A novel prognostic model based on AFP, tumor burden score and Albumin-Bilirubin grade for patients with hepatocellular carcinoma undergoing radiofrequency ablation

Abstract

Purpose

The aim of this study was to develop prognostic scores, including the tumor burden score (TBS) and albumin-bilirubin (ALBI) grade, for evaluating the outcomes of hepatocellular carcinoma (HCC) patients after radiofrequency ablation (RFA).

Materials and Methods

This retrospective study enrolled treatment-naïve HCC patients with BCLC 0-A who underwent RFA between January 2009 and December 2019. Regular follow-up was conducted after RFA to determine progression-free survival (PFS) and overall survival (OS). The patients were randomly allocated to the training or validation datasets in a 1:1 ratio. Preoperative prognostic scores were developed based on the results of multivariate analysis. The discriminatory ability of the scores was assessed using time-dependent AUC and compared with other models.

Results

Serum alpha-fetoprotein (AFP) level and TBS were identified as independent prognostic factors for PFS, while serum AFP, TBS, and ALBI were identified as independent prognostic factors for OS in HCC patients after RFA. The time-dependent AUCs of the AFP-TBS score for the 1-, 3-, and 5-year PFS were 0.651, 0.667, and 0.620, respectively, in the training set, and 0.657, 0.687, and 0.704, respectively, in the validation set. For the 1-, 3-, and 5-year OS, the time-dependent AUCs were 0.680, 0.712, and 0.666, respectively, in the training set, and 0.712, 0.706 and 0.726 in the validation set for the AFP-TBS-ALBI score (ATA). The C-indices and AIC demonstrated that the scores provided better clinical benefits compared to other models.

Conclusion

The ATA/AT score, derived from clinical and objective laboratory variables, can assist in individually predicting the prognosis of HCC patients undergoing curative RFA.

Keywords:

• Tumor burden
• alpha fetoprotein
• prognosis
• hepatocellular carcinoma
• radiofrequency ablation

1. Introduction

Hepatocellular carcinoma (HCC), which accounts for approximately 75% of all primary liver cancers, was the sixth most commonly diagnosed cancer and the third leading cause of cancer-related deaths worldwide in 2018 [1]. Advances in diagnostic imaging and the widespread application of tumor screening in high-risk populations have improved the early detection of HCC. According to HCC practice guidelines, minimally invasive ablation is recommended as a curative treatment for very early or early-stage HCC, alongside surgical resection and liver transplantation, as it can preserve the residual normal liver tissue [2,3]. Radiofrequency ablation (RFA), the most commonly used thermal therapy, is typically indicated for patients with HCC within the Milan criteria [4].

Although the Barcelona Clinical Liver Cancer (BCLC) stage is a well-recognized categorization system for HCC, it may not sufficiently predict prognosis, despite its usefulness in stratifying at-risk patients [5]. Previous studies have demonstrated heterogeneous outcomes among patients in the same BCLC stage, potentially due to differences in liver function and tumor burden assessment within the system [6,7]. Furthermore, while the local recurrence rate of RFA has decreased with technological advancements, the rates of intra- or extra-hepatic recurrence after five years following RFA for HCC remain high, exceeding 70% [8]. Therefore, selecting suitable candidates for ablation is still challenging. Improved risk stratification schemes are needed to enhance patient selection and identify individuals with a high risk of recurrence, enabling the development of individualized surveillance strategies after HCC ablation.

Pathological markers are valuable in predicting the outcome of HCC, but obtaining such information before RFA is difficult [9–11]. Several preoperative prognostic factors, including tumor burden profiles, liver function, and biomarkers, have been identified as important predictors of outcomes among HCC patients. Recently, new indicators, such as, the tumor burden score (TBS) and albumin-bilirubin (ALBI) grade, have been developed to quantitatively evaluate tumor burden and liver reserve, respectively [12,13]. Previous studies have widely verified the prognostic value of TBS or ALBI in HCC [14–17]. However, it remains unclear whether combining these indicators can further improve the prognostic performance in HCC patients undergoing RFA.

