Tumor markers and multimodal magnetic resonance imaging in predicting rectal cancer stage and differentiation

Abstract

BACKGROUND

Rectal cancer is one of the common digestive system malignant tumors around the world. Its early diagnosis and staging are crucial for rectal cancer treatment and prognosis. In recent years, tumor markers have gradually received attention in early screening, treatment monitoring and prognostic evaluation of cancer, but their predictive role in rectal cancer staging and differentiation is still unclear.

AIM

To assess the prognostic value of tumor markers alpha-fetoprotein (AFP) cancer antigen 72-4 (CA72-4), carbohydrate antigen 19-9 (CA19-9), and carcinoembryonic antigen (CEA), alongside multimodal magnetic resonance imaging (MRI), for staging and differentiating rectal cancer in patients.

METHODS

This study retrospectively analyzed 167 patients with rectal cancer who were treated at our institution from January 2020 to December 2024. Each patient underwent serological testing and multimodal MRI for diagnosis. Histopathological examination after surgical resection or imaging based on follow-up was used as the gold standard. According to the T stage and differentiation degree, patients were divided into low stage group (T1-T2) and high stage group (T3-T4). In addition, they were divided into low-differentiation groups and high-differentiation groups according to their differentiation degree. We compared the accuracy, sensitivity and specificity of tumor marker levels and MRI in rectal cancer stage and differentiation.

RESULTS

The study's findings indicate that in the context of rectal cancer T staging, there is substantial concordance between MRI and clinicopathological assessments, with a Kappa coefficient of 0.789 (P < 0.001). Similarly, for various degrees of tumor differentiation, MRI and clinicopathological evaluations demonstrated substantial agreement, with a Kappa coefficient of 0.651 (P < 0.001). Notably, the concentrations of tumor markers CA19-9, CA72-4, CEA, and AFP were significantly elevated in the T3-T4 stage compared to the T1-T2 stage. Furthermore, these markers were significantly higher in the low-differentiation group compared to the high-differentiation group (P < 0.05). The combined use of tumor markers and MRI for preoperative T staging of rectal cancer yielded a diagnostic sensitivity of 93.7% and a specificity of 94.6%, as evidenced by the receiver operating characteristic analysis, with an area under the curve of 0.947. For tumor differentiation, the diagnostic sensitivity and specificity were 93.6% and 97.1%, respectively, with an area under the curve of 0.978 (95% confidence interval: 0.946-1.000), surpassing the accuracy of individual detection methods.

CONCLUSION

The CA19-9, CA72-4, CEA and AFP tumor markers combined with multimodal MRI have high sensitivity and specificity in diagnosing rectal cancer stage and differentiation. Their diagnostic efficacy is significantly better than that of single tests, which can effectively improve the predictive ability of rectal cancer stage and differentiation, provide a more reliable diagnostic reference for clinical practice, and have important clinical significance.

Key Words: Multimodal magnetic resonance imaging; Rectal cancer; T stage; Tumor markers; Differentiation

Core Tip: This study included 167 rectal cancer patients who underwent serological tests and multimodal magnetic resonance imaging. Results showed significant concordance between magnetic resonance imaging and pathological T-staging. Combined use of tumor markers (alpha-fetoprotein cancer antigen 72-4, carbohydrate antigen 19-9, and carcinoembryonic antigen) and magnetic resonance imaging had high sensitivity/specificity for staging/differentiation, outperforming single tests, thus enhancing diagnostic accuracy for clinical practice.

INTRODUCTION

Rectal cancer is a frequently occurring malignant tumor in the digestive system, with its incidence increasing annually. Early detection and precise staging are crucial for its prognosis[1]. Approximately 1.8 million new cases are diagnosed annually, leading to 900000 related deaths[2]. Improvements in diagnostic and treatment methods have led to a 5-year survival rate of around 90% for early-stage rectal cancer patients, but this decreases to 13.1% for advanced stage cases[3]. Metastatic colorectal cancer is identified in about 40% to 50% of colorectal cancer patients[4]. In China, there is a growing trend in rectal cancer cases with an earlier age of onset[5]. Early stage rectal cancer is frequently asymptomatic, complicating diagnosis. Patients usually have poor prognoses when they show more advanced symptoms, such as rectal bleeding and abdominal pain[6]. Therefore, prompt identification, diagnosis, and treatment of rectal cancer is vital for improving patient survival and quality of life.

