Beta2-Microglobulin as Predictive Biomarkers in the Prognosis of Hepatocellular Carcinoma and Development of a New Nomogram

Introduction

Hepatocellular carcinoma (HCC) is a widespread solid malignancy and the third leading cause of cancer-related deaths worldwide.^1,2 Its estimated annual incidence is projected to exceed 1,000,000 cases by 2025.³ Resection is the most common cure for HCC worldwide, notably, according to the Chinese official guidelines and the clinical experience of our center and other country centers, intermediate and advanced-stage HCC patients could undergo resection, which is associated with good survival.^4–7

However, the high recurrence and metastasis rate of HCC after operation is still a big challenge for patients’ survival, and the prognosis remains poor.^8,9 The inherent heterogeneity of HCC poses significant challenges in accurately predicting prognoses and devising suitable interventions for affected individuals.¹⁰ Thus, it is imperative to construct a model identifying high-risk patients accurately for adjuvant therapy to improve overall survival (OS).

The Barcelona Clinical Liver Cancer (BCLC) staging system stands as the predominant clinical framework employed for prognosticating HCC outcomes.¹¹ Nonetheless, within tertiary medical centers, a notable proportion of patients, up to 50%, veer away from the treatment recommendations outlined by the BCLC.¹² Alpha-fetoprotein (AFP) is the most extensively used biomarker for determining the prognosis of HCC.¹³ Unfortunately, AFP is negative in approximately 30–40% of HCC patients.^14,15 Therefore, there exists a critical need to identify a practical and reliable prognostic model that can offer more precise predictions of individual HCC patient outcomes.

Alpha1-microglobulin (α1MG) is a ubiquitous protein, that has reductase activity and the binding characteristics of free radicals and heme,^16,17 but its direct involvement in tumor-related processes remains less explored. β2MG, as a member of the major histocompatibility complex class I (MHC-I) and a vital component of the human immune system, assumes a pivotal role in maintaining the physiological equilibrium of the body and exerting its influence on diverse processes occurring in the glomerulus and other nucleated cells.^18,19 The role of β2MG in cancer is diverse, with effects primarily mediated through the immune system, as well as in cell proliferation, invasion and metastasis.^20–22 The expression of β2MG increased in the progression of human prostate cancer, breast cancer, and renal cell carcinoma, bladder cancer.^23–27 Additionally, it has been observed that an elevated serum β2MG level is not only a significant predictor of OS in patients with plasmacytoma, but also a marker of central nervous system metastasis in small cell lung cancer.^28,29 However, there has been no reported evidence regarding the association between α1MG or β2MG and HCC to date.

This study reveals a strong relationship between increased serum β2MG levels and a poor prognosis in HCC patients for the first time. By incorporating β2MG and other readily accessible clinicopathological features, we successfully developed and validated a novel prognostic nomogram specifically tailored for HCC patients undergoing hepatectomy. This nomogram not only simplifies data acquisition but also exhibits robust predictive efficacy.

Materials and Methods

Population and Information Gathering

The study encompassed a period spanning from October 2013 to December 2021. 1976 patients diagnosed with HCC underwent curative resection at the Guangxi Medical University Cancer Hospital (Nanning, China). The following inclusion and exclusion standards were applied to select participants for this study. The inclusion standards for this study were as follows: (1) individuals who had not undergone any prior anti-cancer treatment, (2) no other concurrent malignancies, and (3) comprehensive laboratory, pathological, and follow-up data are available. On the other hand, the exclusion standards consisted of (1) individuals who had received relevant antitumor therapy such as chemotherapy or radiotherapy (n=125), (2) absence of a clear and definitive pathological diagnosis (n=68), (3) presence of malignant tumors other than hepatocellular carcinoma (HCC) (n=132), (4) cases of recurrent HCC (n=115), and (5) comprehensive laboratory, pathological, and follow-up data are not available (n=445). Finally, this study retrospectively covered 1091 patients in total.

