Value of brief hematological characteristics in differentiating carcinoembryonic-antigen-negative colorectal cancer from benign colorectal diseases

Abstract

BACKGROUND

Colorectal cancer (CRC) remains one of the most common malignancies worldwide, with a significant subset of patients exhibiting absence of carcinoembryonic-antigen (CEA) expression. The lack of effective diagnostic method for CEA-negative CRC prevents its early treatment.

AIM

To identify potentially valuable biomarkers for identifying CEA-negative CRC, the hematological characteristics of patients with CEA-negative CRC was investigated.

METHODS

In this retrospective analysis, 74 patients were included who had been pathologically confirmed to have CEA-negative CRC, along with 79 individuals diagnosed with benign colorectal conditions. The utility of various biomarkers was evaluated using analysis of the receiver operating characteristic (ROC) curve.

RESULTS

Compared with patients with benign colorectal diseases, those with CEA-negative CRC had lower hemoglobin-to-red blood cell distribution width ratio (HRR) and lymphocyte-to-red blood cell distribution width ratio (LRR), and higher platelet-to-lymphocyte ratio (PLR) (P < 0.05). Correlation analysis showed that HRR was negatively correlated with T stage (r = -0.237), LRR was negatively correlated with T stage (r = -0.265) and distant metastasis (r = -0.321), and PLR was positively correlated with T stage (r = 0.251) (all P < 0.05). ROC analysis indicated that HRR outperformed LRR and PLR in identifying CEA-negative CRC. Combining HRR and PLR provided the highest area under the curve (area under the curve = 0.808; sensitivity = 82.43%; specificity = 68.35%) for distinguishing CEA-negative CRC from benign colorectal diseases.

CONCLUSION

HRR, LRR, and PLR alone or in combination could be used to distinguish CEA-negative CRC from benign colorectal diseases.

Key Words: Hematological parameters; Carcinoembryonic-antigen-negative; Colorectal cancer; Biomarkers; Inflammatory cell ratios

Core Tip: A retrospective analysis was used to investigate the hematological characteristics of carcinoembryonic-antigen (CEA)-negative colorectal cancer (CRC) with the aim of identifying potentially valuable biomarkers for distinguishing between CEA-negative CRC and benign colorectal diseases. We found that hematological parameters were associated with CEA-negative CRC and benign colorectal diseases. Moreover, hematological ratios were related to clinicopathological features of CEA-negative CRC. The results suggest that hemoglobin-to-red blood cell distribution width ratio, lymphocyte-to-red blood cell distribution width ratio, and platelet-to-lymphocyte ratio either individually or in combination could effectively differentiate CEA-negative CRC from benign colorectal diseases.

INTRODUCTION

Globally, colorectal cancer (CRC) ranks as one of the predominant malignancies and is the second most common cause of cancer-related mortality[1]. Projections by GLOBOCAN for morbidity and mortality suggest that by 2040, as dietary and lifestyle patterns evolve, the incidence of CRC is expected to escalate to 3.2 million new cases and result in 1.6 million deaths[2]. Currently, the primary methods of eradicating benign colorectal polyps and stage I-II CRC are colonoscopic polypectomy and surgical resection, respectively[3]. Typically, the early clinical indicators of CRC lack prominence, which results in delayed diagnoses as weight loss, anemia, abdominal pain, and bloody stools often emerge only during the later stages of the disease[4]. Stage IV CRC patients who are treated with radical surgical interventions face a grim prognosis, with a five-year survival rate at just 13.0%. Conversely, detecting CRC during its early stages significantly enhances the survival rate, raising it to nearly 90.0%[5]. Therefore, early diagnosis of CRC which can reduce advanced-stage disease and mortality is especially critical.

Common clinical tools for early screening and diagnosis of CRC are fecal occult blood test (FOBT) and sigmoidoscopy, which appear to reduce mortality[6]. However, there are still limitations. FOBT is a non-invasive, economical, and broadly used screening method, but its results are vulnerable to dietary and pharmacological influences. What’s more, it could only be detected in the presence of pathological tissue bleeding[7]. To date, direct observation with colonoscopy maintains the gold standard for screening[8]. Nevertheless, this procedure is expensive and invasive, necessitating extensive bowel preparation and posing risks of severe complications, such as bowel perforation[9]. Thus, the development of a reliable and less invasive biomarker for early CRC screening and diagnosis is crucial.