Therefore, the objective of this study was to develop a categorized score based on pretreatment indicators with prognostic significance before RFA. Additionally, a comparison was made between the categorized score and other predictive models including BCLC stage, Cancer Liver Italian Program (CLIP) score, Japan Integrated System (JIS) staging systems, the Chinese University Prognostic Index (CUPI), and the Okuda staging system, to determine if the categorized score can effectively predict individual survival and stratify patient risk among ideal candidates for RFA [18–21].

2. Materials and methods

This retrospective study (Ethics approval number: [2023]507) was approved by the Institutional Review Board of the First Affiliated Hospital of Sun Yat-sen University. The requirement for informed consent from patients was waived because of the retrospective nature of the study.

2.1. Study population and patients’ eligibility

A total of 848 treatment-naïve HCC patients who underwent curative‐intent RFA at the First Affiliated Hospital of Sun Yat-sen University, between January 2009 and December 2019 were retrospectively studied. Patients who met the following criteria were included in this study:(1) a solitary nodule less than 7.0 cm in diameter or up to three nodules 3.0 cm in diameter or smaller (within BCLC 0-A) [22]; (2) no pre-anti-tumor therapy; (3) no liver transplantation afterwards; and (4) availability of complete follow-up data. Eligible patients were randomly assigned to the training or validation datasets using a computer-generated randomized number at a 1:1 ratio (Figure 1).

Figure 1. Patient inclusion flowchart.

2.2. Variables, definitions and follow-up

The patients’ baseline demographics, virological data, the extent of tumor involvement (tumor size and number), the severity of the liver injury, and serum biochemistry were comprehensively recorded. HCC was diagnosed by either imaging evaluation according to the AASLD or EASL guidelines [2,3]. The tumor number and size were assessed using computerized tomography (CT) or magnetic resonance imaging (MRI) before ablation. According to BCLC staging in 2022, a solitary HCC smaller than 2 cm without vascular invasion or extrahepatic spread is classified as BCLC-0. A solitary HCC irrespective of size or as a multifocal HCC with up to 3 nodules (none of them >3 cm), without macrovascular invasion, extrahepatic spread is classified as BCLC-A [5]. Diagnosis and treatment were discussed by the multidisciplinary board of our hospital. The RFA procedure and device of RFA have been previously described in detail [23,24].

The TBS was defined as the distance from the origin of a Cartesian plane and comprised two variables: maximum tumor size (x-axis) and number of tumors (y-axis), so that TBS²=(maximum tumor diameter)² + (number of tumors)², as previously described [12]. TBS was calculated based on radiographic tumor characteristics, as imaging-based TBS was comparable to pathology-based TBS [25]. In the case of multiple nodules, the tumor size was defined as the size of the largest lesion. The cutoff values of TBS were selected to divide the cohort into three groups: low TBS (score <2.42), medium TBS (score between 2.42 to 3.07), and high TBS (score >3.07) (Calculator for TBS in supplemental Excel).

The ALBI grade was calculated according to the following formula = 0.66 × log₁₀bilirubin (μmol/L) − 0.085 × albumin (g/L) [13]. The ALBI score was divided into three groups as follows: grade 1 (score ≤ −2.60), grade 2 (score > −2.60 and ≤–1.39), and grade 3 (score > −1.39) (Calculator for ALBI in supplemental Excel).

Dynamic CT or MRI was done one month after all tumors were ablated by RFA. All patients were followed up every 3 months during the first 2 years after ablation and every 3-6 months thereafter. Every patient underwent a routine liver function review, serum alpha-fetoprotein (AFP) analysis, and imaging studies, including contrast-enhanced ultrasound (CEUS), CT and/or MRI during follow-up. When a recurrence was suspected on surveillance ultrasound, it was further confirmed using enhanced CT or MRI.