Random assignment divided the patients into two groups: the training cohort (n = 767) and the validation cohort (n = 324). Data collection was by two independent investigators, QML and ZWJ, with validation by a third investigator, MD. General information like age and sex, serum AFP, CA199, α1MG and β2MG levels, and histopathological information like node number, differentiation, and vascular infiltration were all clinical factors that were gathered. The flow of the research program is depicted in Figure 1.

Figure 1 The flow of study. Abbreviations: HCC, hepatocellular carcinoma; HR, hazard ratio; CI, confidence interval; β2MG, beta2-microglobulin; ROC, receiver operating characteristic.

The Barcelona Clinic Liver Cancer (BCLC) staging system was used to determine the tumor stage.^30,31 Microvascular invasion (MVI) is defined as the presence of nests of over 50 cells within the endothelial vessel lumen when examined microscopically.³²

Follow-Up

Patients were followed up, either by phone or through an outpatient surveillance system, to determine their condition or the date of death if the patient had already died. Patients underwent regular monitoring at specific intervals following their surgery. The OS was determined by calculating the time between the surgical intervention and either the date of death or the date of the last follow-up. The last follow-up date of this study was in September 2022.

Statistical Analysis

The statistical analyses were performed using IBM SPSS Statistics software version 26.0 and R version 4.2.1. To determine the optimal cutoff values for α1MG and β2MG, we conducted receiver operating characteristic (ROC) curve analysis. To compare categorical variables between the training and validation cohorts, the chi-square test was utilized. Survival curves were plotted using the Kaplan-Meier method and compared using the Log rank test. In the subsequent multivariable Cox regression analysis, only variables with a significance level of p < 0.05 in the univariate analysis were included. Independent prognostic factors were determined based on their statistical significance (p < 0.05) in the multivariable Cox regression analysis. Based on the multivariable analysis-determined independent prognostic factors, we constructed a predictive nomogram using the rms package in R to predict individual OS at 1, 3, and 5 years. Calibration curves were produced to visually evaluate the congruence between the actual OS outcomes and the OS predictions derived from the nomogram. DCA analysis was performed via the R package “stdca.R”. The predictive accuracy and discriminative ability of the nomogram were assessed through various measures, including the C-index, ROC curves, decision curves, and DCA analysis, and compared with other systems. Statistical tests were conducted utilizing a two-sided approach, wherein a significance level of p < 0.05 denoted statistical significance.

Results

Patient Characteristics

Patients included in the analysis were randomly divided into two groups in a 7:3 ratio, with 767 from the training cohort and 324 from the validation cohort. The baseline characteristics of all patients recruited are summarized in Table 1. Most of the enrollees were male (n = 956; 87.63%), and 754 (69.11%) patients were aged ≥45 years old. No significant differences were observed between the training and validation cohorts concerning age, BCLC stage, liver cirrhosis, Edmondson grade, node number, MVI, CEA, AFP, α1MG, and β2MG (p = 0.062–0.897), except for sex and CA199 (p = 0.002–0.036).

Table 1 Patient Demographics and Clinical Characteristics

A Higher Concentration of Microglobulin is Related to the Poor Prognosis of HCC

In order to evaluate the prognostic value of microglobulin in HCC, we compared the survival differences among patients with different concentrations of microglobulin by drawing the Kaplan-Meier survival curve. The results showed that in the training cohort, the OS of the ɑ1MG≥19.750mg/L group was significantly shortened (p = 0.027, Figure 2A), and the β2MG≥2.395mg/L group was also significantly correlated with the poor overall survival (p < 0.001, Figure 2B). In the validation cohort, there was no significant difference in OS between the two groups of α1MG (p = 0.168, Figure 2C), and the β2MG≥2.395mg/L group was still significantly related to the poor OS (p = 0.012, Figure 2D). Kaplan -Meier survival curve of ɑ1MG showed different results in training and validation cohorts, which may be caused by insufficient samples in the validation cohort.

