metricas
covid
Buscar en
Annals of Hepatology
Toda la web
Inicio Annals of Hepatology Comparison of diagnostic performance of AFP, DCP and two diagnostic models in he...
Información de la revista
Vol. 28. Núm. 4.
(julio - agosto 2023)
Compartir
Compartir
Descargar PDF
Más opciones de artículo
Visitas
2272
Vol. 28. Núm. 4.
(julio - agosto 2023)
Original article
Acceso a texto completo
Comparison of diagnostic performance of AFP, DCP and two diagnostic models in hepatocellular carcinoma: a retrospective study
Visitas
2272
Yongwu Chena,#, Xiuli Yangb,#, Yanfei Shaob, Hongying Zhaob, Jinying Jiangb, Ping Huangb,c, Yi Lud,
Autor para correspondencia
lyysz@zju.edu.cn

Corresponding authors.
, Zixue Xuanb,c,
Autor para correspondencia
xuanzixue0222@163.com

Corresponding authors.
a Department of Pharmacy, The First Affiliated Hospital of USTC, Division of Life Sciences and Medicine, University of Science and Technology of China, Hefei, 230036, China
b Center for Clinical Pharmacy, Cancer Center, Department of Pharmacy, Zhejiang Provincial People's Hospital, Affiliated People's Hospital, Hangzhou Medical College, Hangzhou, 310014, China
c Key Laboratory of Endocrine Gland Diseases of Zhejiang Province, Zhejiang Provincial People's Hospital, Affiliated People's Hospital, Hangzhou Medical College, Hangzhou, 310014, China
d General Surgery, Cancer Center, Department of Hepatobiliary & Pancreatic Surgery and Minimally Invasive Surgery, Zhejiang Provincial People's Hospital, Affiliated People's Hospital, Hangzhou Medical College, Hangzhou, 310014, China
Este artículo ha recibido
Información del artículo
Resumen
Texto completo
Bibliografía
Descargar PDF
Estadísticas
Figuras (3)
Mostrar másMostrar menos
Material adicional (1)
ABSTRACT
Introduction and Objectives

Hepatocellular carcinoma (HCC) may be diagnosed using the GAAP and ASAP models; our goal was to verify and evaluate their diagnostic effectiveness compared to alpha-fetoprotein (AFP), des-gamma-carboxy prothrombin (DCP), and AFP & DCP for both HCC and HCC caused by the hepatitis B virus (HBV).

Patients and Methods

GAAP and ASAP models were validated and compared using a retrospective investigation of 938 patients from our hospital between July 2020 and July 2021.

Results

Both the GAAP and ASAP models had better diagnostic efficacy than AFP, DCP, AFP & DCP. The GAAP model achieved better performance in section A for the detection of HCC and in section C for the detection of HBV-HCC than the ASAP model. The Hosmer-Lemeshow test showed that the GAAP and ASAP models were well-calibrated for the diagnoses of these two groups. To be more specific, the area under curve (AUC) of the GAAP model for HCC detection in section A was 0.862 [95% confidence interval (CI): 0.838-0.883], and that of the ASAP model was 0.850 [95% CI: 0.826-0.872]. The AUC of the GAAP model for HBV-HCC detection in section C was 0.897 [95% CI: 0.872-0.918], and that of the ASAP model was 0.878 [95% CI: 0.852-0.902].

Conclusions

The GAAP model was more accurate and reliable than the AFP, DCP, AFP and DCP, as well as the ASAP model in section A for the detection of HCC and in section C for the detection of HBV-HCC.

Keywords:
Diagnosis
GAAP model
ASAP model
DCP
Hepatocellular carcinoma
Abbreviations:
TRIPOD
ROC
AUC
US
CT
MRI
AFP
DCP
PIVKA-II
ALT
AST
ALP
TBIL
PLT
HCC
CHB
HBV
LC
HBV-LC
HC
Texto completo
1Introduction

