR, MLR, and FAR as independent predictors (Table S5); however, this model exhibited more moderate predictive capability, with a validation C-index of 0.733 (Table S7, Figure S3).
To further enhance diagnostic precision, we developed the integrated Model 3 by combining the most significant inflammatory and tumor markers. Multivariable analysis confirmed Age, Gender, CA50, CA125, CEA, PLR, FAR, and WBC as independent determinants of disease status (Table S6), with CA19-9 retained for its clinical relevance (Fig. 2A). As anticipated, this comprehensive model demonstrated superior performance compared to the single-category models, achieving the highest C-indices of 0.886 in the training set and 0.895 in the validation set (Table S7). The nomogram attained a validation sensitivity of 84.9% (95% CI: 77.2–90.8), specificity of 73.4% (95% CI: 60.9–83.7), and accuracy of 80.9% (95% CI: 74.4–86.3), with PPV of 85.6% and NPV of 72.3% (Supplementary Table S8), confirming its robust diagnostic performance for CEA-negative CRC. Calibration plots revealed excellent agreement between predicted probabilities and actual observations (Fig. 2B). Furthermore, DCA and CIC confirmed that Model 3 offered the greatest net clinical benefit across a wide range of threshold probabilities, outperforming both Model 1 and Model 2 (Fig. 2C-D).
Fig. 2.
Development and evaluation of the combined diagnostic nomogram (Model 3). A The diagnostic nomogram integrating tumor markers and inflammatory indices to predict the probability of malignancy in CEA-negative patients. B Calibration plots assessing the agreement between predicted and observed probabilities of CRC occurrence. C Decision Curve Analysis (DCA) comparing the net clinical benefit of the three different models (Model 1, 2, and 3). D Clinical Impact Curve (CIC) for the integrated Model 3
Identification of prognostic factors for overall survival
Among 553 CEA-negative CRC patients with a median follow-up of 57.7 months (range 3.6–92.9), 76 deaths occurred. To build a prognostic model, 27 clinicopathological variables were screened using Kaplan-Meier and log-rank tests, and significant ones entered a Cox proportional hazards model (Table 2 and S9). Multivariable analysis identified five independent determinants of OS: Age (> 60 years; HR = 1.941, P = 0.039), Lymph Node Ratio (LNR) (Positive; HR = 3.312, P < 0.001), MLR (> 0.228; HR = 2.373, P = 0.002), TSP (> 50%; HR = 1.827, P = 0.024), and Tumor Location (Left vs. Right Colon; HR = 1.988, P = 0.008). We further explored CEA’s prognostic value within the normal range. As a continuous variable, it was significantly associated with OS (HR = 1.270, P = 0.012), but significance was lost when dichotomized (HR = 1.409, P = 0.188), justifying its exclusion from the final nomogram (Table 2).
Table 2.
Univariable and multivariable analyses of clinicopathologic variables in relation to OS in CEA-negative CRC patients
| Variable | Univariate analysis | Multivariate analysis | ||||
|---|---|---|---|---|---|---|
| HR | 95% CI | P value | HR | 95% CI | P value | |
| Gender | 0.899 | 0.572–1.413 | 0.644 | - | - | - |
| Age | 1.920 | 1.106–3.333 | 0.020 | 1.941 | 1.035–3.642 | 0.039 |
| Histological grade | 0.438 | 0.260–0.737 | 0.002 | 0.647 | 0.356–1.175 | 0.152 |
| TSP | 1.820 | 1.088–3.043 | 0.022 | 1.827 | 1.084–3.080 | 0.024 |
| Size | 1.603 | 0.993–2.588 | 0.053 | 1.438 | 0.807–2.562 | 0.218 |
| Location | 1.897 | 1.207–2.982 | 0.006 | 1.988 | 1.201–3.292 | 0.008 |
| Nerve invasion | 2.287 | 1.434–3.648 | 0.001 | 1.421 | 0.798–2.531 | 0.233 |
| Vessel invasion | 2.246 | 1.430–3.529 | 0.000 | 1.416 | 0.795–2.522 | 0.238 |
| LNR | 3.087 | 1.942–4.906 | 0.000 | 3.312 | 1.983–5.533 | 0.000 |