Colonoscopy is the main method for diagnosing rectal cancer, whereas computed tomography (CT) and positron emission tomography (PET) scans are commonly used to assess the spread of rectal cancer tumors[7]. Magnetic resonance imaging (MRI) is highly effective in diagnosing rectal cancer, particularly lesions under 1 cm, due to its excellent soft tissue resolution, various scanning techniques, and lack of radiation risks[8]. Multi-modal MRI has recently become crucial for diagnosing and staging rectal cancer because of its excellent resolution and tissue contrast. Imaging techniques alone may not completely uncover the biological traits of the tumor, particularly when evaluating rectal cancer staging and differentiation[9]. Tumor markers are crucial for cancer diagnosis and treatment monitoring. Blood markers like alpha-fetoprotein (AFP), cancer antigen 72-4 (CA72-4), carbohydrate antigen 19-9 (CA19-9), and carcinoembryonic antigen (CEA) are overexpressed in various tumors, indicating growth, invasion, and metastasis[10]. These biomarkers correlate with tumor behavior and clinical progression, aiding in diagnosis and staging[11]. However, relying solely on imaging or biomarker detection can lead to misdiagnosis or missed diagnosis in rectal cancer. This study aims to evaluate the diagnostic value of combining tumor markers (AFP, CA19-9, CEA, and CA72-4) with enhanced MRI for preoperative staging and differentiation of rectal cancer, thereby improving preoperative assessment.

MATERIALS AND METHODS

Study population

This retrospective study involved the selection of 167 patients diagnosed with rectal cancer via colonoscopy and proctoscopy at our institution between January 2020 and December 2024. These patients underwent surgical resection within 1 week following imaging examinations. After applying the inclusion and exclusion criteria, 109 patients were deemed eligible and subsequently screened, as depicted in the patient screening flowchart in Figure 1. Oncological variables and outcomes were documented, with tumors staged according to the tumor-node-metastasis classification of the Union for International Cancer Control. Patients were categorized into low-stage (T1-T2, n = 59) and high-stage (T3-T4, n = 50) groups. Furthermore, tumors were classified into poorly differentiated (n = 52) and well-differentiated (n = 57) groups, with poorly and moderately differentiated tumors being collectively categorized as poorly differentiated.

Figure 1 Patient selection flowchart. MRI: Magnetic resonance imaging.

Inclusion and exclusion criteria

Inclusion criteria: (1) Diagnosis in accordance with the “National Comprehensive Cancer Network Guidelines for Diagnosis and Treatment of Rectal Cancer”[12]; (2) Availability of comprehensive clinical, pathological, and imaging data for assessment; (3) Underwent preoperative multimodal MRI; and (4) Absence of prior chemotherapy, radiotherapy, or any other treatments before study enrollment.

Exclusion criteria: (1) Presence of other malignant tumors; (2) Severe cardiovascular disease, hepatic or renal failure, or other significant systemic diseases; (3) Incomplete imaging or biomarker data; (4) Prior treatment for rectal cancer; (5) Suboptimal MRI image resolution; and (6) Pregnant or lactating women.

Ethical considerations

The ethical standards of the local ethics committee and the Declaration of Helsinki (1964) and its subsequent updates were adhered to in all procedures. Informed consent was not required for retrospective image data, as approved by the institutional review board.

Detection methods

All patients fasted for 8-12 hours before the examination. Patients with intestinal obstruction symptoms underwent bowel preparation.

Tumor marker detection

Venous blood samples were obtained from patients, and serum was isolated via centrifugation. CA199, CA72-4, CEA, and AFP concentrations were quantified using an enzyme-linked immunosorbent assay (ELISA). On the second day following admission, 4-5 mL of fasting blood was collected and centrifuged at 3000 rpm for 10 minutes within 1 hour of collection.