China has the highest frequency of liver cancer, primarily hepatocellular carcinoma (HCC), based on the Globocan 2020 report by the International Agency for Research on Cancer (IARC), a section of the World Health Organization (WHO). The majority (80%) of patients with HCC in China are associated to infections with the hepatitis B virus (HBV) [1]. The prognosis of subjects with HCC depends greatly on earlier detection [2,3]. Conventionally, HCC diagnosis predominantly depends on abdominal ultrasound (US), multidetector-row spiral computed tomography (CT), magnetic resonance imaging (MRI) [4], and some blood markers, like alpha-fetoprotein (AFP), des-gamma-carboxy prothrombin (DCP). AFP is a recognized biomarker that has been widely applied in the serologic screening of early HCC [5]. With more sensitivity and specificity to AFP, DCP sometimes referred to as a protein triggered by vitamin K absence or antagonist II (PIVKA-II), has lately become a key diagnostic tool for liver malignancy [6,7]. However, whether a mixture of AFP and DCP could elevate the efficiency in the detection of early HCC remains to be discussed [8]. Liu et al. constructed a model based on gender, age, AFP, and DCP (GAAP model) in a single-center cohort consisting of 525 cancer cases (242 HCC, 187 cirrhosis, and 96 chronic hepatitis) [9], and found that the AUC of this model was 0.924 (95% CI, 0.895–0.952). Yang et al. also created the ASAP model, a diagnostic nomogram model depending on age, gender, AFP, and PIVKA-II, which exhibited slightly stronger diagnostic performance in the identification of HBV-HCC relative to the GALAD model [10,11]. Nevertheless, there is a paucity of validation in external sets for the discrimination and calibration of these two newer models. It is also unclear whether the models could exhibit favorable diagnostic value in other domestic medical facilities, and no studies are comparing the diagnostic and predictive efficacy of the two models. In this investigation, we discussed the diagnostic and predictive significance of AFP, DCP, AFP & DCP, the GAAP and ASAP models in patients with HCC, based on 938 cases from our hospital between July 2020 and July 2021,

2Patients and Methods2.1Design and patients

All patients were obtained in the investigation between July 2020 and July 2021, when they were admitted to the hospital. A dataset containing 262 cases with HCC (199 individuals with HBV-HCC or 143 HCC cases with liver cirrhosis (LC)), 173 subjects with LC (115 participants with HBV-LC), 393 individuals with chronic hepatitis B (CHB), and 110 healthy controls (HC) was utilized to examine the GAAP and ASAP models (Figure 1). The clinical database was employed to obtain demographic information, clinical features, diagnostic information, and laboratory findings (such as the participants' AFP and DCP levels, liver function, and standard blood testing). The Chinese Society of Hepatology's most recent recommendations for the prevention and management of CHB infection were followed in this investigation while establishing the diagnosis of CHB and associated cirrhosis [12,13]. Inclusion conditions for the Healthy control cohort were as follows: (1) no history of liver-related illness diagnosis or treatment, no family history of malignancy; and (2) serological indicators showed no current or prior HBV infection and an anti-HCV antibody analysis was negative; and (3) regular blood assessments, liver function testing, and kidney function analyse all revealed normal findings; and (4) no disorders were seen during the ultrasonography of the liver or gallbladder systems, and (5) the liver fibroscan findings showed no disorders. HCC was diagnosed by histopathological examination of the biopsy or met the following criteria: imaging tests (US, CT, MRI, or other imaging tests) showing typical imaging injuries of HCC, and the lesion tissue had typical changes of blood flow. The inclusion criteria were: (1) HCC not treated with surgery, radiation, chemotherapy, or ablation; (2) no missing data for AFP or DCP. The following were the exclusion conditions: (1) current warfarin therapy; (2) in patients with multiple admissions, only the first admission was included.

Figure 1.

Study diagram. Firstly, 1151 participants screened form our hospital between July 2020 and July 2021, then 108 repeat participants, 103 participants with incomplete or missing data, and 2 participants with warfarin takers were excluded. So 938 participants, consisting of 262 patients with HCC (199 patients with HBV-HCC or 143 HCC patients with liver cirrhosis (LC)), 173 patients with LC (115 patients with HBV-LC), 393 patients with chronic hepatitis B (CHB), and 110 healthy controls (HC), were included in this study. Section A included 262 HCC, 393 CHB, and 173 LC patients and 110 HC. Section B included 199 HBV-HCC, 393 CHB, 173 LC, and 110 HC. Section C included 199 HBV-HCC, 393 CHB, 115 HBV-LC. Section D included 143 HCC with LC and 173 LC.

(0.21MB).

In this study, the predictors included in the ASAP and GAAP models were age, gender, AFP, and DCP. We estimated the sample size based on the need for 10 positive outcome events per predictor. In addition, as a single-level model, at least 100 events are required. We take the larger of the two as the lower limit of the sample size. Therefore, the estimated sample size of this study is 100.

