Prediction and analysis of albumin-bilirubin score combined with liver function index and carcinoembryonic antigen on liver metastasis of colorectal cancer

Abstract

BACKGROUND

Colorectal cancer (CRC) is a common malignant tumor, and liver metastasis is one of the main recurrence and metastasis modes that seriously affect patients’ survival rate and quality of life. Indicators such as albumin bilirubin (ALBI) score, liver function index, and carcinoembryonic antigen (CEA) have shown some potential in the prediction of liver metastasis but have not been fully explored.

AIM

To evaluate its predictive value for liver metastasis of CRC by conducting the combined analysis of ALBI, liver function index, and CEA, and to provide a more accurate liver metastasis risk assessment tool for clinical treatment guidance.

METHODS

This study retrospectively analyzed the clinical data of patients with CRC who received surgical treatment in our hospital from January 2018 to July 2023 and were followed up for 24 months. According to the follow-up results, the enrolled patients were divided into a liver metastasis group and a nonliver metastasis group and randomly divided into a modeling group and a verification group at a ratio of 2:1. The risk factors for liver metastasis in patients with CRC were analyzed, a prediction model was constructed by least absolute shrinkage and selection operator (LASSO) logistic regression, internal validation was performed by the bootstrap method, the reliability of the prediction model was evaluated by subject-work characteristic curves, calibration curves, and clinical decision curves, and a column graph was drawn to show the prediction results.

RESULTS

Of 130 patients were enrolled in the modeling group and 65 patients were enrolled in the verification group out of the 195 patients with CRC who fulfilled the inclusion and exclusion criteria. Through LASSO regression variable screening and logistic regression analysis. The ALBI score, alanine aminotransferase (ALT), and CEA were found to be independent predictors of liver metastases in CRC patients [odds ratio (OR) = 8.062, 95% confidence interval (CI): 2.545-25.540], (OR = 1.037, 95%CI: 1.004-1.071) and (OR = 1.025, 95%CI: 1.008-1.043). The area under the receiver operating characteristic curve (AUC) for the combined prediction of CRLM in the modeling group was 0.921, with a sensitivity of 78.0% and a specificity of 95.0%. The H-index was 0.921, and the H-L fit curve had χ² = 0.851, a P value of 0.654, and a slope of the calibration curve approaching 1. This indicates that the model is extremely accurate, and the clinical decision curve demonstrates that it can be applied effectively in the real world. We conducted internal verification of one thousand resamplings of the modeling group data using the bootstrap method. The AUC was 0.913, while the accuracy was 0.869 and the kappa consistency was 0.709. The combination prediction of liver metastasis in patients with CRC in the verification group had an AUC of 0.918, sensitivity of 85.0%, specificity of 95.6%, C-index of 0.918, and an H-L fitting curve with χ² = 0.586, P = 0.746.

CONCLUSION

The ALBI score, ALT level, and CEA level have a certain value in predicting liver metastasis in patients with CRC. These three criteria exhibit a high level of efficacy in forecasting liver metastases in patients diagnosed with CRC. The risk prediction model developed in this work shows great potential for practical application.

Key Words: Albumin-bilirubin, Carcinoembryonic antigen, Colorectal cancer, Tumor metastasis, Prediction model

Core Tip: This study investigated the application of the albumin bilirubin (ALBI) and liver function indices (aspartate aminotransferase, alanine aminotransferase, total bilirubin, etc.) combined with carcinoembryonic antigen (CEA) for the prediction of liver metastasis in patients with colorectal cancer (CRC). Through retrospective analysis of clinical data, we explored the associations of the ALBI and CEA level with liver metastasis in patients with CRC and established predictive models. The results of this study will provide clinicians with a simple and effective way to assess the risk of liver metastasis in patients with CRC and guide the development of treatment options that can help improve patient prognosis and survival.