Multimodal MRI examination

Patients were positioned supine on the scanning table for initial localization imaging[13]. The primary imaging sequences included T2-weighted imaging for axial and sagittal plane scans to assess the tumor’s location, size, and penetration through the intestinal wall, utilizing scanning parameters of repetition time = 4000 millisecond/echo time = 80 milliseconds and a 3 mm slice thickness. MRI examinations were performed using a 3.0-T scanner (GE Discovery MR750), with a spatial resolution of 0.5 mm × 0.5 mm × 3 mm for T2-weighted sequences. Diffusion-weighted imaging was used to differentiate tumor tissue, with b values set at 0, 500, and 1000 second/mm² and a 3 mm slice thickness. Dynamic contrast-enhanced T1-weighted imaging was conducted following intravenous administration of a contrast agent to evaluate the degree of tumor vascularization, using scanning parameters of repetition time = 3.5 millisecond/echo time = 1.5 milliseconds and a 3 mm slice thickness. Upon completion of all imaging sequences, two radiologists, each with over 10 years of experience, conducted meticulous image analysis. The evaluation was based on the tumor’s morphology, size, invasion, and vascular characteristics, leading to a comprehensive assessment of T stage and degree of differentiation. Detailed diagnostic reports were subsequently prepared, outlining the specific location, size, stage, and degree of differentiation of the tumor as determined by MRI results. These reports provide a precise foundation for clinical treatment decisions (Table 1).

Table 1 Magnetic resonance imaging staging criteria.

MRI staging criteria	Description
T1 stage	The tumor is limited to the mucosal layer, presenting as a polypoid structure within the lumen or as a local mass
T2 stage	The thickness of the wall exceeds 0.5 cm, but the tumor has not yet extended to the surrounding fat tissue, with a clear distinction from surrounding fat
T3 stage	The tumor extends to the surrounding fat tissue but has not reached the fascia. This stage is defined as tumor penetration through the wall layer and invasion into surrounding fat
T4 stage	The tumor further invades the pelvic region and may be accompanied by distant metastasis

MRI: Magnetic resonance imaging.

Pathological examination

Pathological specimens were preserved in formalin and diagnosed by two pathologists, each with over 10 years of experience. Imaging results were compared with the gold standard, which comprised surgical intervention with histopathological analysis and follow-up imaging. The follow-up and comparative imaging modalities included CT, MRI, and fluorine-18-fluorodeoxyglucose-positron emission tomography (PET) CT. Tumor differentiation was classified based on World Health Organization 2019 criteria[14]: (1) High-grade: G1 (well-differentiated) and G2 (moderately differentiated); and (2) Low-grade: G3 (poorly differentiated) and G4 (undifferentiated).

Observation indicators

The following comparisons were made: (1) Concordance between MRI findings and the pathological diagnostic gold standard in patients with varying stages and degrees of differentiation of rectal cancer; (2) The comparison of CA199, CA72-4, CEA, and AFP levels in patients with different stages and degrees of differentiation of rectal cancer; (3) The construction of receiver operating characteristic (ROC) curves, with the calculation of the area under the curve (AUC) and 95% confidence intervals to assess diagnostic accuracy. The sensitivity and specificity of various detection methods were compared, with a P value less than 0.05 considered statistically significant.

Statistical analysis

Data processing was performed in SPSS version 26.0. For normally distributed quantitative data, the results were articulated as mean ± SD, with intergroup comparisons conducted via independent samples t-test. For data deviating from a normal distribution, results were represented as the median accompanied by the interquartile range, and the Mann-Whitney U test was utilized for intergroup comparisons. Categorical variables were delineated as frequency and percentage, with group differences evaluated using the χ² test or Fisher’s exact test.