Figure 2 Kaplan-Meier curve of microglobulin. (A) Kaplan-Meier curve of α1MG in the training cohort. (B) Kaplan-Meier curve of β2MG in the training cohort. (C) Kaplan-Meier curve of α1MG in the validation cohort. (D) Kaplan-Meier curve of β2MG in the validation cohort. Abbreviations: α1MG, alpha1- macroglobulin; β2MG, beta2-microglobulin.

Univariate Analysis and Multivariate Analysis

Besides β2MG, the univariate Cox regression analysis also revealed that BCLC stage [p < 0.001, hazard ratio (HR)=2.862], Edmondson grade (p < 0.001, HR=1.751), node number (p = 0.001, HR=1.556), MVI (p < 0.001, HR=2.485), AFP (p < 0.001, HR=1.622), CA199 (p = 0.010, HR=1.451), α1MG (p = 0.028, HR=1.282), and β2MG (p < 0.001, HR=1.595) were all identified as risk factors for the OS of HCC patients. Furthermore, the multivariate Cox regression analysis confirmed that BCLC stage (p < 0.001, HR=2.185), Edmondson grade (p = 0.001, HR=1.477), MVI (p < 0.001, HR=1.906), CA199 (p = 0.015, HR=1.428), and β2MG (p = 0.001, HR=1.444) were independent risk factors for the OS of HCC patients. The study’s data underwent univariate and multivariate Cox regression analyses, and the results are presented in Table 2.

Table 2 Univariate and Multivariable Analysis for Overall Survival of the Training Cohort

Prognostic Nomogram for OS

Based on multivariate Cox regression analysis in the training cohort, a nomogram was constructed for OS prediction and involved all independent prognostic factors that were mentioned above in the training cohort (Figure 3A) and validation cohort (Figure 3B). This nomogram aimed to predict the OS of HCC patients who underwent resection at 1, 3, and 5 years. The constructed model demonstrated a high level of accuracy in predicting the OS rate of HCC, as indicated by a C-index of 0.712 (95% CI, 0.698–0.726). The performance of the nomogram was evaluated using calibration curves and AUC values. Figure 4A and B display calibration curves for predicting 1-year, 3-year, and 5-year OS, indicating good calibration. In the training cohort, the area under the receiver operating characteristic curve (AUC) values for predicting 1-year, 3-year, and 5-year OS were 0.763, 0.732, and 0.710, respectively (Figure 4C). In the validation cohort, the corresponding AUC values for predicting 1-year, 3-year, and 5-year OS were 0.811, 0.714, and 0.711, respectively (Figure 4D). Both the calibration curves and ROC curves demonstrated effective prediction of the 1-, 3-, and 5-year OS rates. In conclusion, the nomogram’s accuracy, as indicated by the C-index and AUC values, suggests its potential as a useful tool in clinical practice for prognostic assessment in HCC patients.

Figure 3 Nomograms for the prediction of 1-, 3- and 5-year OS. (A) The nomogram of the training cohort; (B) The nomogram of the validation cohort. Abbreviations: MVI, microvascular invasion; CA199, carbohydrate antigen 199; β2MG, beta2-microglobulin.

Figure 4 Evaluation of the performance of the nomograms in the training cohort and validation cohort. (A) Calibration curves of the nomogram predicting 1-, 3- and 5-year OS in the training cohort. (B) Calibration curves of the nomogram predicting 1-, 3- and 5-year OS in the validation cohort. (C) ROC curves of the nomogram predicting 1-, 3- and 5-year OS in the training cohort. (D) ROC curves of the nomogram predicting 1-, 3- and 5-year OS in the validation cohort. Abbreviations: OS, overall survival; ROC, receiver operating characteristic; AUC, area under the curve.