INTRODUCTION

According to the statistics of National Cancer Center of China in 2023, there are approximately 4.08 million newly diagnosed colorectal cancer (CRC) patients and 196000 deaths in China each year, ranking second and fourth in malignant tumors, respectively[1-4]. The main reason for the poor prognosis of CRC patients is liver metastasis, which occurs in approximately 50% of CRC patients at the initial visit or immediately after surgery. The median survival time of patients who underwent radical resection of liver metastases was 35 months, and the 5-year survival rate was 30%-57%[5]. The median survival time of patients without radical resection of liver metastases was less than 7 months, and the 5-year survival rate was only 5%. Therefore, early detection of liver metastasis can significantly improve the long-term prognosis and reduce the mortality of CRC patients[6-8].

At present, imaging is the main method for detecting liver metastasis in CRC patients, but it has high equipment requirements, is strongly affected by the professional level of imaging doctors, and has high examination costs[9]. Therefore, a simple, economical, and objective detection method is urgently needed. Clinical studies have shown that abnormal changes in liver function indexes occur when malignant tumors develop liver metastases[10-12], so routine liver function index detection is expected to detect liver metastases early[13]. A new model for evaluating liver function, the albumin bilirubin (ALBI) score, which is composed of bilirubin and serum albumin (ALB) levels, has been proposed[14]. The liver function was divided into 3 levels: The higher the level was, the worse the liver function was. Recent studies have shown that ALBI scores correlate with the prognosis of a variety of cancers, including CRC with liver metastases, resectable gastric cancer, and resectable pancreatic cancer[15]. Carcinoembryonic antigen (CEA) is a widely recognized broad-spectrum tumor marker[16]. The detection of serum CEA levels before and after surgery can predict hepatic and occult metastasis of CRC. Several studies have shown that patients with a serum CEA concentration ≥ 15 μg/L after CRC surgery are at increased risk of distant metastasis[17-19].

Currently, there are no studies that have integrated the ALBI score with traditional liver function markers and CEA to predict CRC hepatic metastases. This study aimed to investigate the predictive value of a certain factor for CRC liver metastasis and to offer novel insights for the clinical identification and prediction of such metastasis.

MATERIALS AND METHODS

Study subjects and categorization

This retrospective analysis involved CRC patients who underwent surgery in our hospital between January 2018 and July 2023. The inclusion criteria for patients were as follows: (1) Had CRC confirmed through surgical pathology; (2) Had complete laboratory blood routine examination and tumor marker detection data; and (3) Had no prior treatment with chemoradiotherapy or hormone therapy before admission.

The exclusion criteria were as follows: (1) Had a follow-up duration of less than 24 months; (2) Had liver lesions in patients with liver metastasis confirmed as noncolorectal metastatic carcinoma by pathology or history combined with imaging; (3) Had a history of other malignant tumors or blood system diseases; and (4) Had primary malignant tumors at other sites.

According to the incidence of liver metastasis within 24 months after discharge, CRC patients were divided into a liver metastasis group and a nonliver metastasis group and randomly divided into a modeling group and a verification group at a ratio of 2:1. This research was approved by the Ethics Committee of Xiangya Hospital of Central South University (Approval number: 2023A-402), and patients who provided informed consent were excluded.

Data collection

The basic information of the patients, including age, sex, body mass index, primary tumor location, past history, prehospital treatment, preoperative laboratory examination, imaging results, and postoperative pathological diagnosis, was obtained through the electronic medical records system. The laboratory test results included aspartate aminotransferase/alanine aminotransferase transferase (AST/ALT), total protein (TP), ALB, globulin (GLO), ALB/GLO (A/G), gamma-glutamyltransferase (GGT), alkaline phosphatase (ALP), alpha-fetoprotein (AFP), CEA, carbohydrate antigen 125 (CA125), and CA19-9 levels. The ALBI score was calculated based on the results of laboratory tests. Follow-up information was obtained via electronic medical records or telephone consultations.