The primary outcome of interest was overall survival (OS), defined as the time interval between the date of HCC ablation and the date of death or the last follow-up evaluation (January 2022). The secondary outcome was progression-free survival (PFS), defined as the time between ablation of HCC and the date of recurrence or the last follow-up.

2.3. Statistical analysis

Quantitative variables were presented as medians (ranges) or means (standard deviation), while categorical variables were shown as counts (%). No missing data were reported for the investigated covariates. Statistical analyses were conducted using appropriate tests such as Student’s t test, Mann–Whitney U tests, chi-squared test, or Fisher’s exact probability test to compare the training and validation cohorts. Variables that showed potential significance in the univariate analysis were selected for multivariate analysis using the Cox proportional hazards model to identify independent predictors of PFS and OS. The assumption of proportional hazards was checked using a time-dependent Cox regression model. Prognostic scores for risk stratification were developed based on the aforementioned analyses. The distribution of both PFS and OS in the validation cohort was depicted using the Kaplan–Meier method, and differences in survival rates were compared using the log-rank test. The discrimination and calibration of the prognostic scores were evaluated using the concordance index (C-index), area under the time-dependent receiver operating characteristic curve (time-dependent AUC), and calibration plots. Prognostic scores were further assessed with decision curve analysis (DCA). Comparisons were made between the prognostic scores for PFS and OS and existing scoring systems such as BCLC, CLIP, JIS, CUPI, and Okuda using the C-index and Akaike information criterion (AIC) [26]. A higher C-index indicates greater accuracy in prognostic predictions, while the lowest AIC value indicates better discriminatory ability. All p-values were two-sided, with p < 0.05 considered statistically significant. Statistical analyses were performed using SPSS version 23 (SPSS Inc., Chicago, IL, USA) and R 4.2.1 (http://www.R-project.org).

4. Results

4.1. Baseline characteristics

Finally, 716 patients with HCC were enrolled, the majority of whom (606/716, 84.6%) were male, with a median age of 57 years (range, 20–90 years). Among the enrolled patients, 85.2% (610/716) had single HCC, with tumor size ranging from 0.6 cm to 6.2 cm. The median TBS was 2.69, with 36.3%, 30.7%, and 33.0% falling into the low, medium, and high TBS categories, respectively. A total of 48.7% (349/716) of patients had elevated serum AFP levels, among whom 14.2% (102/716) had serum AFP levels ≥400 ng/mL. Child-Pugh class A was observed in 95.0% (680/716) of the patients, and 46.2% (331/716), 52.5% (376/716) and 1.3% (9/716) of the patients were classified into ALBI grades 1, 2, and 3, respectively. The distribution of patients according to different cancer staging systems is presented in Table 1. There were no significant baseline differences observed between the training and validation cohorts (all p > 0.05). The median PFS after RFA for the enrolled patients was 31 months (range, 1–153 months), whereas the median OS after RFA was 54 months (range, 1–153 months).

Table 1. Baseline demographics and clinical characteristics in 716 patients.