RESULTS

Baseline characteristics

The baseline characteristics of the 109 rectal cancer patients were as follows: Most patients (62.39%) were 60 years of age or older, while 37.61% were under 60. There was a slight male predominance (55.96%) compared to females (44.04%). In terms of lymph node involvement, 36.70% of patients exhibited lymphatic metastasis, whereas 63.30% did not. Vascular and nerve infiltration were relatively rare, occurring in 6.42% and 5.50% of cases, respectively. Most tumors were moderately differentiated (84.40%), with 11.93% classified as poorly differentiated and 3.67% as highly differentiated. The distribution of T staging was as follows: T1 (17.43%), T2 (42.20%), T3 (33.94%), and T4 (6.42%). Further details are provided in Table 2. MRI and clinicopathological examination showed a high degree of consistency between the two T staging methods (Kappa coefficient = 0.789, P < 0.001). Similarly, the diagnostic results for the degree of differentiation were largely consistent between the two methods (Kappa coefficient = 0.651, P < 0.001), as demonstrated in Tables 3 and 4.

Table 2 Demographic characteristics of patients with rectal cancer, n (%).

Clinicopathological parameters	Cases
Age
< 60	41 (37.61)
≥ 60	68 (62.39)
Gender
Male	61 (55.96)
Female	48 (44.04)
Lymphatic metastasis
Yes	40 (36.70)
No	69 (63.30)
Vascular invasion
Yes	7 (6.42)
No	102 (93.58)
Nerve infiltration
Yes	6 (5.50)
No	103 (94.50)
Degree of differentiation
Poorly	13 (11.93)
Moderately	92 (84.40)
Highly	4 (3.67)
T staging
T1	19 (17.43)
T2	46 (42.20)
T3	37 (33.94)
T4	7 (6.42)

Table 3 Consistency between magnetic resonance imaging and clinical pathological examinations in the diagnosis of T-stage of rectal cancer.

MRI staging	Pathological staging (gold standard)		Total (n)
	T1-T2	T3-T4
T1-T2	60	6	66
T3-T4	5	38	43
Total (n)	65	44	109

P < 0.001. Consistency statistics: Cohen’s κ = 0.789 (95% confidence interval: 0.692-0.886). Agreement level: Substantial. MRI: Magnetic resonance imaging.

Table 4 Consistency between magnetic resonance imaging and clinicopathological examinations in the diagnosis of different degrees of rectal cancer.

MRI grading	Pathological grading (gold standard)		Total
	High-grade	Low-grade
High-grade	3	2	5
Low-grade	1	103	104
Total	4	105	109

P < 0.001. Consistency statistics: Cohen’s κ = 0.651 (95% confidence interval: 0.012-1.290). Agreement strength: Substantial (Landis and Koch criteria). Definition of terms: High-grade: Well/moderately differentiated tumors; Low-grade: Poorly differentiated/undifferentiated tumors; MRI: Magnetic resonance imaging.

Tumor marker expression in rectal cancer patients across T stages

AFP, CEA, CA72-4, and CA19-9 Levels were significantly higher in patients at the T3-T4 stages compared to those at the T1-T2 stages (P < 0.05), as presented in Table 5.

Table 5 Comparison of carbohydrate antigen 19-9, cancer antigen 72-4, carcinoembryonic antigen, and alpha-fetoprotein levels betweenT1-T2 and T3-T4 stages, mean ± SD.

Indicator	T1-T2 (n = 65)	T3-T4 (n = 44)	P value
CA19-9 (kU/L)	10.37 ± 2.05	43.31 ± 4.52	< 0.001
CA72-4 (kU/L)	6.43 ± 1.71	12.62 ± 2.14	< 0.001
CEA (μg/L)	13.28 ± 3.75	23.26 ± 2.32	< 0.001
AFP (μg/L)	16.12 ± 3.83	21.16 ± 2.53	< 0.001

CA19-9: Carbohydrate antigen 19-9; CA72-4: Cancer antigen 72-4; CEA: Carcinoembryonic antigen; AFP: Alpha-fetoprotein.

Tumor marker expression in rectal cancer patients with varying degrees of differentiation

During the T3-T4 stages, AFP, CEA, CA72-4, and CA19-9 Levels were markedly elevated compared to the T1-T2 stages, as illustrated in Table 6.

Table 6 Comparison of carbohydrate antigen 19-9, cancer antigen 72-4, carcinoembryonic antigen, and alpha-fetoprotein levels in different degrees, mean ± SD.