Carcinoembryonic-antigen (CEA) is the most widespread serum tumor biomarker to detect CRC. Nevertheless, CEA has low organ specificity and is usually present with high serum levels in most patients with malignant tumors. Therefore, CEA could not be used solely as a biomarker for cancer screening. In addition, it has been found that some CRC cases were CEA-negative, making CEA unavailable for widespread screening[10]. Currently, extensive research is dedicated to identifying reliable biomarkers for CRC diagnosis. Among these, systemic inflammatory cell ratios[11], as well as biomarkers such as circulating tumor DNA, proteins, tumor-derived cells, and microRNAs in blood samples, are being actively studied for their diagnostic potential in CRC[12]. However, few studies have focus on identification of CEA-negative CRC. Thus, there is an urgency for more effective biomarkers for the identification of CEA-negative CRC to assure timely initiation of treatment. In this study, we aimed to comprehensively investigate hematological characteristics of CEA-negative CRC patients so as to identify potentially valuable biomarkers for distinguishing between CEA-negative CRC and benign colorectal diseases.

MATERIALS AND METHODS

Study subjects

Clinical data of 74 CRC patients who had undergone radically surgical resection at our hospital between 1 September 2018 and 30 June 2023 were retrospectively analyzed. Data from 79 patients with benign colorectal diseases were also collected to serve as controls for comparative analysis, including colon polyps, rectal polyps and colorectal adenomas. Inclusion criteria of CRC patients: CRC diagnosed through histopathology, blood parameters, and clinicopathology. Patients with hematological disorders, active infectious diseases, recently received a blood transfusion, parasitic infections, infectious shock, and a history of other primary malignancies were excluded from the study. A histological report following surgery or colonoscopy confirms the histological diagnosis of CRC or colorectal polyps. Following the American Joint Committee on Cancer Tumor-Node-Metastasis Staging System (8^th edition), CRC patients were divided into different cancer staging groups and into groups with or without metastases. The research obtained ethical clearance from the hospital's ethics committee and adhered to the guidelines set forth in the Declaration of Helsinki. Given its retrospective nature, the requirement for obtaining informed consent from the participants was exempted by the ethics committee.

Data collection

Data for this study were sourced from the Hospital Information System and the Laboratory Information System. These data comprised the initial examination results of the patients upon their admission to the hospital. Blood cell counts were tested by the XN-10[B3] Automatic Blood Analyzer (Sysmex Corporation). The collected dataset encompassed absolute measurements of various hematological parameters including white blood cells, lymphocytes (LYM), red blood cells (RBC), hemoglobin (HGB), RBC distribution width (RDW), and platelet (PLT). Ratios such as HGB-to-RDW ratio (HRR), calculated as HGB divided by RDW, LYM-to-RDW ratio (LRR), defined as the absolute LYM count relative to RDW, and PLR, the ratio of absolute PLT count to absolute LYM count, were also derived. The quantification of serum CEA was performed using the ABBOTT ARCHITECT i2000SR Automatic Immunoassay Analyzer from Abbott Laboratories, United States. The CEA concentration of lower than 5.0 ng/mL of CRC patients were considered CEA-negative CRC in our hospital.

Statistical analysis

For the purposes of data analysis and the generation of graphical outputs, this study utilized GraphPad Prism 9 along with SPSS version 21.0 (SPSS Inc./IBM, Chicago, IL, United States). To determine data normality, the Kolmogorov-Smirnov test was employed. Quantitative data adhering to a normal distribution were depicted as mean ± SD. Comparisons involving two datasets utilized the independent t-test, while those involving three datasets were analyzed with one-way analysis of variance, supplemented by Post Hoc multiple comparison tests. When variances were equal, the Least Significant Difference method was applied, and for unequal variances, the Tamhane's T2 method was implemented. Non-normally distributed data were expressed as medians with interquartile ranges [M (Q1, Q3)], analyzed using the Mann-Whitney U test for two datasets and the Kruskal-Wallis H test for three datasets. The significance thresholds for multiple comparisons were adjusted using the Bonferroni correction. Categorical data were presented as counts and percentages n (%), analyzed either by the χ² test or Fisher's exact test as appropriate. The diagnostic utility of hematological parameters in identifying CEA-negative CRC was assessed using receiver operating characteristic (ROC) curve analysis, which evaluated specificity, sensitivity, area under the curve (AUC), and the Youden index. Correlations between hematological parameters and clinicopathological characteristics were explored using Spearman's non-parametric correlation test. A P-value of less than 0.05 was considered statistically significant on a two-sided basis. The sample size was determined using G*Power software (ver. 3.1.9.7), with calculations based on a medium effect size (d = 0.5), a two-sided t-test, an alpha level of 0.05, a beta of 0.2, and a sample allocation ratio of N2/N1 = 79/74 = 1.07, achieving a statistical power of 0.8[13]. This yielded sample sizes of 62 for group 1 and 66 for group 2, demonstrating that the study's sample size was adequately powered for subsequent analyses.