Sample size estimation

The sample size was estimated according to the formula: n = Z² × P × (1-P)/d². Previous studies reported that the incidence of postoperative liver metastasis in CRC patients was 15%-25%, that is, P = 15%, Z = 1.96, d = 0.05, and the minimum sample size of CRC patients was calculated to be 196.

Bias control analysis

The research subjects were selected strictly according to the inclusion and exclusion criteria. Data were entered and checked by two people to minimize subjective bias.

Statistical analysis

SPSS 26.0 (IBM Corp., Armonk, NY, United States) and R 4.2.2 software were used for the statistical analysis. The Wilcoxon rank sum test was used for comparisons between groups. Counting data are expressed as a percentage of cases, and a χ² test was used for comparisons between groups. In the modeling group, with the occurrence of liver metastasis as the dependent variable, least absolute shrinkage and selection operator (LASSO) regression variables were screened using the “glmnet” package, and the best λ value was selected through 10-fold cross-validation. The LASSO logistic regression model was constructed using the forward LR method, the receiver operating characteristic (ROC) curve was plotted, the area under the ROC curve (AUC) was calculated, its differentiation was evaluated, and the calibration curve and clinical decision curve were used to analyze its calibration and clinical benefit. At the same time, the bootstrap method was used to verify the modeling group internally. For the bilateral test, P < 0.05 indicated a statistically significant difference.

RESULTS

General patient data collection and analysis

Finally, 195 CRC patients who met the inclusion and exclusion criteria were enrolled in this study, and the enrollment process is shown in Figure 1. There were 113 males and 82 females. The average age was 60.05 ± 12.3 years, ranging from 26 to 90 years. The differences between the two groups in the primary tumor location, total bilirubin, direct bilirubin, ALT, AST, ALT/AST, TP, ALB, A/G, GGT, ALP, CEA, CA125, CA19-9, and ALBI scores were statistically significant (P < 0.05) (Table 1).

Figure 1 Patient inclusion and randomization flow charts.

Table 1 Baseline data between liver metastasis group and non-liver metastasis group (n = 195).

Index	Liver metastases group (n = 70)	Non-liver metastases group (n = 125)	P value
Age (mean ± SD, yr)	58.3 ± 12.5	61.7 ± 12.0	0.067
Gender, n (%)			0.865
Man	40 (57.1)	73 (58.4)
Female	30 (42.9)	52 (41.6)
Primary tumor location, n (%)			0.042
Colon	38 (54.3)	49 (39.2)
Rectum	32 (45.7)	76 (60.8)
BMI, M (P25, P75), kg/m2	22.5 (21.4, 24.7)	22.1 (20.2, 25.0)	0.28
TBIL, M (P25, P75), μmol/L	14.9 (11.6, 24.7)	12.8 (9.4, 16.9)	0.001
DBIL, M (P25, P75), μmol/L	7.2 (4.3, 10.6)	3.3 (2.3, 5.4)	< 0.001
IBIL, M (P25, P75), μmol/L	9.3 (5.2, 14.3)	8.8 (6.0, 12.1)	0.642
ALT, M (P25, P75), U/L	31.5 (16.0, 54.8)	17.0 (10.0, 25.0)	< 0.001
AST, M (P25, P75), U/L	40.0 (27.8, 87.3)	20.0 (16.0, 25.5)	< 0.001
AST/ALT, M (P25, P75)	1.6 (1.1, 2.2)	1.4 (0.9, 1.7)	0.005
TP, M (P25, P75), g/L	65.1 (58.9, 70.6)	69.4 (64.3, 73.4)	0.001
ALB, M (P25, P75), g/L	36.3 (29.5, 40.2)	41.4 (38.3, 44.1)	< 0.001
GLO, M (P25, P75), g/L	29.3 (24.7, 33.5)	28.3 (24.0, 31.0)	0.124
A/G (mean ± SD)	1.2 ± 0.4	1.5 ± 0.3	< 0.001
GGT, M (P25, P75), U/L	59.0 (25.8, 148.0)	18.0 (13.0, 25.5)	< 0.001
ALP, M (P25, P75), U/L	122.5 (77.5, 211.0)	76.0 (63.0, 90.5)	< 0.001
AFP, M (P25, P75), μg/L	2.6 (1.6, 3.4)	2.5 (1.8, 3.2)	0.845
CEA, M (P25, P75), μg/L	79.5 (12.5, 476.0)	2.9 (1.8, 5.9)	< 0.001
CA125, M (P25, P75), U/mL	21.3 (11.0, 54.7)	11.1 (7.9, 16.7)	< 0.001
CA19-9, M (P25, P75), U/mL	85.5 (17.3, 801.7)	10.9 (7.5, 17.0)	< 0.001
ALBI score, M (P25, P75)	-2.3 (-2.7, -1.6)	-2.8 (-3.0, -2.6)	< 0.001