Variable	Total cohort (n = 716) N or Mean (range)	Training cohort (n = 358) N or Median (range)	Validation cohort (n = 358) N or Median (range)	p-value
Gender, Male/Femal	606/110	296/62	310/48	0.147
Age (years), <55, ≥55	303/413	141/217	162/196	0.112
HBsAg, P/N	619/97	312/46	307/51	0.585
HCVAb, P/N	30/686	14/344	16/342	0.709
HBV DNA (IU/mL), <100, ≥100	465/251	237/121	228/130	0.481
ALB (IU/L)	39 (21.9–53.6)	39 (21.9–49)	39 (22–53.6)	0.877
ALT (IU/L)	30 (3.5–204)	29 (7–204)	31 (3.5-194)	0.390
AST (IU/L)	33 (12–272)	32 (12–272)	33 (13–247)	0.265
TBIL (IU/L)	15.1 (3.8–72.2)	14.8 (3.8–59.9)	15.3 (4.2–72.2)	0.573
AFP (ng/mL), <400, ≥400	614/102	306/52	308/50	0.831
PLT (×10^9/L), <150, ≥150	430/286	213/145	217/141	0.760
Tumor Size (cm), mean ± SD	2.49 ± 0.88	2.43 ± 0.87	2.56 ± 0.90	0.385
Tumor Size (cm), median (IQR)	2.3 (0.6–6.2)	2.3 (0.6–5.9)	2.4(1.0–6.2)	0.066
≤3cm	550	287	263
3.1–5.0cm	159	68	91
≥ 5.1cm	7	3	4
Tumor Number, 1/2/3	610/87/19	301/49/8	309/38/11	0.374
Child-pugh, A5-6/B7	680/36	342/16	338/20	0.494
ALBI grade, 1/2/3	331/376/9	167/187/4	164/189/5	0.928
TBS, low/medium/high	260/220/236	135/111/112	125/109/124	0.603
BCLC staging system, 0/1	219/497	115/243	104/254	0.372
CLIP score, 0/1/2	496/196/24	246/99/13	250/97/11	0.896
CUPI, 1/2	697/19	348/10	349/9	0.816
JIS score, 0/1/2	578/134/4	288/67/3	290/67/1	0.604
Okuda staging system, 1/2/3	665/49/2	332/26/0	333/23/2	0.452
PFS, median (IQR)	31 (1–153)	32 (1–144)	30 (1-153)	0.648
Overall Survival, median (IQR)	54 (1–153)	53 (1–150)	55 (6-153)	0.745

HBsAg: hepatitis B surface antigen; HCVAb: antibody to hepatitis C virus; HBV DNA: hepatitis B virus DNA; ALT: alanine aminotransferase; AST: aspartate aminotransferase; TBIL: total bilirubin; AFP: ɑ-fetoprotein; PLT: platelets; TBS: tumor burden score; ALBI: albumin-bilirubin; AJCC: American joint committee on cancer; BCLC: Barcelona Clinic Liver Cancer; CLIP: the Cancer of the liver Italian program; CUPI: the Chinese university prognostic index; JIS: the Japan iIntegrated staging; PFS: Progression-free Survival; IQR: inter quartile range.

4.2. Development of the progression-free and overall survival Prognostic scores

In the univariate analysis, serum AFP, TBS, and ALBI grade were found to be significant variables associated with PFS. In the multivariate Cox model, AFP ≥400 ng/mL (HR, 1.932; 95% CI, 1.383–2.698; p < 0.001), medium TBS (HR, 1.525; 95% CI, 1.106–2.102; p = 0.010), and high TBS (HR, 2.459; 95% CI, 1.803–3.353; p < 0.001) were identified as independent predictors of PFS (Table 2). A Cox proportional hazards model was constructed using a stepwise approach based on the aforementioned parameters. The AFP-TBS (AT) score was developed with the two predictors from the multivariate model, and the weight score of each variable was determined according to the predicted risk model. In this new AT score, ten points were assigned for high TBS, five points for medium TBS and seven points for AFP ≥400 ng/mL. The total score ranged from 0 to 17 points accordingly (Supplementary Table 1). An open-access online calculator was developed based on the outcomes of multivariate analysis (https://fah-sysu.shinyapps.io/OSHCCRFA/).

Table 2. Univariable and multivariable analyses of factors associated with progression-free survival and overall survival.