2.2AFP and DCP Assays

Utilizing Roche electrochemiluminescence immunoassay (ng/mL units), serum AFP levels were determined in this research. Utilizing the ARCHITECT immunoassay (mAU/mL units) and frozen-thawed serum, DCP, which is formed in tumor tissues as a consequence of an acquired deficiency in posttranslational carboxylation of the prothrombin precursor, was detected [11]. The laboratory tests were conducted by technicians who were not aware of the subjects' diagnosis. There were no negative outcomes associated with the collection of serum samples.

2.3Models for validation

For the reporting and execution of this external validation research, we strictly followed the Transparent Reporting of a Multivariable Prediction Model for Individual Prognosis or Diagnosis (TRIPOD) standards [14].

The GAAP and ASAP models, as described previously [11,15], use the following equations:

  • 1. GAAP score = −11.203 + 0.699 × [Sex (1 for male, 0 for female)] + 0.094 × [Age] + 1.076 × log10 [AFP] +2.376 × log10 [DCP].

  • 2. ASAP score = −7.57711770 + 0.04666357 × [Age] −0.57611693 × [Sex (1 for female, 0 for male)] + 0.42243533 × ln[AFP]+1.10518910 × ln[DCP].

2.4Statistical analysis

Program for the analyses, graphics, or both was created employing GraphPad Prism version 8.0.1, MedCalc version 18.2.1, and SPSS version 24.0. Categorical variables were reported as percentages for demographic data, and the chi-square test was employed to determine differences. Continuous skewed distribution variables were provided as median and interquartile range values, and variations were assessed using either the Mann-Whitney U-test or the Kruskal-Wallis H-test. Except for AFP and DCP, where the cut-off values were based on clinical criteria and were, respectively, AFP 20 ng/ml and DCP 40 mAU/mL, the best cut-off values were calculated depending on the Youden index. The area under the curve (AUC) for AFP and DCP utilized in conjunction with the GAAP or ASAP model as predictors was calculated using ROC curves [10]. To evaluate diagnostic accuracy, three metrics were used: AUC, sensitivity, and specificity. The Delong test was used to compare different AUC values. The Hosmer-Lemeshow analysis was employed to calibrate the model, and calibration plots were used to evaluate it [16]. Binary logistic regression was employed to anticipate the probability of HCC and HBV-HCC to assess the diagnostic precision of the combined AFP and DCP results. At P ≤ 0.05, variations were deemed statistically significant.

2.5Ethical statement

Due to the study's retrospective character, an exemption from the informed consent criteria was authorized. The 1975 Declaration of Helsinki's ethical principles were followed by the research design. The ethics committee of the Zhejiang Provincial People's Hospital authorized the investigation, which attests to this (APPROVAL NUMBER: 2021QT280).

3Results3.1Characteristics of the Study Participants

Typically, 938 patients were eligible and analyzed using GAAP and ASAP models. Table 1 displays the medical and demographic information about the groups. Individuals with HCC were older than participants without HCC (P <0.001). In comparison to the non-HCC group, the HCC group's serum levels of AFP, DCP, AST, ALP, and PLT were greater ( Figure 2. The serum concentrations of DCP and AFP were greater in the HCC and HBV-HCC groups than in other groups (CHB, LC, HBV-LC, HC) (P<0.05).

Table 1.