RESULTS

Baseline characteristics

Table 1 presents the patient-related parameters and baseline characteristics. The study encompassed 74 individuals diagnosed with CEA-negative CRC, aged between 35 and 95 years. According to the grading standards, there were 34 (45.95%) patients in stage I + II and 40 (54.05%) in stage III + IV. There were 12 (16.22%) patients in T1 + T2 stage and 62 (83.78%) patients in T3 + T4 stage. Thirty-five (47.30%) CEA-negative CRC patients were identified as having no lymph node metastasis, while 64 (86.49%) had distant metastasis. A further 79 patients with benign colorectal diseases (range 29-88 years) were also included in the study. No significant differences between CEA-negative CRC and patients with benign colorectal diseases in age, gender, family history, smokers, drinkers, and chronic medical history (P > 0.05). FOBT positivity was higher in CEA-negative CRC compared with benign colorectal diseases (P < 0.001).

Table 1 Baseline characteristics between carcinoembryonic-antigen-negative colorectal cancer and benign colorectal diseases, n (%).

Characteristics	CEA-negative CRC	Benign colorectal diseases	P value
Number	74	79
Age (year), mean ± SD	59.39 ± 13.86	55.63 ± 12.13	0.076
Gender			0.409
Male	42 (56.76)	50 (63.29)
Female	32 (43.24)	29 (36.71)
Family history			0.066
Cancer	4 (5.41)	4 (5.06)
Others	1 (1.35)	8 (10.13)
Habit
Smokers	11 (14.86)	17 (21.52)	0.271
Drinkers	14 (18.92)	23 (29.11)	0.129
Chronic medical history			0.245
Hypertensives	13 (17.57)	17 (21.52)
Others	27 (36.49)	19 (24.05)
Positive FOBT	42 (56.76)	16 (20.25)	< 0.001
TNM stage
I + II	34 (45.95)
III + IV	40 (54.05)
Tumor invasion (T stage)
T1 + T2	12 (16.22)
T3 + T4	62 (83.78)
Lymph node metastasis (N stage)
N0	35 (47.30)
N1 + N2	39 (52.70)
Distant metastasis (M stage)
M0	64 (86.49)
M1	10 (13.51)

FOBT: Fecal occult blood test; CEA: Carcinoembryonic antigen; CRC: Colorectal cancer.

Hematological characteristics

The hematological characteristics of CEA-negative CRC and benign colorectal diseases patients are presented in Table 2. Statistically significant differences in levels of LYM, RBC, HGB, RDW, PLT, CEA, LRR and PLR in the CEA-negative CRC compared with benign colorectal diseases (all P < 0.05). LYM, RBC, HGB, HRR, and LRR were significantly lower in CEA-negative CRC patients compared with benign colorectal diseases patients; however, RDW, PLT, CEA, and PLR were obviously high in the cases. The levels of HRR were lower in stages I-IV, T1-T4, N0-N2, and M0-M1 compared to benign colorectal diseases (all P < 0.05). The levels of LRR were lower and the levels of PLR were higher in stages I-IV, T3 + T4, N0-N2, and M0-M1 compared to benign colorectal diseases (all P < 0.05; Figure 1).