The Decision Curve Analysis

The decision curve analysis was conducted to assess the net benefit of the models in predicting OS of HCC patients. Based on Figure 5A, it is clear that the nomogram model (represented by the red line) exhibited the highest net benefit when compared to the BCLC stage model (blue line). The higher position of the red line (nomogram) compared to the blue line (BCLC stage) indicates that the nomogram model offers a greater net benefit across a range of threshold probabilities. This suggests that the nomogram model is more effective in predicting OS and provides better decision-making support for treatment strategies in HCC patients. The results of the analysis indicated that the nomogram outperformed not only the BCLC stage model but also the Edmondson grade model and the combined BCLC stage and Edmondson grade model in predicting OS. In the validation cohort, this superiority was seen across a larger range of threshold probabilities (Figure 5B). The decision curves demonstrated the higher net clinical benefit associated with the nomogram, further supporting its superior predictive performance in determining the OS of HCC patients.

Figure 5 Decision curve analysis for 5-year survival predictions. (A) The decision curve of the training cohort; (B) The decision curve of the validation cohort.

Risk Stratification of OS

Cut-off values for total points were determined using the X-tile program to stratify patients in both the training and validation cohorts into three risk groups based on OS.³³ The three risk categories were defined as follows: low-risk group (3.00–93.00 points), intermediate-risk group (96.00–125.00 points), and high-risk group (127.00–160.00 points). In the training cohort, the low-risk group had an OS rate of 80.4%, the intermediate-risk group had an OS rate of 50.4%, and the high-risk group had an OS rate of 21.3%. This difference in OS rates among the three risk groups was statistically significant (p < 0.001), as shown in Figure 6A. Similarly, in the validation cohort, the low-risk group had an OS rate of 72.5%, the intermediate-risk group had an OS rate of 49.7%, and the high-risk group had an OS rate of 20.8%. Again, this difference in OS rates among the risk groups was statistically significant (p < 0.001), as depicted in Figure 6B. In both the training and validation cohorts, the classification scheme based on total points accurately distinguished survival outcomes between the three risk categories. Furthermore, the risk stratification provided by this scheme holds significant value in predicting and monitoring the prognosis of patients with HCC.

Figure 6 Survival graph for all three groups based on the predictor from the nomogram model. (A) Kaplan-Meier curve in the training cohort. (B) Kaplan-Meier curve in the validation cohort.

Discussion

This study presents, for the very first time, a remarkable correlation between increased serum β2MG levels and an adverse prognosis in individuals diagnosed with HCC. Through the integration of β2MG and other easily obtainable clinicopathological characteristics, we have effectively constructed and validated a pioneering prognostic nomogram designed specifically for HCC patients undergoing hepatectomy. This nomogram not only streamlines data collection but also demonstrates exceptional predictive accuracy.

Currently, the prognostic value of serum β2MG has been extensively studied in various types of lymphoproliferative disorders, such as non-gastric mucosa-associated lymphoid tissue lymphoma,³⁴ mucosa-associated lymphoid tissue lymphoma,³⁵ extranodal natural killer/T-cell lymphoma,^36,37 Hodgkin’s lymphoma,³⁸ follicular lymphoma,³⁹ diffuse large B-cell lymphoma,^40,41 and multiple myeloma.⁴² The robustness of serum β2MG as a prognostic indicator has led to its incorporation into prognostic staging systems for multiple myeloma and follicular lymphoma.^39,42 However, the prognostic value of serum β2MG in HCC has not been reported. Our study is the first to reveal a correlation between elevated serum β2MG levels and poor prognosis in HCC patients, suggesting that β2MG holds promise as a prognostic marker for HCC. β2MG plays a crucial role in tumor progression by exerting significant influence on the immune microenvironment, maintaining stem cell-like tumor populations, enhancing migration and invasion capabilities, promoting proliferation, and acting as a signaling and growth-promoting factor for bone metastasis.^21–23,43 Although β2MG has been associated with human tumors, its significance in HCC has not been fully investigated. Therefore, further research is warranted to elucidate the precise mechanisms by which β2MG promotes HCC.