Characteristics of the study subjects used to evaluate the GAAP and ASAP model

Characteristics  HCC(n=262)  non-HCC(n=676)  CHB(n=393)  HBV-HCC(n=199)  LC(n=173)  HBV-LC(n=115)  non-HBV-LC(n=58)  HC(n=110)  Z  P  H  P 
Age  61.0 (52.0, 68.0)  48.0 (40.0, 58.0)  45.0 (36.0, 53.0)  58.0 (51.0, 68.0)  57.0 (49.0, 65.5)  52.0 (46.0, 59.0)  68.0 (60.2, 76.0)  51.5 (42.8, 64.0)  -10.606  <0.001  298.157  <0.001 
AFP (ng/mL)  6.2 (2.7, 162.6)  2.8 (2.0, 4.8)  2.6 (1.9, 3.7)  8.9 (2.8, 198.4)  3.5 (2.2, 7.2)  3.8 (2.2, 7.8)  3.2 (1.9, 6.1)  2.9 (2.1, 4.7)  -9.328  <0.001  203.714  <0.001 
DCP(mAU/mL)  91.4 (25.2, 1534.7)  25.0 (20.2, 31.4)  24.7 (20.4, 29.5)  108.5 (25.2, 1608.8)  25.2 (18.6, 37.1)  25.2 (18.8, 34.1)  26.7 (17.9, 49.4)  26.2 (21.9, 33.1)  -12.128  <0.001  256.228  <0.001 
ALT(U/L)  26.0 (18.0, 46.5)  27.0 (19.0, 45.0)  27.0 (19.0, 45.3)  27.5 (19.0, 45.3)  30.0 (19.0, 43.0)  31.0 (21.0, 43.0)  27.0 (16.0, 44.8)  29.0 (19.0, 66.0)  1.406  0.160  4.789  0.686 
AST(U/L)  35.0 (25.0, 63.5)  27.0 (21.0, 41.0)  25.0 (21.0, 33.0)  36.0 (25.0, 62.0)  36.0 (25.3, 56.0)  34.0 (24.3, 54.0)  39.0 (30.5, 74.8)  25.5 (19.3, 54.8)  -5.400  <0.001  112.837  <0.001 
TBIL(mmol/L)  17.4 (12.4, 27.2)  16.0 (12.2, 23.5)  15.4 (11.8, 20.4)  17.2 (12.3, 24.7)  20.3 (14.8, 34.3)  19.4 (14.8, 29.8)  25.4 (14.4, 48.3)  14.6 (11.1, 24.7)  -1.456  0.145  60.812  <0.001 
PLT(× 109/L)  126.0 (89.0, 175.0)  159.0 (10.2, 208.0)  180.0 (146.0, 220.0)  120.0 (80.3, 172.8)  101.0 (56.0, 145.0)  101.0 (51.0, 146.0)  100.0 (62.8, 144.0)  207.5 (159.0, 263.5)  4.542  <0.001  171.460  <0.001 

Mann-Whitney U-tests were used for comparison HCC group with non-HCC group; Kruskal–wallis H-tests were used for comparisons among groups. The P value represents the statistical difference between two groups or among multiple groups. Z and H values represent the statistics of Mann-Whitney U-tests and Kruskal–wallis H-tests, respectively.

Note: AFP showed alpha-fetoprotein, DCP showed abnormal prothrombin, ALT showed alanine aminotransferase, AST showed aspartic acid aminotrans- ferase, ALP showed Alkaline phosphatase, TB showed total bilirubin, PLT showed platelet, HCC showed hepatocellular carcinoma, CHB showed chronic hepatitis B, HBV-HCC showed HBV-related hepatocellular carcinoma, LC showed liver cirrhosis, HBV-LC showed HBV-related liver cirrhosis.

Figure 2.

Serum AFP and DCP in HCC and non-HCC groups. Comparison of AFP and DCP among HCC, HBV-related HCC, Cirrhosis, HBV-related cirrhosis, chronic hepatitis B, and HC groups. The three horizontal bars represent median with interquartile range values. For AFP and DCP, Mann-Whitney U-tests were used for comparisons among groups; ***P <0.001, **P <0.01, *P <0.05, ns P>0.05.

(0.22MB).
3.2Performance of GAAP and ASAP Models for the detection of HCC and HBV-HCC

In Table 2 and Table 3, section A included 262 HCC, 393 CHB, 173 LC patients, and 110 HC. Section B included 199 HBV-HCC, 393 CHB, 173 LC, and 110 HC. Section C included 199 HBV-HCC, 393 CHB, and 115 HBV-LC. Section D included 143 HCC with LC and 173 LC. Both models had greater AUC values than the patient markers DCP, AFP, and AFP&DCP in the population of sections A to D (Figure 3A-D). The AUC of the GAAP model for HCC detection in the population of section A was 0.862 (95% confidence interval [CI): 0.838-0.883] which was superior to that of AFP (0.655, p<0.0001), DCP (0.746, p<0.0001), AFP&DCP (0.781, p<0.0001), and the ASAP model (0.850, p=0.0077) (Table 2; Figure 3A). At an optimal cut-off of -0.7995, the GAAP score had a sensitivity of 74.43% and a specificity of 81.36% for HCC detection. The AUC of the GAAP model for HBV-HCC detection in the population of section C was 0.897 (95% confidence interval (CI): 0.872-0.918) which was superior to that of AFP (0.668, p<0.0001), DCP (0.773, p<0.0001), AFP and DCP (0.784, p<0.0001), and the ASAP model (0.878, p=0.0006) (Table 3; Figure 3C). At an optimal threshold of -0.7995, the GAAP score had a sensitivity of 75.38% and a specificity of 88.19% for HBV-HCC detection. The Hosmer-Lemeshow test showed that the GAAP and ASAP models had a good calibration for the identification of HCC in the people of Section A (P=0.179 for GAAP; P=0.486 for ASAP) and HBV-HCC in the population of section C (P=0.428 for GAAP; P=0.455 for ASAP). The GAAP model and the ASAP model had a poor calibration for the determination of HBV-HCC in the participants of section B (P=0.020 for GAAP; P=0.017 for ASAP), and for the identification of HCC in the people of section D (P=0.029 for GAAP; P=0.132 for ASAP) (Supplementary Figure S1). These findings showed that the GAAP model achieved a better performance in section A for the detection of HCC and in section C for the detection of HBV-HCC than the ASAP model.