Figure 1 Comparison of hemoglobin-to-red blood cell distribution width ratio, lymphocyte-to-red blood cell distribution width ratio, and platelet-to-lymphocyte ratio between the cancer stage and benign colorectal diseases groups. A: Hemoglobin-to-red blood cell distribution width ratio (HRR) in tumor-node-metastasis (TNM) stage and benign colorectal diseases groups; B: HRR in T stage and benign colorectal diseases groups; C: HRR in N stage and benign colorectal diseases groups; D: HRR in M stage and benign colorectal diseases groups; E: Lymphocyte-to-red blood cell distribution width ratio (LRR) in TNM stage and benign colorectal diseases groups; F: LRR in T stage and benign colorectal diseases groups; G: LRR in N stage and benign colorectal diseases groups; H: LRR in M stage and benign colorectal diseases groups; I: Platelet-to-lymphocyte ratio (PLR) in TNM stage and benign colorectal diseases groups; J: PLR in T stage and benign colorectal diseases groups; K: PLR in N stage and benign colorectal diseases groups; L: PLR in M stage and benign colorectal diseases groups. ^aP < 0.05 and ^aP < 0.01 vs benign colorectal diseases group; ^bP < 0.05 vs stage T1 + T2; ^cP < 0.05 vs stage M0; HRR: Hemoglobin-to-red blood cell distribution width ratio; LRR: Lymphocyte-to-red blood cell distribution width ratio; PLR: Platelet-to-lymphocyte ratio.

Table 2 Hematological characteristics between carcinoembryonic-antigen-negative colorectal cancer and benign colorectal diseases.

Characteristics	CEA-negative CRC (n = 74)	Benign colorectal diseases (n = 79)	P value
WBC (× 109/L)	6.43 ± 2.07	6.99 ± 2.21	0.109
LYM (× 109/L)	1.52 ± 0.62	1.77 ± 0.82	0.033
RBC (× 1012/L)	4.45 ± 0.65	4.67 ± 0.56	0.029
HGB (g/L)	120.41 ± 22.55	137.91 ± 14.23	< 0.001
RDW (%)	14.00 (13.00, 15.00)	13.00 (12.00, 13.00)	< 0.001
PLT (× 109/L)	293.88 ± 104.06	244.51 ± 60.51	0.001
CEA (ng/mL)	2.08 (1.55, 2.90)	1.67 (1.16, 2.64)	0.031
HRR	8.61 ± 2.42	10.95 ± 1.55	< 0.001
LRR	0.11 ± 0.05	0.14 ± 0.07	< 0.001
PLR	194.56 (137.11, 252.46)	137.01 (106.11, 204.93)	0.001

Normally distributed data are expressed as mean ± SD; non-normally distributed data are expressed as median and interquartile spacing [M (Q1, Q3)]. WBC: White blood cells; LYM: Lymphocyte; RBC: Red blood cells; HGB: Hemoglobin; RDW: Red blood cell distribution width; PLT: Platelet; CEA: Carcinoembryonic antigen; HRR: Hemoglobin-to-red blood cell distribution width ratio; LRR: Lymphocyte-to-red blood cell distribution width ratio; PLR: Platelet-to-lymphocyte ratio.

Association between hematological parameters and clinicopathological features

The association between biomarkers and cancer staging was investigated to further probe the predictive value of HRR, LRR and PLR. As shown in Figure 1 and Figure 2, among the subgroups of CEA-negative CRC, the significant differences in HRR were T3 + T4 stage < T1 + T2 stage (P < 0.05), in LRR were stage M1 < M0 (P < 0.05). Correlation analysis demonstrated that HRR was negatively correlated with T stage (r = -0.237, P = 0.042), LRR was negatively correlated with T stage (r = -0.265, P = 0.022) and distant metastasis (r = -0.321, P = 0.005), and PLR was positively correlated with T stage (r = 0.251, P = 0.031).

Figure 2 Correlation between hematological parameters and clinic-pathological features in carcinoembryonic-antigen-negative colorectal cancer patients. A: Correlation between hemoglobin-to-red blood cell distribution width ratio and T stage; B: Correlation between lymphocyte-to-red blood cell distribution width ratio (LRR) and T stage; C: Correlation between LRR and M stage; D: Correlation between platelet-to-lymphocyte ratio and T stage. HRR: Hemoglobin-to-red blood cell distribution width ratio; LRR: Lymphocyte-to-red blood cell distribution width ratio; PLR: Platelet-to-lymphocyte ratio.