BMI: Body mass index; TBIL: Total bilirubin; DBIL: Direct bilirubin; IBIL: Indirect bilirubin; ALT: Alanine aminotransferase; AST: Aspartate aminotransferase; TP: Total protein; ALB: Albumin; GLO: Globulin; A/G: Albumin/globulin; GGT: Gamma-glutamyltransferase; ALP: Alkaline phosphatase; AFP: Alpha-fetoprotein; CEA: Carcinoembryonic antigen; CA125: Carbohydrate antigen 125; ALBI: Albumin bilirubin.

The patients were randomly divided into a modeling group (130 patients) and a verification group (65 patients) at a ratio of 2:1. In the modeling group, there were 50 patients with liver metastasis and 80 patients without liver metastasis. In the verification group, there were 20 patients with liver metastasis and 45 patients without liver metastasis. There was no significant difference in the basic data between the two groups (all P > 0.05), as shown in Table 2.

Table 2 Baseline data between modeling group and validation group (n = 195).

Index	Modeling group (n = 130)	Validation group (n = 65)	P value
Age (mean ± SD, yr)	60.3 ± 11.8	60.8 ± 31.2	0.772
Gender, n (%)			0.473
Man	73 (56.2)	40 (61.5)
Female	57 (43.8)	25 (38.5)
Primary tumor location, n (%)			0.222
Colon	62 (47.7)	25 (38.5)
Rectum	68 (52.3)	40 (61.5)
BMI, M (P25, P75), kg/m2	22.3 (20.3, 24.9)	22.0 (20.6, 25.0)	0.992
TBIL, M (P25, P75), μmol/L	13.3 (10.2, 18.0)	14.0 (9.6, 18.7)	0.825
DBIL, M (P25, P75), μmol/L	4.7 (2.8, 7.3)	3.9 (2.7, 7.1)	0.462
IBIL, M (P25, P75), μmol/L	8.7 (5.5, 11.9)	9.6 (5.6, 13.0)	0.412
ALT, M (P25, P75), U/L	17.5 (11.0, 31.0)	20.0 (11.0, 33.0)	0.518
AST, M (P25, P75), U/L	24.5 (17.0, 36.0)	24.0 (17.5, 38.0)	0.846
AST/ALT, M (P25, P75)	1.4 (0.9, 1.8)	1.4 (0.9, 1.9)	0.715
TP, M (P25, P75), g/L	68.3 (63.8, 72.9)	67.1 (61.8, 73.0)	0.348
ALB, M (P25, P75), g/L	40.1 (35.8, 43.5)	39.1 (36.8, 42.7)	0.666
GLO, M (P25, P75), g/L	28.7 (24.4, 32.4)	28.4 (24.0, 31.4)	0.556
A/G (mean ± SD)	1.4 ± 0.4	1.4 ± 0.3	0.991
GGT, M (P25, P75), U/L	22.5 (14.0, 52.3)	22.0 (14.5, 40.5)	0.828
ALP, M (P25, P75), U/L	80.0 (63.8, 111.0)	86.0 (67.0, 110.0)	0.606
AFP, M (P25, P75), μg/L	2.5 (1.7, 3.3)	2.7 (1.6,3.2)	0.607
CEA, M (P25, P75), μg/L	4.7 (2.0, 33.3)	5.3 (2.5, 35.3)	0.514
CA125, M (P25, P75), U/mL	12.6 (8.6, 23.2)	31.2 (9.1, 22.7)	0.792
CA19-9, M (P25, P75), U/mL	13.9 (8.1, 52.6)	15.2 (8.2, 52.4)	0.697
ALBI score, M (P25, P75)	-2.7 (-2.9, -2.3)	-2.7 (-2.9, -2.4)	0.522