| CDX2 | 0.690 | 0.298–1.599 | 0.387 | - | - | - |
| MLH1 | 1.076 | 0.513–2.259 | 0.846 | - | - | - |
| MSH2 | 0.587 | 0.290–1.188 | 0.139 | - | - | - |
| KI-67* | 0.481 | 0.283–0.817 | 0.007 | - | - | - |
| ERCC1 | 0.887 | 0.470–1.676 | 0.712 | - | - | - |
| CEA | 1.409 | 0.845–2.348 | 0.188 | - | - | - |
| CA199 | 1.679 | 1.071–2.634 | 0.024 | 1.341 | 0.810–2.218 | 0.254 |
| CA50 | 1.169 | 0.632–2.160 | 0.619 | - | - | - |
| CA125 | 1.486 | 0.802–2.753 | 0.208 | - | - | - |
| CA724 | 1.045 | 0.656–1.667 | 0.852 | - | - | - |
| CA242* | 2.210 | 1.170–4.176 | 0.015 | - | - | - |
| NLR | 1.521 | 0.758–3.052 | 0.238 | - | - | - |
| PLR | 1.395 | 0.875–2.223 | 0.162 | - | - | - |
| MLR | 2.596 | 1.581–4.263 | 0.000 | 2.373 | 1.370–4.110 | 0.002 |
| FAR | 1.638 | 1.023–2.621 | 0.040 | 1.064 | 0.620–1.823 | 0.822 |
| CAR* | 1.561 | 0.836–2.914 | 0.162 | - | - | - |
Note:*Due to partial data loss, it was not included in the multivariate analysis
Bold indicates statistically significant (𝑃<0.05)
Construction and clinical value of the prognostic nomogram
Based on these five independent predictors, a prognostic nomogram was established (Fig. 3A). The model displayed robust discrimination, with C-indices of 0.749 (training) and 0.829 (validation). The corrected C-indices were consistent at 0.714 and 0.799, respectively (Table S10). Calibration curves demonstrated excellent agreement between the nomogram predicted survival and actual outcomes in both the training and total cohorts (Fig. 3B-D).
Fig. 3.
Development and calibration of the prognostic nomogram. ABCDA Prognostic nomogram for predicting 1-, 2-, and 3-year overall survival, (OS) in CEA-negative CRC patients based on Age, MLR, LNR, TSP, and Tumor Location. The sum of points for each variable corresponds to a specific survival probability on the total points axis., B-D Calibration curves assessing the agreement between nomogram-predicted and observed 1-year OS probabilities in the training set, (B), internal validation set, (C), and total cohort, (D). The dashed diagonal line represents the ideal prediction, while the solid line represents the actual performance of the nomogram; closer alignment indicates better calibration
Stratification capability and comparison with TNM staging
To rigorously benchmark the clinical utility of our prognostic nomogram against the standard AJCC 8th edition TNM staging system, we first evaluated its risk stratification capability. Based on the calculated nomogram scores, patients were stratified into distinct low- and high-risk subgroups. Kaplan-Meier survival analysis demonstrated robust discrimination, with the low-risk group exhibiting significantly superior overall survival compared to the high-risk group across the training, validation, and total cohorts (P < 0.010; Fig. 4A-C). Subsequently, DCA was employed to compare the net clinical benefit of the two systems. The results indicated that our nomogram achieved a higher net benefit than the TNM staging system across the majority of threshold probabilities (Fig. 4D-F). This finding suggests that by integrating biological indicators such as TSP and systemic inflammatory indices, our multi-dimensional model offers a highly favorable clinical net benefit for CEA-negative patients compared to traditional anatomical staging alone.
Fig. 4.
Risk stratification and clinical utility of the prognostic nomogram. AB-D(B(C(D(A-C) Kaplan-Meier survival curves comparing overall survival between high- and low-risk groups in the training set (A), validation set (B), and total cohort (C). (D-F) Decision Curve Analysis (DCA) comparing the prognostic net benefit of the nomogram versus the AJCC 8th TNM staging system in the training set (D), validation set (E), and total cohort (F).