Table 2.

AFP, DCP and 2 models for the diagnosis of HCC and HBV-HCC in the whole population

Section ASection B
Model/biomarker  Cut-off value  AUC (95% CI)  P value (VS GAAP)  sensitivity,%  specificity,%  Calibrationof model,P value  PPV%  NPV%  Model/biomarker  Cut-off value  AUC (95% CI)  P value (VS GAAP)  sensitivity,%  specificity,%  Calibrationof model,P value  PPV%  NPV% 
AFP  20ng/ml  0.655 (0.624 to 0.686)  P < 0.0001  37.40  93.64  69.50  79.42  AFP  20ng/ml  0.663 (0.632 to 0.693)  P < 0.0001  40.70  91.88  57.45  85.19 
DCP  40 mAU/mL  0.746 (0.717 to 0.774)  P < 0.0001  59.92  89.35  68.56  85.19  DCP  40 mAU/mL  0.741 (0.711 to 0.768)  P < 0.0001  62.31  85.79  54.15  89.42 
AFP&DCP  0.19774  0.781 (0.753 to 0.807)  P < 0.0001  67.56  81.95  59.19  86.70  AFP&DCP  0.09025  0.773 (0.745 to 0.799)  P < 0.0001  68.34  80.11  48.07  90.38 
ASAP  -0.43923  0.850 (0.826 to 0.872)  P = 0.0077  72.52  82.99  0.486  62.30  88.63  ASAP  -0.22321  0.829 (0.804 to 0.853)  P = 0.1080  69.85  81.73  0.017  50.73  90.96 
GAAP  -0.79952  0.862 (0.838 to 0.883)  74.43  81.36  0.179  60.75  89.14  GAAP  -0.81783  0.837 (0.812 to 0.861)  75.88  76.45  0.020  46.46  92.17 

CI, confdence interval; NPV, negative prediction value; PPV, positive prediction value.

Table 3.

AFP, DCP and 2 models for the diagnosis of HCC with HBV etiology or LC.

Section CSection D
Model/biomarker  Cut-off value  AUC (95% CI)  P value(VS GAAP)  sensitivity, %  specificity, %  Calibrationof model,P value  PPV%  NPV%  Model/biomarker  Cut-off value  AUC (95% CI)  P value (VS GAAP)  sensitivity, %  specificity,%  Calibrationof model,P value  PPV%  NPV% 
AFP  20ng/ml  0.668(0.632 to 0.703)  P < 0.0001  40.70  92.91  69.22  80.00  AFP  20ng/ml  0.645(0.589 to 0.698)  P < 0.0001  40.56  88.44  74.36  64.29 
DCP  40 mAU/mL  0.773(0.740 to 0.804)  P < 0.0001  62.31  92.32  76.07  86.21  DCP  40 mAU/mL  0.744(0.692 to 0.791)  P = 0.0022  71.33  77.46  72.34  76.58 
AFP&DCP  0.18461  0.784(0.751 to 0.813)  P < 0.0001  67.84  89.96  72.58  87.71  AFP&DCP  0.09025  0.785(0.736 to 0.829)  P = 0.2394  68.34  80.11  73.96  75.38 
ASAP  -0.43923  0.878(0.852 to 0.902)  P = 0.0006  72.86  87.99  0.455  70.39  89.22  ASAP  -0.22321  0.815(0.768 to 0.857)  P = 0.1336  69.85  81.73  0.132  75.96  76.63 
GAAP  -0.79952  0.897(0.872 to 0.918)  75.38  88.19  0.428  71.43  90.14  GAAP  -0.81783  0.805(0.757 to 0.847)  75.88  76.45  0.029  72.70  79.32 
Figure 3.