Performance of single and combined parameters for identifying patients with CEA-negative CRC

Comparison of AUC values for single and combined biomarkers are shown in Table 3 and Figure 3. ROC analysis showed that the AUC of CEA was 0.601, with a high sensitivity of 77.03%, but the specificity was only 44.30%. ROC analysis showed that HRR had the highest efficacy in differentiating between CEA-negative CRC and benign colorectal disease, the AUC being 0.796 (0.727-0.865). The AUC of LRR and PLR were 0.660 (0.572-0.747) and 0.653 (0.565-0.740), respectively. The HRR, LRR and PLR in conjunction with other biomarkers was further analyzed. As we expected, the combination of biomarkers improved the value of clinical applications. Moreover, the combined HRR and PLR had the largest AUC in distinguishing between CEA-negative CRC and colorectal disease (AUC: 0.808, 95%CI: 0.741-0.875; sensitivity: 82.43%; specificity: 68.35%).

Figure 3 Receiver operating characteristic evaluation of the value of hematological parameters for diagnosis of patients with carcinoembryonic-antigen-negative colorectal cancer. A: Receiver operating characteristic (ROC) curve analysis of carcinoembryonic antigen (CEA), hemoglobin-to-red blood cell distribution width ratio (HRR), lymphocyte-to-red blood cell distribution width ratio (LRR), and platelet-to-lymphocyte ratio (PLR); B: ROC curve analysis of HRR + CEA, HRR + PLR, LRR + CEA, PLR + CEA, and HRR + PLR + LRR + CEA. HRR: Hemoglobin-to-red blood cell distribution width ratio; LRR: Lymphocyte-to-red blood cell distribution width ratio; PLR: Platelet-to-lymphocyte ratio; CEA: Carcinoembryonic antigen.

Table 3 The receiver operating characteristic curve assay for hematological parameters.

Indicators	AUC (95%CI)	P value	Cut-off value	Sensitivity (%)	Specificity (%)	Youden index
CEA (ng/mL)	0.601 (0.511-0.691)	0.031	1.54	77.03	44.30	0.21
HRR	0.796 (0.727-0.865)	< 0.001	9.42	58.11	86.08	0.44
LRR	0.660 (0.572-0.747)	0.001	0.13	70.27	65.82	0.36
PLR	0.653 (0.565-0.740)	0.001	150.20	72.97	59.49	0.32
HRR + CEA	0.799 (0.730-0.868)	< 0.001		62.16	86.08	0.48
HRR + PLR	0.808 (0.741-0.875)	< 0.001		82.43	68.35	0.51
LRR + CEA	0.673 (0.589-0.759)	< 0.001		68.92	63.29	0.32
PLR + CEA	0.655 (0.569-0.741)	< 0.001		75.68	50.63	0.26
HRR + PLR + LRR + CEA	0.807 (0.740-0.874)	< 0.001		85.14	64.56	0.50

CEA: Carcinoembryonic antigen; HRR: Hemoglobin-to-red blood cell distribution width ratio; LRR: Lymphocyte-to-red blood cell distribution width ratio; PLR: Platelet-to-lymphocyte ratio; AUC: Area under the receiver operating characteristic curve.

DISCUSSION

CRC in its initial stages is often asymptomatic, with the disease typically remaining undetected for over a decade[3]. It is commonly found that most diagnosed patients are at advanced stages III or IV, which frequently involve metastatic spread to various tissues, resulting in a five-year survival rate that falls below 10%[14]. Currently, the main methods of CRC screening are imaging and faecal testing, but both have some limitations[15]. Moreover, the detection of the tumor marker CEA is one of the main methods of early screening for CRC. However, increased concentrations of CEA seldom occur in the early stages of disease and are normally found in serious tumors[16]. Therefore, the identification of new, convenient, safe, more sensitive and specific, and non-invasive CRC biomarkers is important for early screening, diagnosis, treatment outcome, and prognosis of CEA-negative CRC patients.