Discussion
Serum CEA levels have long been the cornerstone for monitoring CRC [5, 32, 33]. However, a significant diagnostic blind spot exists, as 60.0% to 79.6% of CRC patients present with normal preoperative CEA levels [30, 34]. Consistent with this, 61.8% of the 895 patients in our cohort were CEA-negative. To address this urgent need, we developed two novel nomograms: a diagnostic model integrating CA50, CA125, and inflammatory indices, and a prognostic model incorporating TSP and MLR, which demonstrates a superior clinical net benefit compared to the AJCC 8th TNM staging system.
Given the high prevalence of CEA-negativity, standard CEA thresholds remain insufficient for early detection [12, 35]. While lowering the CEA cut-off (e.g., 1.765 ng/mL in our study) improved sensitivity [10, 36, 37], a multi-marker panel approach remains superior. Aligning with Butvilovskaya et al. [38] and Zhao et al. [39], our multivariate model confirms that integrating multiple markers significantly enhances diagnostic accuracy. Despite some debates [37], CA125 and CA50 emerged as robust independent predictors in our CEA-negative subgroup, likely reflecting distinct biological pathways, such as mucinous differentiation, that capture tumors missed by CEA.
The host immune response is increasingly recognized as critical for tumor identification. While the link between CRP and CRC is debated [19, 40], ratios like NLR, PLR, and MLR show diagnostic promise in other malignancies such as gliomas and lung cancer [41, 42]. In our specific cohort, isolated inflammation markers had limited diagnostic utility, but the FAR and CRP-to-Albumin Ratio (CAR) proved more effective, serving as stable indicators of systemic inflammatory and nutritional status [43, 44]. Crucially, our integrated model (Model 3) achieved the highest predictive performance, indicating that systemic inflammation acts synergistically with tumor antigens rather than as a standalone diagnostic tool. Additionally, while our prognostic cohort includes a small subgroup of Stage IV patients to capture the full real-world spectrum of CEA-negative CRC, this clinical inclusiveness may introduce minor survival heterogeneity. Nevertheless, our model demonstrated robust and stable predictive performance across the entire population. These findings suggest that, although pathological confirmation remains the gold standard, our diagnostic nomogram serves as a non-invasive pre-endoscopic triage tool. It is particularly useful for prioritizing urgent colonoscopies, assessing risk in patients with suspicious imaging pending pathology, and supporting repeat biopsies when initial results are inconclusive.
Regarding prognosis, our study highlights a shift from traditional tumor burden markers to microenvironment indicators. Unlike previous studies suggesting elevated conventional markers (e.g., CA125, CA242) link to poor survival [30, 45], these markers lost independent prognostic significance in our CEA-negative multivariate analysis. Instead, the systemic MLR and local TSP drove survival outcomes. A high MLR indicates an immunosuppressive shift facilitating immune escape [46, 47], while high TSP (> 50%) signifies a desmoplastic barrier hindering immune infiltration [31]. This supports a refined Seed and Soil hypothesis: when the seed (tumor biomarkers) is silent, the outcome is dictated by the soil-the systemic immune competence and local stromal permissiveness. However, given that OS is confounded by postoperative treatments, recurrence management, comorbidities, and other non-tumor factors, these findings require cautious interpretation, and our nomogram should be viewed as a composite risk indicator, not a pure measure of tumor biology.
Several limitations warrant mention. First, our retrospective, single-center design introduces unavoidable selection bias. Furthermore, deriving optimal cut-offs and developing models within the same cohort carries a substantial risk of overfitting. Although internally validated, the external validity of our nomograms remains limited, necessitating rigorous validation in independent, multi-center cohorts before clinical implementation. Second, our discrepancy with Dai et al. [30] regarding the prognostic value of CA242 may be attributed to missing data in our cohort, which necessitated its exclusion from the final multivariate model. Third, our analysis relied on static preoperative baselines; future studies should incorporate longitudinal monitoring to assess how these markers evolve during treatment. Fourth, because our diagnostic nomogram incorporates systemic inflammatory indices, its reliability may be compromised in patients with acute or active chronic inflammation (e.g., diverticulitis, infectious colitis, active IBD), which can artificially elevate these markers and cause false-positive results. Therefore, this nomogram should be applied to patients with suspicious lesions only after active systemic inflammatory confounders have been excluded. Fifth, because OS is confounded by subsequent treatments and comorbidities, our nomogram reflects a composite clinical risk rather than pure tumor biology. By contrast, RFS would offer a more direct measure and should be employed in future prospective validation.