ROC curves of AFP, DCP, AFP&DCP, GAAP and ASAP models for the diagnosis of HCC and HBV-HCC. (A): Discriminating HCC in the whole population; (B): Discriminating HBV-HCC in the whole population; (C): Discriminating HBV-HCC within HBV etiology; (D): Discriminating HCC in the LC population.

(0.48MB).
4Discussion

Early identification of liver cancer is vitally important to improve treatment efficacy and survival outcomes [17]. Currently, biomarkers that are commonly used in early HCC diagnosis are AFP, DCP, and AFP-L3 [18]. DCP was first discovered in Japan and then applied in the clinical diagnosis and screening of liver malignancy [19]. When the threshold was established at 40 mAU/mL, it was shown that DCP had a sensitivity of 66% and a specificity of 89% in identifying HCC in a meta-analysis including 31 trials [20]. Furthermore, it has been observed that DCP is more effective than AFP in diagnosing liver cancer. Compelling evidence was exhibited by a multi-center retrospective study on primary HCC [21], with an AUC of 0.939 for DCP and 0.817 for AFP (p<0.05). According to a meta-analysis, AFP-L3 had a sensitivity and specificity of 48.3% and 92.9%, respectively, in the diagnosis of HCC [22]. As AFP-L3% and AFP are cooperative and complementary, a combination of the two fails to largely increase the diagnostic efficiency for liver cancer [23]. Moreover, such a strategy is limited by instability in testing, high cost, and other issues.

We also discovered that DCP had better ROC-AUC values than AFP in the present investigation for both HCC (0.746 vs 0.655, with a sensitivity and specificity of 59.92% and 89.35%) and HBV-HCC (0.773 vs 0.668, with a sensitivity and specificity of 62.31% and 92.32). Notably, the combination of DCP and AFP contributed to higher ROC-AUC values in the populations of HCC (0.781) and HBV-HCC (0.784), and the sensitivity and specificity increased to 67.56%, 67.84%, and 81.95%, 89.96%, respectively. These outcomes suggested that the mixture of DCP and AFP can make tumor detection more reliable.

Recently, studies showed that models based on AFP, DCP, and other clinical features, can be used to predict HCC. For instance, Johnson et al. developed the GALAD model to identify liver cancer in a UK cohort depending on sex, age, AFP-L3, AFP, and DCP. They found that the model's sensitivity and specificity were 85.6% and 93.3%, respectively [24]. In contrast to the GALAD model, both the GAAP and ASAP models, do not contain AFP-L3, which may be economic and practicable [11]. Importantly, the ASAP and GAAP models were respectively applied to risk prediction in two Chinese populations: HBV-HCC in individuals with CHB or LC, and HCC in patients with chronic liver disease [11,15]. Because the two models are not validated and compared using other external data, so GAAP and ASAP models were validated and compared using a retrospective study of 938 cases from our hospital. As we know, discrimination and calibration are two recognized factors that are frequently used in the judgment of a predictive model for practicability and accuracy in differentiating between cohorts of different outcomes [25]. Generally, the validation of a model includes three aspects: internal, time, and external, with external validation considered the most efficient in identifying the practicability of a model [26]. Here, external validation was adopted, and we found the GAAP model was superior to the ASAP model and AFP&DCP in predicting HBV-HCC in individuals with CHB or HBV-LC, as well as in predicting HCC in the group of subjects with chronic liver disease. In addition, the Hosmer-Lemeshow test showed that the GAAP and ASAP models were well-calibrated for the recognition of HCC in the population of section A and HBV-HCC in the population of section C. Therefore, these findings indicated that the GAAP model was more effective than the ASAP model and AFP&DCP, in the population of section A for the identification of HCC and in the population of section C for the detection of HBV-HCC. However, the GAAP model and the ASAP model had a poor calibration for the recognition of HBV-HCC in the population of section B, and for the detection of HCC in the population of section D, indicating the GAAP and ASAP models are not available for the diagnosis of these two groups.