	OS				PFS
Variable	Univariate p-value	Univariate HR (95%CI)	Multivariate p-value	Multivariate HR (95%CI)	Univariate p-value	Univariate HR (95%CI)	Multivariate p-value	Multivariate HR (95%CI)
sex	0.467	0.567-1.297			0.411	0.615-1.220
Age ≥ 55	0.008	1.120–2.166			0.064	0.986–1.663
HBsAg (+)	0.342	0.506–1.266			0.302	0.822–1.881
HCVAb (+)	0.550	0.567–2.903			0.817	0.454–1.864
HBV DNA ≥100	0.114	0.534–1.069			0.478	0.694–1.187
ALT (IU/L)	0.322	0.998–1.007			0.586	0.997–1.005
AST (IU/L)	0.126	0.999–1.008			0.348	0.998–1.006
AFP (ng/mL), ≥400	<0.001	1.408–2.999	<0.001	1.525–3.284	<0.001	1.313–2.548	<0.001	1.383–2.698
PLT (×10^9/L), ≥150	0.030	0.511–0.965			0.091	0.61–1.037
ALBI
<–2.60	1				1
−2.60– −1.39	<0.001	1.456–2.764	<0.001	1.358–2.600	0.030	1.028–1.719	0.085
>–1.39	0.014	1.450–25.157	0.009	1.622–28.736	0.883	0.274–4.509	0.912
TBS
<2.42	1				1
2.42–3.07	0.015	1.107–2.550	0.026	1.059–2.474	0.020	1.062–2.013	0.010	1.106–2.102
>3.07	<0.001	1.954–4.259	<0.001	2.050–4.525	<0.001	1.755–3.255	<0.001	1.803–3.353

HBsAg: hepatitis B surface antigen; HCVAb: antibody to hepatitis C virus; HBV DNA: hepatitis B virus DNA; ALT: alanine aminotransferase; AST: aspartate aminotransferase; TBIL: total bilirubin; AFP: ɑ-fetoprotein; PLT: platelets; TBS: tumor burden score; ALBI: albumin-bilirubin.

In univariate analysis, age, serum AFP, platelets count, ALBI grade, and TBS were found to be significant variables associated with OS. In the multivariate Cox model, AFP ≥ 400 ng/mL (HR, 2.238, 95% CI, 1.525–3.284, p < 0.001), medium TBS (HR, 1.619, 95% CI: 1.059–2.474, p = 0.026), high TBS (HR, 3.046, 95% CI, 2.050–4.525, p < 0.001), ALBI grade 2 (HR, 1.879, 95% CI, 1.358–2.600, p < 0.001), and ALBI grade 3 (HR, 6.828, 95% CI, 1.622–28.736, p = 0.009) were identified as independent poor prognostic predictors of OS (Table 2). The AFP-TBS–ALBI (ATA) score was developed based on the three predictors from the multivariate model, and the weight score of each variable was determined according to the predicted risk model. In this new ATA score, two points were assigned for medium TBS, four points for high TBS; three points for ALBI grade 2 and AFP ≥ 400 ng/mL; five points for ALBI grade 3. The total score ranged from 0 to 12 points (Supplementary Table 2). An open-access online calculator based on the outcomes of multivariate analysis was developed (https://fah-sysu.shinyapps.io/PFSHCCRFA/).

4.3. Validation of the AFP-TBS and AFP-TBS-ALBI scores

The predictive performance of these two scores was evaluated using validation sets. From the aspects of 1-, 3-, 5-year PFS, the time-dependent AUCs of the training set were 0.651, 0.667, and 0.620, respectively (Figure 2(A)). The AUCs for the validation set were 0.657, 0.687, and 0.704, respectively (Figure 2(B)). Using a cutoff value of 7 for the AT score in relation to PFS, 206 (57.5%) and 152 (42.5%) patients in the validation cohort were classified into low-risk and high-risk groups, respectively, with median PFS durations of 55 (95% CI, 43–87) and 17 (95% CI, 13–21) months (p < 0.001) (Figure 3(A)). The 1-, 3-, and 5-year PFS rates were 83.5% (95% CI, 0.786–0.887), 62.2% (95% CI, 0.558–0.692) and 48.9% (95% CI, 0.422–0.567) for the low score group, and 60.5% (95% CI, 0.532–0.688), 26.9% (95% CI, 0.293–0.447) and 16.5% (95% CI, 0.114–0.239) for the high score group, respectively. The calibration curves showed good consistency between the observed and AT score-predicted probabilities for 1-, 3-, and 5-year PFS in the validation sets (Supplementary Figure 1(A–C)). DCA exhibited significant positive net benefits in the predictive model for most of the threshold probabilities at different time points, indicating a favorable potential clinical effect of the AT score (Supplementary Figure 2(A)).