This retrospective study pioneered the evaluation of the effectiveness of HRR, LRR, and PLR in detecting patients with CEA-negative CRC. Our findings indicate that RDW, PLT, CEA and PLR are elevated in CEA-negative CRC compared to benign colorectal disease; whereas LYM, RBC, HGB, HRR and LRR are significantly lower in CEA-negative CRC. Furthermore, our findings indicated a positive association between the PLR and T stage, while the LRR was inversely correlated with both T stage and the occurrence of distant metastases. Additionally, a positive correlation was noted between the HRR and T stage. Results of these indicate that the three parameters could be used as early screening biomarkers of CEA-negative CRC. Furthermore, our findings showed that HRR had a better ability to identify CEA-negative CRC predictive value than LRR and PLR. When used in conjunction with PLR, HRR may improve the effectiveness of differentiating benign diseases from CEA-negative CRC. And the AUC of LRR also increases when combined with CEA. Our findings could be further explained by the following factors.

On the one hand, cancer is widely recognized as a consequence of chronic inflammation[17]. Systemic inflammation is present preoperatively in about 20%-40% of CRC patients, which is a marker of poor prognosis[18]. On the other hand, hematological parameter in non-invasive blood routine test (BRT) have long been considered markers of systemic inflammatory response which are commonly used for routine testing in the early diagnosis of a variety of diseases, such as infections, anemia, and cancers[19]. A hypoxic tumor microenvironment promotes tumor invasion and metastasis, inducing immunosuppression and resistance[20]. RDW is an indicator in BRT that measures the degree of change in RBC size, and RDW have been shown to be diagnostic and indicated prognostic biomarkers for CRC[21,22]. LYM are an essential component in anti-tumor immunity, recognizing as well as killing cancer cells or releasing cytokines to trigger anti-tumor immunity, and high LYM counts have been shown to be a favorable factor for good survival in many human cancers[23]. As is known, bone marrow megakaryocytes produce PLTs, which are non-nucleated cells involved in the coagulation cascade and linked to thrombosis and inflammation[24]. In addition, activated PLTs are implicated with cancer progress and metastasis[25].

Existing literature has identified that the HRR, along with the RDW-to-LYM ratio (RLR) and the PLR, may serve as biomarkers for systemic inflammation. Specifically, the HRR is recognized for its potential to reflect the degree of oxidative stress and the overall systemic inflammatory response of an organism. It has been observed that diminished levels of HRR are associated with an unfavorable prognosis in individuals diagnosed with primary CRC and hepatocellular carcinoma[26]. Li et al[27] observed that CRC patients with low values of HRR had a poorer survival outcome and a negative correlation with the tumor stage of CRC. A study suggests that RLR could be used to predict HBV-related cirrhosis and that its use could reduce the need of frequent liver biopsies for chronic hepatitis B patients[28]. Additionally, PLR may reflect the balance between tumor inflammation and anti-tumor immunity. Retrospective studies have found PLR related to the diagnosis and prognosis of various malignancies such as gastric and rectal cancers[29,30]. Moreover, Domerecka et al[31] found that RLR and PLR may be helpful in detecting autoimmune hepatitis and looking for features that have progressed to cirrhosis in its course.

Above all, based on the numerous previous studies on systemic inflammatory cell ratios, it seems reasonable to explore the predictive efficacy of HRR, LRR, and PLR for CRC. Most importantly, this study is the first to investigate the use of HRR, LRR, and PLR alone or in combination with other biomarkers for differentiating CEA-negative CRC from benign colorectal diseases. Regrettably, there are inherent limitations within this study. Although every effort was made to gather a comprehensive range of clinical cases, the small sample size and the retrospective nature of the study are significant limitations. Thus, it is essential to undertake larger, prospective studies to robustly confirm and extend the findings presented here. Second, only participants of our nationality were included in this study, so the findings cannot be generalized to other ethnic groups. Third, we recommend that more experts conduct more relevant studies in different laboratories on populations with high racial, geographic, and national diversity to confirm the results of this study.

CONCLUSION

Early screening and diagnosis of CRC and its differential diagnosis from benign colorectal diseases is essential to improve prognosis. CEA alone is insufficient for this purpose, especially for the detection of CEA-negative CRC. HRR, LRR, and PLR are promising screening biomarkers for CEA-negative CRC. In contrast, the combination of HRR + PLR or HRR + PLR + LRR + CEA significantly improves the performance in differentiating CEA-negative CRC from benign colorectal diseases.

ACKNOWLEDGEMENTS

We thank the Department of Clinical Laboratory, the Guangxi International Zhuang Medicine Hospital Affiliated to Guangxi University of Chinese Medicine, China.