This investigation has some restrictions. For example, this was a single-center retrospective analysis with small sample size, and tumor size and phase were not stratified. In the future, retrospective or prospective multi-center large sample investigations are necessary to further verify the prediction ability of the ASAP and GAAP models.

5Conclusions

In this study, the diagnostic efficacies of the ASAP and GAAP models were externally verified, and the results showed that the GAAP model was more effective than the AFP, DCP, AFP&DCP, as well as the ASAP model in the whole population for the identification of HCC and in the HBV subset for the detection of HBV-HCC.

Funding

This work was funded by Zhejiang Province Natural Science Foundation of China (Grant No. LYY21H310008 and No. LYY19H310009), Zhejiang Medical Technology Plan Project (Grant No.2022KY060, No.2020KY034, No.2019KY319), and the “10,000 Talents Plan” of Zhejiang Province (Grant No. 2020R52029).

Data sharing statement

This article has all the data that were created or evaluated during this investigation.

Author contributions

(I) Conception and design: (ZX.X, P.H); (II) Administrative support: (ZX.X, P.H); (III) Provision of trial materials or cases: (XL.Y, YF.S, HY.Z, JY.J); (IV) Gathering and assembly of information: (YW.C, ZX.X); (V) Data examination and interpretation: (YW.C, Y.L, ZX.X); (VI) Manuscript writing: All authors; (VII) Final authorization of manuscript: All authors.

Appendix
Supplementary materials

Supplementary Figure S1. Calibration plots for the GAAP and ASAP models for predicting the HCC and HBV-HCC. The red line represents perfect calibration. (1) ASAP model; (2) GAAP model. (A): Discriminating HCC in the whole population; (B): Discriminating HBV-HCC in the whole population; (C): Discriminating HBV-HCC within HBV etiology; (D): Discriminating HCC in the LC population.