Figure 2. Graphs Depict performance of the Cox proportional hazards model of the at and ATA scores. (A) From the 1,3,5-year progression-free survival aspect, the time-dependent AUC of the training set was 0.651,0.667, and 0.620, respectively. (B) while the AUC of the validation set was 0.657, 0.687, and 0.704, respectively. (C) From the 1,3,5-year overall survival aspect, the time-dependent AUC of the training set was 0.680,0.712, and 0.666, respectively. (D) while the AUC of the validation set was 0.712, 0.706, and 0.726, respectively.

Figure 3. Graphs Show comparison Kaplan–Meier curves of (A) progression-free survival (PFS) between low-risk (at score constructed from AFP and TBS [at score] < 7) and high-risk (at score ≥7) patients; (B)overall survival (OS) between low-risk (ATA score constructed from AFP, TBS, and ALBI [ATA score] < 5) and high-risk (ATA score ≥5) patients.

Considering the 1-, 3-, and 5-year OS aspects, the time-dependent AUCs of the training set were 0.680, 0.712 and 0.666, respectively (Figure 2(C)). The AUCs for the validation set were 0.712, 0.706, and 0.726, respectively (Figure 2(D)). Using a cutoff value of 5 for OS, 255 (71.2%) and 103 (28.8%) patients in the validation cohort were classified into the low-risk and high-risk groups, respectively, with median OS durations of 91 (95% CI, 83–141) and 47 (95% CI,40–57) months (p < 0.001) (Figure 3(B)). The 1-, 3-, and 5-year OS rates were 99.6% (95% CI, 0.988–1.0), 91.6% (95% CI, 0.882–0.951), and 71.6% (95% CI, 0.658–0.764) for the low score group, and 95.2% (95% CI, 0.910–0.994), 66.9% (95% CI, 0.584–0.785), and 36.9% (95% CI, 0.282–0.483) for the high score group, respectively. The calibration curves showed good consistency between the observed and ATA score-predicted probabilities for 1-, 3-, and 5-year OS in the validation sets (Supplementary Figure 1D-F). DCA exhibited significant positive net benefits in the predictive model for most of the threshold probabilities at different time points, indicating a favorable potential clinical effect of the ATA score (Supplementary Figure 2(B)). The prognostic performance of the predictive scores was compared with that of other models, including BCLC, CLIP, CUPI, JIS and Okuda (Table 3). Both the AT and ATA scores demonstrated the highest homogeneity and the lowest AIC, indicating better prognostic performance compared to the other models.

Table 3. Prognostic performance of different staging systems in the validation cohort.

Models for PFS	C-index (95% CI)	AIC	Models for OS	C-index (95% CI)	AIC
AT	0.643 (0.606–0.680)	2508.747	ATA	0.693 (0.650–0.736)	1805.383
BCLC	0.567 (0.536–0.598)	2544.844	BCLC	0.578 (0.545–0.611)	1860.631
CLIP	0.579 (0.548–0.610)	2541.083	CLIP	0.587 (0.548–0.626)	1858.075
CUPI	0.514 (0.502–0.526)	2569.156	CUPI	0.524 (0.508–0.540)	1863.978
JIS	0.53 (0.505–0.555)	2560.309	JIS	0.534 (0.501–0.567)	1873.307
Okuda	0.524 (0.506–0.542)	2562.536	Okuda	0.526 (0.504–0.548)	1869.451

AT: ɑ-fetoprotein and tumor burden score; ATA: ɑ-fetoprotein, tumor burden score and albumin-bilirubin score; BCLC: Barcelona Clinic Liver Cancer; CLIP: the Cancer of the liver Italian program; CUPI: the Chinese university prognostic index; JIS: the Japan Integrated staging.