Finally, our nomograms should be viewed as complementary to emerging ctDNA-based assays. While ctDNA enables highly sensitive detection of molecular residual disease and systemic genetic alterations [48], our models uniquely capture the local tumor microenvironment (e.g., TSP) and host immune response using routine, cost-effective parameters. This makes them ideal for immediate, first-line triage, particularly in resource-limited settings, where they can prioritize patients for subsequent ctDNA testing [49]. Future integration of ctDNA into this multidimensional framework could yield next-generation models that comprehensively combine tumor genetics with microenvironmental and host factors.
In conclusion, this study provides a hypothesis-generating framework for the management of CEA-negative CRC. We demonstrate that integrating serum tumor markers (CA125, CA50) with systemic inflammatory (MLR) and stromal (TSP) features significantly enhances both diagnosis and risk stratification. Our developed nomograms offer a practical, non-invasive tool to resolve the diagnostic uncertainty in this blind spot population, guiding more personalized clinical decision-making.
Supplementary Information
Supplementary Material 1: Supplemental Table S1. The R packages used in this study. Supplemental Table S2. ROC curve analysis of blood tumor markers. Supplemental Table S3. ROC curve analysis of blood inflammation indicators. Supplemental Table S4. Binary logistic regression analysis based on cut-off values of tumor markers to determine the complexity of CRC disease (model 1). Supplemental Table S5. Binary logistic regression analysis based on cut-off values of inflammatory indicators to determine the complexity of CRC disease (model 2). Supplemental Table S6. Binary logistic regression analysis based on cut-off values of blood indicators to determine the complexity of CRC disease (model 3). Supplemental Table S7. Comparison between 3 models in CEA-negative CRC disease. Supplemental Table S8. Diagnostic performance of the integrated nomogram in the training and validation cohorts. Supplemental Table S9. Survival times were estimated by the Kaplan-Meier method and compared with the use of a two-sided log-rank test. Supplemental Table S10. The C-index of the Nomogram Model. Supplemental Figure S1. Distribution and sensitivity of biomarkers in colorectal cancer (CRC) and benign colorectal disease (BCD) patients. Supplemental Figure S2. Diagnostic performance and threshold optimization of serum biomarkers. Supplemental Figure S3. Construction of a nomogram prediction model using individual blood tumor markers and inflammation indicators.
Acknowledgements
We would like to express our sincere gratitude to the Department of Pathology at Shanghai General Hospital for their strong support and invaluable assistance with the pathological evaluations in this study.
Authors’ contributions
C.H. (Chen Huang), J. Q. and ZM. F. conceived this experiments and revised the manuscript. ZM. F. and Z. L. contributed to the manuscript writing. ZM. F., YT. X., RC. Z. and Z. L. contributed to the raw data gathering and patients follow-up. C. H. (Chao Han), and Y. Z. contributed to bioinformatics analysis. Z. L. and C. H. (Chao Han) contributed to Statistical Analysis. All authors read and approved the final paper.
Funding
This work was supported by the National Natural Science Foundation of China (Nos. 82072662 to C. Huang, 82403429 to ZM. Fu, and 82103052 to C. Han); the Shanghai Municipal Education Commission-Gaofeng Clinical Medicine Grant Support (Grant No. 20191425); and the Shanghai Three-year Action Plan to Promote Clinical Skills and Clinical Innovation in Municipal Hospitals (Grant No. SHDC2020CR4022).
Data availability
No datasets were generated or analysed during the current study.
Declarations
Ethics approval and consent to participate
This retrospective study adhered to the Declaration of Helsinki and received approval from the Clinical Research Ethics Committee of Shanghai General Hospital.