Indicator	Low differentiation (n = 105)	High differentiation (n = 4)	P value
CA19-9 (kU/L)	55.43 ± 5.12	9.62 ± 2.71	< 0.001
CA72-4 (kU/L)	14.15 ± 3.63	7.34 ± 1.25	< 0.001
CEA (μg/L)	14.82 ± 2.16	7.73 ± 1.48	< 0.001
AFP (μg/L)	3.94 ± 0.58	3.26 ± 0.51	0.022

CA19-9: Carbohydrate antigen 19-9; CA72-4: Cancer antigen 72-4; CEA: Carcinoembryonic antigen; AFP: Alpha-fetoprotein.

Diagnostic value of tumor markers combined with MRI in preoperative rectal cancer T staging

ROC analysis revealed that the combination of tumor markers with MRI for preoperative T staging of rectal cancer achieved a diagnostic sensitivity and specificity of 93.7% and 94.6%, respectively, with an AUC of 0.947, as depicted in Table 7 and Figure 2A.

Figure 2 Receiver operating characteristic curve assessing the diagnostic efficacy of tumor markers combined with magnetic resonance imaging. A: Receiver operating characteristic curve assessing the diagnostic efficacy tumor markers combined with magnetic resonance imaging in preoperative T stage of rectal cancer; B: Receiver operating characteristic curve assessing the diagnostic efficacy of tumor markers combined with magnetic resonance imaging in rectal cancer differentiation. AFP: Alpha-fetoprotein; MRI: Magnetic resonance imaging; CA72-4: Cancer antigen 72-4; CA19-9: Carbohydrate antigen 19-9; CEA: Carcinoembryonic antigen.

Table 7 Diagnostic value of tumor markers combined with magnetic resonance imaging in preoperative T stage of rectal cancer.

Indicator	AUC	95%CI	P value	Sensitivity (%)	Specificity (%)
AFP	0.669	0.604-0.735	< 0.001	73.5	76.7
MRI	0.771	0.709-0.833	< 0.001	83.1	85.8
CA72-4	0.789	0.741-0.836	< 0.001	65.2	78.6
CA19-9	0.846	0.802-0.889	< 0.001	89.5	82.6
CEA	0.879	0.838-0.921	< 0.001	85.3	91.2
Combined detection	0.947	0.899-0.994	< 0.001	93.7	94.6

AUC: Area under the curve; CI: Confidence interval; AFP: Alpha-fetoprotein; MRI: Magnetic resonance imaging; CA72-4: Cancer antigen 72-4; CA19-9: Carbohydrate antigen 19-9; CEA: Carcinoembryonic antigen.

Diagnostic value of tumor markers combined with MRI in the degree of differentiation of rectal cancer

ROC analysis indicated that the combination of tumor markers and MRI achieved a diagnostic sensitivity of 93.6% and specificity of 97.1% for rectal cancer differentiation, with an AUC of 0.978, as illustrated in Table 8 and Figure 2B.

Table 8 Diagnostic value of tumor markers combined with magnetic resonance imaging in differentiation degree of rectal cancer patients.

Indicator	AUC	95%CI	P value	Sensitivity (%)	Specificity (%)
AFP	0.815	0.721-0.896	< 0.001	75.4	76.2
CA19-9	0.887	0.817-0.955	< 0.001	89.2	91.2
CEA	0.894	0.842-0.964	< 0.001	90.5	88.1
CA72-4	0.916	0.866-0.975	< 0.001	89.6	89.4
MRI	0.923	0.873-0.975	< 0.001	91.3	93.1
Combined detection	0.978	0.946-0.998	< 0.001	93.6	97.1

AUC: Area under the curve; CI: Confidence interval; AFP: Alpha-fetoprotein; MRI: Magnetic resonance imaging; CA72-4: Cancer antigen 72-4; CA19-9: Carbohydrate antigen 19-9; CEA: Carcinoembryonic antigen.