BMI: Body mass index; TBIL: Total bilirubin; DBIL: Direct bilirubin; IBIL: Indirect bilirubin; ALT: Alanine aminotransferase; AST: Aspartate aminotransferase; TP: Total protein; ALB: Albumin; GLO: Globulin; A/G: Albumin/globulin; GGT: Gamma-glutamyltransferase; ALP: Alkaline phosphatase; AFP: Alpha-fetoprotein; CEA: Carcinoembryonic antigen; CA125: Carbohydrate antigen 125; ALBI: Albumin bilirubin.

Construction of the LASSO logistic regression prediction model

In the modeling group, variable screening was performed using LASSO regression with whether patients developed liver metastases as the dependent variable, and the optimal λ value was selected through 10-fold cross-validation (Figure 2). In this study, λmin was selected as the best λ value, LASSO regression and the forward LR method were applied for logistic analysis (Table 3). Based on the results of the statistical regression analysis, three variables, ALBI, ALT, and CEA, were included in the statistical regression prediction model.

Figure 2 Feature variable selection based on least absolute shrinkage and selection operator regression. A: Tenfold cross-validation chart; B: Shrinkage coefficient chart.

Table 3 Multivariate Logistic regression analysis of factors influence liver metastasis of colorectal cancer.

Variable	β	Error	Wald	Freedom	P value	OR	95%CI
							Lower limit	Upper limit
ALT (U/L)	0.036	0.017	4.734	1	0.030	1.037	1.004	1.071
CEA (μg/L)	0.025	0.009	8.017	1	0.005	1.025	1.008	1.043
ALBI score	2.087	0.588	12.588	1	0.000	8.062	2.545	25.540
Constant term	3.047	1.555	3.839	1	0.050	21.058	-

ALT: Alanine aminotransferase; CEA: Carcinoembryonic antigen; ALBI: Albumin bilirubin; OR: Odds ratio; CI: Confidence interval.

Internal evaluation and internal verification of the prediction models

In the modeling group, the AUC of the ALBI score and the combination of ALT and CEA for the prediction of liver metastasis in patients with CRC was 0.921 (Figure 3), the sensitivity was 78.0%, the specificity was 95.0% (Table 4). For the data of the modeling group, the internal verification of 1000 samples was carried out by the bootstrap method (Figures 4 and 5). The accuracy was 0.869, the kappa consistency was 0.709, and the AUC was 0.913. When the ALT, CEA, and ALBI scores were used to diagnose CRC liver metastases alone, the bottom of the curve of CEA was the largest (AUC = 0.897), and the combined performance of the three was the highest in diagnosing CRC liver metastases (Table 4). The combined prediction of the AUC of the verification group was 0.918 (Figure 3), the sensitivity of spirit was 85.0%, the specificity was 95.6% (Table 5), the C-index was 0.918, and the H-L fitting curve χ² = 0.586, P = 0.746.

Figure 3 Receiver operating characteristic curve of the model for predicting liver metastasis in patients with colorectal cancer. A: Modeling group; B: Verification group. ALT: Alanine aminotransferase; CEA: Carcinoembryonic antigen; ALBI: Albumin-bilirubin.