Competing interests
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.
Zhongmao Fu, Zai Luo and Chao Han contributed equally to this work.
Contributor Information
Jing Qin, Email: qin.jing@zs-hospital.sh.cn.
Chen Huang, Email: huangchen0204@sjtu.edu.cn.
References
- 1.Sung H, Ferlay J, Siegel RL, Laversanne M, Soerjomataram I, Jemal A, et al. Global cancer statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA-Cancer J Clin. 2021;71(3):209–49. [DOI] [PubMed] [Google Scholar]
- 2.Morgan E, Arnold M, Gini A, Lorenzoni, Cabasag CJ, Laversanne M, et al. Global burden of colorectal cancer in 2020 and 2040: incidence and mortality estimates from GLOBOCAN. Gut. 2023;72(2):338–44. [DOI] [PubMed] [Google Scholar]
- 3.Siegel RL, Wagle NS, Cercek A, Smith RA, Jemal A. Colorectal cancer statistics, 2023. CA-Cancer J Clin. 2023;73(3):233–54. [DOI] [PubMed] [Google Scholar]
- 4.Baqar AR, Wilkins S, Staples M, Lee CHA, Oliva K, McMurrick P. The role of preoperative CEA in the management of colorectal cancer: A cohort study from two cancer centres. Int J Surg. 2019;64:10–5. [DOI] [PubMed] [Google Scholar]
- 5.Konishi T, Shimada Y, Hsu M, Tufts L, Jimenez-Rodriguez R, Cercek A, et al. Association of Preoperative and Postoperative Serum Carcinoembryonic Antigen and Colon Cancer Outcome. JAMA Oncol. 2018;4(3):309–15. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Park JW, Lim SB, Kim DY, Jung KH, Hong YS, Chang HJ, et al. Carcinoembryonic antigen as a predictor of pathologic response and a prognostic factor in locally advanced rectal cancer patients treated with preoperative chemoradiotherapy and surgery. Int J Radiat Oncol Biol Phys. 2009;74(3):810–7. [DOI] [PubMed] [Google Scholar]
- 7.Aggarwal C, Meropol NJ, Punt CJ, Iannotti N, Saidman BH, Sabbath KD, et al. Relationship among circulating tumor cells, CEA and overall survival in patients with metastatic colorectal cancer. Ann Oncol. 2013;24(2):420–8. [DOI] [PubMed] [Google Scholar]
- 8.Paku M, Uemura M, Kitakaze M, Miyoshi N, Takahashi H, Mizushima T, et al. Clinical Significance of Preoperative and Postoperative Serum CEA and Carbohydrate Antigen 19 – 9 Levels in Patients Undergoing Curative Resection of Locally Recurrent Rectal Cancer. Dis Colon Rectum. 2023;66(3):392–400. [DOI] [PubMed] [Google Scholar]
- 9.Huang SH, Tsai WS, You JF, Hung HY, Yeh CY, Hsieh PS, et al. Preoperative Carcinoembryonic Antigen as a Poor Prognostic Factor in Stage I-III Colorectal Cancer After Curative-Intent Resection: A Propensity Score Matching Analysis. Ann Surg Oncol. 2019;26(6):1685–94. [DOI] [PubMed] [Google Scholar]
- 10.Ren GM, Li RK, Zheng GZ, Du KL, Dan HJ, Wu HZ, et al. Prognostic value of normal levels of preoperative tumor markers in colorectal cancer. Sci Rep. 2023;13(1):12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Chern YJ, Hsu HY, Hsu YJ, Hsu LY, Tsai WS, Liao CK, et al. Tumor Marker Trajectories and Survival Analysis in Patients With Normal Carcinoembryonic Antigen Ranges After Colorectal Cancer Resection. Dis Colon Rectum. 2024;67(1):62–72. [DOI] [PubMed] [Google Scholar]
- 12.Nicholson BD, Shinkins B, Mant D. Blood Measurement of Carcinoembryonic Antigen Level for Detecting Recurrence of Colorectal Cancer. JAMA-J Am Med Assoc. 2016;316(12):1310–1. [DOI] [PubMed] [Google Scholar]
- 13.Diakos CI, Charles KA, McMillan DC, Clarke SJ. Cancer-related inflammation and treatment effectiveness. Lancet Oncol. 2014;15(11):E493–503. [DOI] [PubMed] [Google Scholar]
- 14.Kasmann L, Bolm L, Schild SE, Janssen S, Rades D. Neutrophil-to-Lymphocyte Ratio Predicts Outcome in Limited Disease Small-cell Lung Cancer. Lung. 2017;195(2):217–24. [DOI] [PubMed] [Google Scholar]