5. Discussion

Ablation therapy is a radical treatment recommended by mainstream guidelines for early-stage HCC. However, patient heterogeneity poses a challenge for prognosis. Therefore, accurately predicting high-risk patients is crucial for selecting patients for ablation before surgery and determining postoperative surveillance intensity. The current study demonstrated that AT and ATA scores can accurately predict short-term and long-term prognosis, respectively. Thus, these scores can be used as tools for preoperative treatment selection. According to the results of multivariate analysis, serum AFP, TBS, and ALBI levels were integrated into the prognostic score. Notably, the novel AT and ATA scores performed relatively well in the validation cohort (C-index, 0.643; 95% CI, 0.606–0.680; C-index, 0.693; 95% CI, 0.650–0.736, respectively). Both AT and ATA scores exhibited the highest homogeneity and lowest AIC compared to other models.

Serum AFP is widely recognized as a surrogate of tumor biomarkers of tumor aggressiveness because it is closely linked to cell proliferation and cancer progression [27]. A higher AFP level is typically associated with a large tumor burden and poor prognosis [28,29]. In this study, preoperative serum AFP >400 ng/mL had prognostic value for both PFS and OS in HCC receiving ablation. Additionally, tumor burden has been considered a major predictor of outcomes and determining the stage. The diameter of the largest nodule and the total number of nodules represent the main tumor-related morphological variables traditionally used. Recently, Sasaki et al. established a novel tumor burden score (TBS) using the Pythagorean theorem, which incorporates both tumor number and size as continuous variables and can minimize the heterogeneity of tumor burden on long-term outcomes in patients with colorectal liver metastases [12]. Previous studies have demonstrated that TBS can also differentiate the prognosis of HCC patients undergoing resection or liver transplantation [30,31]. Due to the relatively small tumor burden of HCC eligible for ablation, we adjusted the TBS cutoff value accordingly, similar to the TBS cutoff value used by Ho, et al. [32]. Our results indicate that TBS can also be used as a prognostic surrogate for patients with HCC undergoing RFA.

The severity of liver dysfunction is another crucial outcome predictor of treating HCC. The Child–Pugh classification, as a traditional tool to assess the liver injury, has intrinsic drawbacks owing to its subjective variables with arbitrary cutoff values and inter-correlation of serum albumin and ascites. The ALBI grade is a simple and objective indicator of liver functional reserve based solely on serum levels of albumin and bilirubin and has been validated by several research groups [33,34]. In our study, ALBI grade was an independent prognostic indicator for OS but not PFS in HCC patients receiving RFA. It is understandable that recurrence is more associated with the intrinsic aggressiveness of the tumor, while liver reserve function plays a more important role in the quality of life and prognosis of patients with HCC.

The present study had several limitations that should be considered when interpreting the results. First, due to the retrospective nature of this study, the risk of selection bias is unavoidable. Second, our findings are based on a single medical center, and external validation with more patients from other institutions or in a prospective setting is required before applying our findings to patients with different etiologies of liver cirrhosis and HCC. Third, the present study only analyzed patients with BCLC-0/A HCC; thus, the data may not be applicable to patients with more advanced HCC (i.e. BCLC-B/C HCC).

6. Conclusion

The ATA/AT score, which is derived from clinical and objective laboratory variables, is easy to use and provides superior prognostic performance compared to other staging systems for HCC. Therefore, the ATA/AT score could serve as a novel tool for making clinical decisions in patients with HCC. We anticipate further refinement and adjustment of our model following additional external validation in the near future, including patients of different races, etiologies, and countries of origin.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by Guangdong Basic and Applied Basic Research Foundation (grant:2020A1515010653), National Natural Science Foundation of China (grant: 82102047, 82102163), and the Major Research Plan of the National Natural Science Foundation of China (grant: 92059201).