Figure 4 Calibration curves of the least absolute shrinkage and selection operator logistic regression model. A: Modeling group; B: Verification group.

Figure 5 Decision curve analysis of the least absolute shrinkage and selection operator logistic regression model. A: Modeling group; B: Verification group.

Table 4 Diagnostic efficacy between individual and combined detection of alanine aminotransferase, carcinoembryonic antigen, and albumin bilirubin score in modeling group.

Testing index	AUC	Sensitivity (%)	Specificity (%)	P value
ALT	0.704	58.0	85.0	< 0.001
CEA	0.897	84.0	87.5	< 0.00
ALBI score	0.825	84.0	72.5	< 0.001
ALT combined with CEA	0.896	80.0	912	< 0.00
ALT combined with ALBI score	0.858	80.0	78.7	< 0.001
CEA combined with ALBI score	0. 91	82.0	86.3	< 0.001
Combination of the three	0.921	78.0	95.0	< 0.001

ALT: Alanine aminotransferase; CEA: Carcinoembryonic antigen; ALBI: Albumin bilirubin; AUC: Area under the receiver operating characteristic curve.

Table 5 Diagnostic efficacy between individual and combined detection of alanine aminotransferase, carcinoembryonic antigen, and albumin bilirubin score in verification group.

Testing index	AUC	Sensitivity (%)	Specificity (%)	P value
ALT	0.774	75.0	73.3	< 0.001
CEA	0.864	75.0	88.9	< 0.001
ALBI score	0.659	45.0	97.8	0.042
ALT combined with CEA	0.916	85.0	95.6	< 0.001
ALT combined with ALBI score	0.784	60.0	91.1	< 0.001
CEA combined with ALBI score	0.834	70.0	100.0	< 0.001
Combination of the three	0.918	85.0	95.6	< 0.001

ALT: Alanine aminotransferase; CEA: Carcinoembryonic antigen; ALBI: Albumin bilirubin; AUC: Area under the receiver operating characteristic curve.

Visualization of risk prediction models

Based on the results of the statistical regression analysis, a column graph of liver metastasis in CRC was drawn, as shown in Figure 6. For example, if the ALBI score of a certain research object is -3.0, 100 μg/L CEA, and 50 U/L ALT are projected vertically to the scoring axis, and the obtained scores are added: 7 + 10 + 4 = 21. The corresponding position of 21 points on the total score axis was found, and the predicted risk value projected vertically down to the risk axis of CRC liver metastasis was approximately 0.70.

Figure 6 Nomogram prediction model for liver metastasis of colorectal cancer. ALT: Alanine aminotransferase; CEA: Carcinoembryonic antigen; ALBI: Albumin-bilirubin.

DISCUSSION

Based on the analysis of the patient’s ALBI score, conventional liver function indicators, tumor markers, etc., this study concluded that the ALBI score, ALT level, and CEA level were independent predictors of the occurrence of liver metastases in CRC patients. The prediction model was then established by statistical logistic regression[20-22]. H-L fitting curve: χ² = 0.586, P = 0.746. These results indicate that the combined prediction of CRC liver metastasis is effective, and the risk prediction model constructed by the three methods has good clinical application prospects.