- 15.Zhang ZF, Wang DZ, Zhang JJ, Ruan YL, Zhao L, Yang L, et al. Comparison of the effectiveness of chemotherapy combined with immunotherapy and chemotherapy alone in advanced biliary tract cancer and construction of the nomogram for survival prediction based on the inflammatory index and controlling nutritional status score. Cancer Immunol Immunother. 2023;72(11):3635–49. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Fang L, Yan FH, Liu C, Chen J, Wang D, Zhang CH, et al. Systemic Inflammatory Biomarkers, Especially Fibrinogen to Albumin Ratio, Predict Prognosis in Patients with Pancreatic Cancer. Cancer Res Treat. 2021;53(1):131–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Starzer AM, Steindl A, Mair MJ, Deischinger C, Simonovska A, Widhalm G, et al. Systemic inflammation scores correlate with survival prognosis in patients with newly diagnosed brain metastases. Br J Cancer. 2021;124(7):1294–300. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Bilen MA, Martini DJ, Liu Y, Lewis C, Collins HH, Shabto JM, et al. The prognostic and predictive impact of inflammatory biomarkers in patients who have advanced-stage cancer treated with immunotherapy. Cancer. 2019;125(1):127–34. [DOI] [PubMed] [Google Scholar]
- 19.Shoji F, Takeoka H, Kozuma Y, Toyokawa G, Yamazaki K, Ichiki M, et al. Pretreatment prognostic nutritional index as a novel biomarker in non-small cell lung cancer patients treated with immune checkpoint inhibitors. Lung Cancer. 2019;136:45–51. [DOI] [PubMed] [Google Scholar]
- 20.Ishihara H, Tachibana H, Takagi T, Kondo T, Fukuda H, Yoshida K, et al. Predictive Impact of Peripheral Blood Markers and C-Reactive Protein in Nivolumab Therapy for Metastatic Renal Cell Carcinoma. Target Oncol. 2019;14(4):453–63. [DOI] [PubMed] [Google Scholar]
- 21.Cai H, Li JC, Chen Y, Zhang Q, Liu Y, Jia HJ. Preoperative inflammation and nutrition-based comprehensive biomarker for predicting prognosis in resectable colorectal cancer. Front Oncol. 2023;13:9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Lehtomäki K, Heervä E, Kellokumpu-Lehtinen PL, Mustonen H, Salminen T, Joensuu H, et al. Transient Changes in Serum CEA, CA19-9, CRP, YKL-40, and IL-6 during Adjuvant Chemotherapy and Survival of Patients with Colorectal Cancer. Int J Mol Sci. 2023;24(7):17. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Zhang H, Lin F, Wang ZC. Mean platelet volume/platelet count ratio in combination with tumor markers in colorectal cancer: a retrospective clinical study. BMC Cancer. 2023;23(1):8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Morita S, Nomura T, Fukushima Y, Morimoto T, Hiraoka N, Shibata N. Does serum CA19-9 play a practical role in the management of patients with colorectal cancer? Dis Colon Rectum. 2004;47(2):227–32. [DOI] [PubMed] [Google Scholar]
- 25.Li CX, Zhang DF, Pang XL, Pu HJ, Lei M, Fan BB, et al. Trajectories of Perioperative Serum Tumor Markers and Colorectal Cancer Outcomes: A Retrospective, Multicenter Longitudinal Cohort Study. EBioMedicine. 2021;74:8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Li CX, Zhao K, Zhang DF, Pang XL, Pu HJ, Lei M, et al. Prediction models of colorectal cancer prognosis incorporating perioperative longitudinal serum tumor markers: a retrospective longitudinal cohort study. BMC Med. 2023;21(1):13. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Gao YF, Wang JP, Zhou Y, Sheng S, Qian SY, Huo XW. Evaluation of Serum CEA, CA19-9, CA72-4, CA125 and Ferritin as Diagnostic Markers and Factors of Clinical Parameters for Colorectal Cancer. Sci Rep. 2018;8:9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Kuusela P, Haglund C, Roberts PJ, Jalanko H, Comparison of, ca-50. a new tumor-marker, with carcinoembryonic antigen (CEA) and alpha-fetoprotein (AFP) in patients with gastrointestinal-diseases. Br J Cancer. 