We constructed a nomogram based on β2MG, CA199, and clinical characteristics to predict the prognosis of HCC patients. Taking the BCLC stage, Edmondson grade, and the combination of the BCLC stage with Edmondson grade as comparatives, the nomogram exhibited a higher C-index (0.712) compared to the BCLC stage (0.660), Edmondson grade (0.579), and the combination of BCLC stage with Edmondson grade (0.681). The results suggested the nomogram had a superior predictive performance than the BCLC stage, Edmondson grade, and the combination of the BCLC stage with Edmondson grade. Such a superior predictive performance attributed to hurdle the limitations of the BCLC staging system, making the nomogram an effective tool for determining disease prognosis benefitting improving the OS.⁴⁴ Moreover, we compared our nomogram with other similar models for HCC patients published in the previous literature. These models include a nomogram based on NLR (neutrophil to lymphocyte ratio), GMWG (the geometric mean of GGT and white WBC count), and tumor size with a C-index value of 0.70,⁴⁵ a prognostic model developed by parameters of Aspartate Transaminase/Alanine Transaminase Ratio and clinical features with a C-index value of 0.680,⁴⁶ and a prognostic model constructed by platelet count and clinical features with a C-index value of 0.660.⁴⁷ Our model achieved a C-index of 0.712 (95% CI, 0.698–0.726), indicating higher predictive accuracy compared to these previously published models. In recent years, several scholars, including Long et al,⁴⁸ Tang et al,⁴⁹ and Chen et al,⁵⁰ have developed prognostic nomograms for predicting OS in HCC patients using costly genomic data. However, it is worth noting that their respective nomogram models exhibited lower C-index values and incurred higher costs compared to our developed nomogram. Specifically, the C-index values reported in their studies were 0.69, 0.66, and 0.703, respectively, whereas our nomogram outperformed them in terms of predictive accuracy and discriminative ability. These results indicated that our predictive model based on β2MG was able to reliably predict the prognosis of patients with HCC after liver resection using inexpensive and easily accessible variables. Additionally, our strategy successfully divided HCC patients into high-risk, intermediate-risk, and low-risk groups, leading to personalized adjuvant therapy for high-risk patients to achieve improved survival outcomes.

In conclusion, the exceptional predictive performance above all indicated we successfully developed and validated a groundbreaking prognostic nomogram specifically designed for HCC patients undergoing liver resection by integrating β2MG, CA199, and pathological features. Furthermore, this study had the key advantage of introducing a novel prognostic indicator, ie, β2MG, for predicting HCC. Given the ease of obtaining β2MG and CA199 levels through routine preoperative laboratory tests, our nomogram had the potential for daily practice in the clinic, providing a useful tool for clinical decision-making and personalized adjuvant therapy management for HCC patients. However, our study also had some limitations. For example, due to its retrospective nature, selection bias was inevitable. Besides, our study only assessed the prognostic value of the nomogram for predicting OS in HCC patients and did not evaluate its impact on predicting disease-free survival (DFS). The combination of OS and DFS in the nomogram would have better clinical utility. Additionally, the lack of validation using an independent dataset poses a limitation to our study, as it hinders the confirmation of the external validity of our findings. Therefore, the prognostic nomogram we constructed will need to be validated in the future on a larger scale and in multicenters to be more consistent with clinical application.

Conclusion

In summary, our study not only establishes a significant association between elevated serum β2MG levels and an unfavorable prognosis in HCC patients but also presents a novel prognostic nomogram tailored specifically for HCC patients undergoing hepatectomy. This nomogram not only simplifies data acquisition but also demonstrates exceptional predictive efficacy. Our findings hold substantial promise for advancing clinical practice and optimizing patient care in the field of HCC management. However, the mechanism by which β2MG promotes the progression of HCC and multi-center validation still requires further research.