References
[1]
H Sung, J Ferlay, RL Siegel, M Laversanne, I Soerjomataram, A Jemal, et al.
Global Cancer Statistics 2020: GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries.
CA Cancer J Clin, 71 (2021), pp. 209-249
[2]
J Fu, H. Wang.
Precision diagnosis and treatment of liver cancer in China.
Cancer Lett, 412 (2018), pp. 283-288
[3]
JA Marrero, LM Kulik, CB Sirlin, AX Zhu, RS Finn, MM Abecassis, et al.
Diagnosis, Staging, and Management of Hepatocellular Carcinoma: 2018 Practice Guidance by the American Association for the Study of Liver Diseases.
Hepatology, 68 (2018), pp. 723-750
[4]
C Ayuso, J Rimola, R Vilana, M Burrel, A Darnell, Á García-Criado, et al.
Diagnosis and staging of hepatocellular carcinoma (HCC): current guidelines.
Eur J Radiol, 101 (2018), pp. 72-81
[5]
PR Galle, F Foerster, M Kudo, SL Chan, JM Llovet, S Qin, et al.
Biology and significance of alpha-fetoprotein in hepatocellular carcinoma.
Liver Int, 39 (2019), pp. 2214-2229
[6]
NI Zakhary, SM Khodeer, HE Shafik, CA. Abdel Malak.
Impact of PIVKA-II in diagnosis of hepatocellular carcinoma.
J Adv Res, 4 (2013), pp. 539-546
[7]
N Poté, F Cauchy, M Albuquerque, H Voitot, J Belghiti, L Castera, et al.
Performance of PIVKA-II for early hepatocellular carcinoma diagnosis and prediction of microvascular invasion.
J Hepatol, 62 (2015), pp. 848-854
[8]
P Kotwani, W Chan, F Yao, N. Mehta.
DCP and AFP-L3 Are Complementary to AFP in Predicting High-Risk Explant Features: Results of a Prospective Study.
Clin Gastroenterol Hepatol, (2021),
[9]
M Liu, R Wu, X Liu, H Xu, X Chi, X Wang, et al.
Validation of the GALAD Model and Establishment of GAAP Model for Diagnosis of Hepatocellular Carcinoma in Chinese Patients.
J Hepatocell Carcinoma, 7 (2020), pp. 219-232
[10]
B Li, Y Zhao, W Cai, A Ming, H Li.
Validation and update of a multivariable prediction model for the identification and management of patients at risk for hepatocellular carcinoma.
Clin Proteomics, 18 (2021), pp. 21
[11]
T Yang, H Xing, G Wang, N Wang, M Liu, C Yan, et al.
A Novel Online Calculator Based on Serum Biomarkers to Detect Hepatocellular Carcinoma among Patients with Hepatitis B.
Clin Chem, 65 (2019), pp. 1543-1553
[12]
JL Hou, W. lai.
[The guideline of prevention and treatment for chronic hepatitis B: a 2015 update].
Zhonghua Gan Zang Bing Za Zhi, 23 (2015), pp. 888-905
[13]
L Wei, JL. Hou.
[The guideline of prevention and treatment for hepatitis C: a 2015 update].
Zhonghua Gan Zang Bing Za Zhi, 23 (2015), pp. 906-923
[14]
RE Patzer, AH Kaji, Y. Fong.
TRIPOD Reporting Guidelines for Diagnostic and Prognostic Studies.
JAMA Surg, 156 (2021), pp. 675-676
[15]
PJ Johnson, SJ Pirrie, TF Cox, S Berhane, M Teng, D Palmer, et al.
The detection of hepatocellular carcinoma using a prospectively developed and validated model based on serological biomarkers.
Cancer Epidemiol Biomarkers Prev, 23 (2014), pp. 144-153
[16]
G Nattino, ML Pennell, S. Lemeshow.
Assessing the goodness of fit of logistic regression models in large samples: A modification of the Hosmer-Lemeshow test.
Biometrics, 76 (2020), pp. 549-560
[17]
Y Zhu, LX. Qin.
Strategies for improving the efficacy of immunotherapy in hepatocellular carcinoma.
Hepatobiliary Pancreat Dis Int, 21 (2022), pp. 420-429
[18]
K Yamamoto, H Imamura, Y Matsuyama, Y Kume, H Ikeda, GL Norman, et al.
AFP, AFP-L3, DCP, and GP73 as markers for monitoring treatment response and recurrence and as surrogate markers of clinicopathological variables of HCC.
J Gastroenterol, 45 (2010), pp. 1272-1282
[19]
T Suehiro, T Matsumata, H Itasaka, A Taketomi, K Yamamoto, K. Sugimachi.
Des-gamma-carboxy prothrombin and proliferative activity of hepatocellular carcinoma.
Surgery, 117 (1995), pp. 682-691
[20]
H Xing, YJ Zheng, J Han, H Zhang, ZL Li, WY Lau, et al.
Protein induced by vitamin K absence or antagonist-II versus alpha-fetoprotein in the diagnosis of hepatocellular carcinoma: A systematic review with meta-analysis.
Hepatobiliary Pancreat Dis Int, 17 (2018), pp. 487-495
[21]
JY Choi, SW Jung, HY Kim, M Kim, Y Kim, DG Kim, et al.
Diagnostic value of AFP-L3 and PIVKA-II in hepatocellular carcinoma according to total-AFP.
World J Gastroenterol, 19 (2013), pp. 339-346
[22]
X Yi, S Yu, Y. Bao.
Alpha-fetoprotein-L3 in hepatocellular carcinoma: a meta-analysis.
Clin Chim Acta, 425 (2013), pp. 212-220
[23]
F Qi, A Zhou, L Yan, X Yuan, D Wang, R Chang, et al.
The diagnostic value of PIVKA-II, AFP, AFP-L3, CEA, and their combinations in primary and metastatic hepatocellular carcinoma.
J Clin Lab Anal, 34 (2020), pp. e23158
[24]
S Berhane, H Toyoda, T Tada, T Kumada, C Kagebayashi, S Satomura, et al.
Role of the GALAD and BALAD-2 Serologic Models in Diagnosis of Hepatocellular Carcinoma and Prediction of Survival in Patients.
Clin Gastroenterol Hepatol, 14 (2016), pp. 875-886
[25]
JS Chiu, FC Yu, YC. Li.
Discrimination and calibration are concurrently required for model comparison.
Int J Cardiol, 112 (2006), pp. 245-246
[26]
ME Matheny, L Ohno-Machado, FS. Resnic.
Discrimination and calibration of mortality risk prediction models in interventional cardiology.
J Biomed Inform, 38 (2005), pp. 367-375

These authors contributed equally to this work.

Copyright © 2023. Fundación Clínica Médica Sur, A.C.
Descargar PDF
Opciones de artículo
es en pt

¿Es usted profesional sanitario apto para prescribir o dispensar medicamentos?

Are you a health professional able to prescribe or dispense drugs?

Você é um profissional de saúde habilitado a prescrever ou dispensar medicamentos