1987;55(6):673–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Li QG, Dai WX, Li YQ, Xu Y, Li XX, Cai SJ. Nomograms for predicting the prognostic value of serological tumor biomarkers in colorectal cancer patients after radical resection. Sci Rep. 2017;7:10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Dai X, Wang HR, Lu YQ, Chen Y, Liu Y, Huang SY. Development and Validation of a Nomogram to Predict Overall Survival in Stage I-III Colorectal Cancer Patients after Radical Resection with Normal Preoperative Serum Carcinoembryonic Antigen. Cancers. 2023;15(23):16. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Li TF, Yu ZK, Yang Y, Fu ZM, Chen ZA, Li Q, et al. Rapid multi-dynamic algorithm for gray image analysis of the stroma percentage on colorectal cancer. J Cancer. 2021;12(15):4561–73. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Tong GJ, Xu W, Zhang GY, Liu J, Zheng ZZ, Chen Y, et al. The role of tissue and serum carcinoembryonic antigen in stages I to III of colorectal cancer-A retrospective cohort study. Cancer Med. 2018;7(11):5327–38. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Pu HJ, Yang W, Liu MM, Pang XL, Chen YX, Xiong QX. Elevated postoperative carcinoembryonic antigen guides adjuvant chemotherapy for stage II colon cancer: a multicentre cohort retrospective study. Sci Rep. 2024;14(1):13. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Ramphal W, Boeding JRE, van Iwaarden M, Schreinemakers JMJ, Rutten HJT, Crolla R, et al. Serum carcinoembryonic antigen to predict recurrence in the follow-up of patients with colorectal cancer. Int J Biol Markers. 2019;34(1):60–8. [DOI] [PubMed] [Google Scholar]
- 35.Palmqvist R, Engarås B, Lindmark G, Hallmans G, Tavelin B, Nilsson O, et al. Prediagnostic levels of carcinoembryonic antigen and CA 242 in colorectal cancer:: A matched case-control study. Dis Colon Rectum. 2003;46(11):1538–44. [DOI] [PubMed] [Google Scholar]
- 36.Beom SH, Shin SJ, Kim CG, Kim JH, Hur H, Min BS, et al. Clinical Significance of Preoperative Serum Carcinoembryonic Antigen Within the Normal Range in Colorectal Cancer Patients Undergoing Curative Resection. Ann Surg Oncol. 2020;27(8):2774–83. [DOI] [PubMed] [Google Scholar]
- 37.Thomas DS, Fourkala EO, Apostolidou S, Gunu R, Ryan A, Jacobs I, et al. Evaluation of serum CEA, CYFRA21-1 and CA125 for the early detection of colorectal cancer using longitudinal preclinical samples. Br J Cancer. 2015;113(2):268–74. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Butvilovskaya VI, Popletaeva SB, Chechetkin VR, Zubtsova ZI, Tsybulskaya M, Samokhina LO, et al. Multiplex determination of serological signatures in the sera of colorectal cancer patients using hydrogel biochips. Cancer Med. 2016;5(7):1361–72. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Zhao R, Xia DG, Chen YW, Kai ZT, Ruan FY, Xia CR, et al. Improved diagnosis of colorectal cancer using combined biomarkers including < i>Fusobacterium nucleatum, fecal occult blood, transferrin, CEA, CA19-9, gender, and age. Cancer Med. 2023;12(13):14636–45. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Ghuan S, Van Hemelrijck M, Garmo H, Holmberg L, Malmström H, Lambe M, et al. Serum inflammatory markers and colorectal cancer risk and survival. Br J Cancer. 2017;116(10):1358–65. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Zheng SH, Huang JL, Chen M, Wang BL, Ou QS, Huang SY. Diagnostic value of preoperative inflammatory markers in patients with glioma: a multicenter cohort study. J Neurosurg. 2018;129(3):583–92. [DOI] [PubMed] [Google Scholar]