DISCUSSION

This study reveals that MRI exhibits certain inconsistencies compared to conventional pathological staging in the preoperative T-staging of rectal cancer. Notably, MRI has a conservative bias in diagnosing T1-T2 stage tumors, largely attributable to its limitations in detecting micro-tumor invasion, particularly in evaluating tumor penetration of the muscularis propria. Additionally, MRI often underestimates tumor aggressiveness in T3-T4 staging. This finding aligns with previous research that identified limitations in the ability of MRI to assess rectal cancer invasion depth[15-17]. Therefore, it is important to recognize the potential restrictions of MRI when using it for T-staging and to consider combining it with other diagnostic methods to improve overall assessment accuracy[18].

Regarding the comparison of tumor marker levels between T1-T2 and T3-T4 stages, AFP, CEA, CA72-4, and CA19-9 Levels were significantly higher in T3-T4 patients than in T1-T2 patients, potentially reflecting larger tumor size and aggressiveness. These data suggest a close relationship between tumor marker levels and its biological behavior, making them valuable indicators for tumor invasiveness and prognosis. This finding is consistent with previous research[19-21]. Therefore, incorporating tumor marker detection can provide more information for rectal cancer staging and treatment decisions.

In diagnosing differentiation grades, MRI successfully identified 103 out of 105 pathologically confirmed low-differentiation cases, and 3 out of 4 pathologically diagnosed high-differentiation cases. This indicates that MRI has relatively high accuracy in diagnosing the differentiation grade of rectal cancer. Previous studies found that MRI is valuable in assessing extramural vascular invasion in rectal cancer, especially in predicting distant metastasis risk[22-25]. Results of this study demonstrate that CA19-9, CA72-4, CEA, and AFP levels in the low-differentiation group were significantly higher than those in the high-differentiation group, potentially reflecting higher tumor burden and aggressiveness. This result suggests that tumor markers can be used both for tumor detection and monitoring and as auxiliary indicators for assessing tumor invasiveness and predicting patient prognosis. Studies have shown that elevated tumor marker levels are associated with aggressive features of rectal cancer[26].

ROC analysis further demonstrates that the combination of tumor markers and multi-modal MRI has high diagnostic efficacy in T-staging and differentiation assessment of rectal cancer. This suggests that a comprehensive application of MRI and tumor marker detection can more accurately assess the invasiveness and differentiation degree of rectal cancer, thereby providing patients with more appropriate treatment plans[27]. The synergy arises from MRI’s ability to delineate anatomical invasion (e.g., muscularis propria penetration) and tumor markers’ reflection of tumor biological aggressiveness (e.g., angiogenesis, proliferation). This combined approach may particularly benefit clinical decisions requiring both structural and molecular insights. The effectiveness of this comprehensive diagnostic approach is consistent with previous research findings[14]. The application of comprehensive diagnostic methods can improve diagnostic sensitivity and specificity and also provide important information for the personalized treatment of rectal cancer patients.

Limitations

As a retrospective single-center study, this study has the following limitations: Firstly, because the data for this study comes from the electronic medical records of patients from a single center in China, there may be bias in the admission rate, potentially impacting the accuracy of the results. Secondly, laboratory indicator data may have differences due to environmental influences. Finally, due to the relatively small sample size, the findings of this study should be approached with caution. Future studies should include more centers and larger cohorts to eliminate selection bias and improve generalizability. Additionally, integrating high-resolution MRI or artificial intelligence-assisted image segmentation could enhance micro-invasion detection in T1-T2 stages.

CONCLUSION

In conclusion, this research investigated the usefulness of combining tumor markers with multi-modal MRI for preoperative T-staging and differential diagnosis of rectal cancer. The findings demonstrate that this combination method has significant advantages in improving the sensitivity and specificity of preoperative diagnosis of rectal cancer. This approach could help clinicians more precisely formulate treatment strategies, such as avoiding overtreatment in early-stage patients while optimizing neoadjuvant therapy for advanced cases. However, the study also revealed the limitations of MRI in accurately identifying early tumor stages and completely matching pathological diagnoses. Future research should further optimize imaging techniques and evaluation criteria, and increase sample size to improve diagnostic accuracy and reliability, so as to better guide personalized treatment strategies for rectal cancer.