Due to the small number and low concentration of tumor cells in CRC liver metastases, routine imaging examinations (such as computed tomography, magnetic resonance imaging, etc.) cannot make an early diagnosis of CRC liver metastases[23]. Therefore, it is urgent to find a better method to accurately identify CRC liver metastases at an early stage[24-26]. CEA is present in malignant tumors of the gastrointestinal tract and pancreatic endoderm-derived epithelium, is overexpressed in CRC patients, and is distributed throughout the cell membrane. Previous studies have shown that high CEA expression is significantly correlated with CRC metastasis[27-29]. Serum CEA increases 6 months before imaging findings of liver metastases, and serum CEA, CA19-9, and AFP are significantly greater in CRC patients with liver metastases than in those without liver metastases[30]. The mechanism underlying the correlation between high CEA expression and CRC liver metastasis is as follows: CEA reduces the death of cancer cells in the blood by inhibiting apoptosis. CEA binds to the Kupffer cell receptor protein and changes the liver microenvironment, which is conducive to the survival of cancer cells. CEA upregulates cell adhesion molecules associated with metastasis. Recent studies have suggested that CEA reexamination every 2-3 months after CRC surgery is helpful for the early detection of liver metastasis[31-33]. Another study reported that the sensitivity of serum CEA in the diagnosis of CRC liver metastasis was only 36.5%[34], which made it difficult to meet the requirements of early detection of liver metastasis. The level of CEA in advanced CRC patients and other gastrointestinal malignancies is significantly increased, but early CEA detection alone may result in false-positive and false-negative results in the diagnosis of CRC liver metastases, and the diagnostic accuracy is low[35-37].

Following the introduction of the ALBI score, several studies have applied it to the prediction of CRC liver metastases, resectable gastric cancer, and resectable pancreatic cancer[38]. Compared with the Child-Pugh score, the most commonly used liver function evaluation index in clinical practice, the ALBI measures two subjective indicators of hepatic encephalopathy and ascites. In addition, data acquisition is more convenient[39]. However, the ALBI score contains only two indicators, there is no upper limit effect, and the ALBI score is biased if the patient has hypoproteinaemia or hyperbilirubinemia (such as obstructive jaundice)[40]. In addition, this score was originally proposed for liver cancer patients without considering the influence of other causes, and further studies are needed to confirm its practicality and accuracy for detecting liver function abnormalities caused by other causes. The above studies suggest that there are certain limitations in predicting CRC liver metastasis based only on tumor markers, liver function, and other single indicators, which may easily lead to misdiagnosis and missed diagnosis[41]. Therefore, in this study, the ALBI score was combined with conventional liver function indicators and CEA to detect CRC liver metastasis at an early stage.

The C-index of the CRC liver metastasis risk prediction model established in this study based on the ALBI score, ALT, and CEA was 0.921 and 0.918, respectively, in the modeling group and verification group, and the correlation between the two curves in the calibration chart was good and consistent[42]. The clinical decision curve also showed good clinical application value. According to the analysis, the AUCs of the ALBI score, ALT level, CEA level, and their combination for predicting CRC liver metastasis were 0.825, 0.704, 0.897, and 0.921, respectively. With the introduction of the ALBI score, several studies have applied it to the prediction of CRC liver metastasis, resectable gastric cancer, and resectable pancreatic cancer[43]. Compared with the Child-Pugh score, which is the most commonly used liver function indicator in clinical practice, the ALBI removes two subjective indicators, namely, hepatic encephalopathy and ascites, and data acquisition is more convenient. However, the ALBI score contains only two indicators, there is no upper limit effect, and the ALBI score is biased if the patient has hypoproteinaemia or hyperbilirubinemia (such as obstructive jaundice). In addition, this score was originally proposed based on liver cancer patients without considering the influence of other causes, and further studies are needed to confirm its practicality and accuracy for detecting liver function abnormalities caused by other causes[44]. The above studies suggest that there are certain limitations in predicting CRC liver metastasis based only on tumor markers, liver function, and other single indicators, which may easily lead to misdiagnosis and missed diagnosis. Therefore, in this study, the ALBI score was combined with conventional liver function indicators and CEA to detect CRC liver metastasis at an early stage.

This study has the following shortcomings: (1) As a single-center retrospective study, there may be bias in case selection; and (2) Because of the lack of multicenter data for external validation of the model, large sample sizes and multicenter clinical data are still needed to improve the validity and reliability of the model.

CONCLUSION

In summary, the ALBI score combined with the ALT and CEA levels has high specificity and accuracy in predicting CRC liver metastasis and will be valuable for improving the diagnosis and treatment of CRC liver metastasis patients.