- 42.Zhu XM, Song HZ, Chen Y, Han FF, Wang Q, Cui YB. Neutrophil-to-Lymphocyte Ratio and Platelet-to-Lymphocyte Ratio in Blood to Distinguish Lung Cancer Patients from Healthy Subjects. Dis Markers. 2020;2020:5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Ying HQ, Chen W, Xiong CF, Wang YY, Li XJ, Cheng XX. Quantification of fibrinogen-to-pre-albumin ratio provides an integrating parameter for differential diagnosis and risk stratification of early-stage colorectal cancer. Cancer Cell Int. 2022;22(1):13. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Ruan GT, Xie HL, Yuan KT, Lin SQ, Zhang HY, Liu CA, et al. Prognostic value of systemic inflammation and for patients with colorectal cancer cachexia. J Cachexia Sarcopenia Muscle. 2023;14(6):2813–23. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.You WQ, Sheng NQ, Yan L, Chen HQ, Gong JF, He ZH, et al. The difference in prognosis of stage II and III colorectal cancer based on preoperative serum tumor markers. J Cancer. 2019;10(16):3757–66. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46.Chan JCY, Chan DL, Diakos CI, Engel A, Pavlakis N, Gill A, et al. The Lymphocyte-to-Monocyte Ratio is a Superior Predictor of Overall Survival in Comparison to Established Biomarkers of Resectable Colorectal Cancer. Ann Surg. 2017;265(3):539–46. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47.Stotz M, Pichler M, Absenger G, Szkandera J, Arminger F, Schaberl-Moser R, et al. The preoperative lymphocyte to monocyte ratio predicts clinical outcome in patients with stage III colon cancer. Br J Cancer. 2014;110(2):435–40. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48.Reinert T, Henriksen TV, Christensen E, Sharma S, Salari R, Sethi H, et al. Analysis of Plasma Cell-Free DNA by Ultradeep Sequencing in Patients With Stages I to III Colorectal Cancer. JAMA Oncol. 2019;5(8):1124–31. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49.Malla M, Loree JM, Kasi PM, Parikh AR. Using Circulating Tumor DNA in Colorectal Cancer: Current and Evolving Practices. J Clin Oncol. 2022;40(24):2846–57. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Supplementary Material 1: Supplemental Table S1. The R packages used in this study. Supplemental Table S2. ROC curve analysis of blood tumor markers. Supplemental Table S3. ROC curve analysis of blood inflammation indicators. Supplemental Table S4. Binary logistic regression analysis based on cut-off values of tumor markers to determine the complexity of CRC disease (model 1). Supplemental Table S5. Binary logistic regression analysis based on cut-off values of inflammatory indicators to determine the complexity of CRC disease (model 2). Supplemental Table S6. Binary logistic regression analysis based on cut-off values of blood indicators to determine the complexity of CRC disease (model 3). Supplemental Table S7. Comparison between 3 models in CEA-negative CRC disease. Supplemental Table S8. Diagnostic performance of the integrated nomogram in the training and validation cohorts. Supplemental Table S9. Survival times were estimated by the Kaplan-Meier method and compared with the use of a two-sided log-rank test. Supplemental Table S10. The C-index of the Nomogram Model. Supplemental Figure S1. Distribution and sensitivity of biomarkers in colorectal cancer (CRC) and benign colorectal disease (BCD) patients. Supplemental Figure S2. Diagnostic performance and threshold optimization of serum biomarkers. Supplemental Figure S3. Construction of a nomogram prediction model using individual blood tumor markers and inflammation indicators.
Data Availability Statement
No datasets were generated or analysed during the current study.