Patient characteristics
We retrospectively incorporated 168 patients with a single HCC, who had undergone radical hepatectomy by the same surgeon during January 2018 to June 2020. On the basis of oncology status during follow-up, study population was divided into nonrecurrence group and recurrence group, incorporating 90 and 78 patients, respectively. Recurrence group had significantly higher TAP (204.5[154.1,228.1] vs 165.1[134.9༌203.0]µm2, respectively; p = 0.002) and distinctly bigger tumor size (5.0 [3.5, 6.0] vs 4.5 [3.0, 5.0] cm, respectively; p = 0.003) than patients in nonrecurrence group (Table 1). However, in terms of demographic characteristics, HBsAg status, MVI, cirrhosis, liver function level, degree of differentiation, operative parameters and Child-Pugh score, no statistical difference was found between groups (P > 0.05 for all) (Table 1).
The outcome of Pearson r analysis between TAP and maximum tumor size was showed in Fig. 1. There was no significant relationship between TAP and maximum tumor size (r = 0.122; P = .117; Fig. 1).
Table 1
Baseline characteristics in recurrence group and non-recurrence group
| Recurrence group | Non-recurrence group | P |
Characteristics | (n = 78) | (n = 90) | |
Age, year | 53.0(45.8,59.0) | 53.0(48.0,63.0) | 0.302 |
Gender(Male/Female) | 59/19 | 72/18 | 0.497 |
HBsAg (+/-) | 56/22 | 65/25 | 0.951 |
Cirrhosis(Yes/No) | 56/22 | 67/23 | 0.699 |
Differentiation(Poor/Well) | 35/43 | 25/63 | 0.092 |
Child-Pugh(A/B) | 76/2 | 86/4 | 0.687 |
AST (U/L) | 33.3(24.0,88.0) | 27.5(22.4,45.0) | 0.078 |
ALT(U/L) | 38.1(17.9,74.9) | 29.1(18.9,53.3) | 0.254 |
TBIL (U/L) | 14.5(10.6,22.9) | 14.2(11.0,19.5) | 0.515 |
BMI (kg/m2) | 22.7(20.8,26.4) | 24.1(21.8,26.7) | 0.28 |
Tumor diameter (cm) | 5.0(3.5,6.0) | 4.5(3.0,5.0) | 0.003 |
MVI (Yes/No) | 36/42 | 30/60 | 0.09 |
TAP (µm2) | 204.5(154.1,228.1) | 165.1(134.9,203.0) | 0.002 |
Pringle maneuver (Yes/No) | 42/36 | 50/40 | 0.824 |
Total blood loss (ml) | 400.0(200.0,700.0) | 400.0(200.0,600.0) | 0.567 |
Operation time (min) | 220.0(200.0,185.0) | 247.5(203.8,300.0) | 0.214 |
Operative procedure | | | 0.902 |
Extended hemihepatotectomy | 7 | 12 | |
Hemihepatotectomy | 15 | 17 | |
Sectionectomy | 30 | 35 | |
Segmentectomy | 14 | 13 | |
Laparoscopic approach | 12 | 13 | |
HBsAg hepatitis B surface antigen, ALT alanine aminotransferase, AST aspartate aminotransferase, TBIL total bilirubin level, MVI microvascular invasion, TAP tomor abnormal protein |
Survival And Risk Factor
According to the cut-off value of TAP level mentioned in previous literatures[13], the whole population was categorized into the TAP high group (TAP ≥ 225 µm2) and TAP low group (TAP < 225 µm2). As shown in Fig. 2A, the RFS rate was distinctly higher in the TAP low group than TAP high group in the whole population (P < 0.001). Furthermore, in the stratification analysis according to maximum tumor diameter (≤ 5 or > 5cm), the RFS rate of TAP high group was distinctly different from that of TAP low group (P < 0.001 and P = 0.001, respectively; Fig. 2B,C).
In univariate analysis, TAP (P < 0.001), presence of MVI (P = 0.008), large tumor size (P < 0.001) and degree of tumor differentiation (p = 0.042) were distinctly associated with RFS in HCC patients who had undergone radical surgery (Table 2). In addition, the multivariate analysis suggested that TAP (hazard ratio [HR], 3.47; 95% CI, 2.18-5.51; P < 0.001), large tumor size (HR, 2.18; 95% CI, 1.36‐3.49; P < 0.001), poor tumor differentiation (HR, 0.53; 95% CI, 0.33‐0.84; P = 0.007) and presence of MVI (HR, 2.03; 95% CI, 1.28‐3.22; P = 0.003) were still independently correlated with RFS (Table 2).
Table 2
Independent prognostic factors predicting RFS in the whole population
Characteristics | Univariate analysis | | Multivariate analysis | |
| HR(95% CI) | P-value | HR(95% CI) | P-value |
Age, year | | | | |
≤ 60,༞60 | 0.70 (0.42–1.19) | 0.189 | | |
Gender | | | | |
Male/Female | 1.07 (0.64–1.80) | 0.787 | | |
HBsAg | | | | |
Positive/Negative | 0.85 (0.52–1.40) | 0.528 | | |
Cirrhosis | | | | |
Yes/No | 0.99 (0.61–1.63) | 0.979 | | |
Differentiation | | | | |
Poor/Well | 0.63 (0.40–0.98) | 0.042 | 0.53(0.33–0.84) | 0.007 |
Child-Pugh | | | | |
A/B | 0.82 (0.20–3.32) | 0.776 | | |
MVI | | | | |
Yes/No | 1.83 (1.17–2.87) | 0.008 | 2.03 (1.28–3.22) | 0.003 |
Pringle maneuver | | | | |
Yes/No | 1.13 (0.73–1.77) | 0.583 | | |
Maximum tumor size (cm) | | | | |
≤ 5,༞5 | 2.24 (1.43–3.49) | ༜0.001 | 2.18 (1.36–3.49) | ༜0.001 |
AST (U/L) | | | | |
≤ 40,༞40 | 1.22 (0.77–1.94) | 0.389 | | |
ALT(U/L) | | | | |
≤ 40,༞40 | 1.36 (0.87–2.12) | 0.178 | | |
TBIL (U/L) | | | | |
≤ 21,༞21 | 1.19 (0.72–1.99) | 0.494 | | |
TAP (µm2) | | | | |
༜225,≥225 | 3.34(2.13–5.23) | ༜0.001 | 3.47 (2.18–5.51) | ༜0.001 |
HBsAg hepatitis B surface antigen, ALT alanine aminotransferase, AST aspartate aminotransferase, TBIL total bilirubin level, MVI microvascular invasion, TAP tomor abnormal protein |
Construction Of A Robust Nomogram
According to abovementioned independent prognostic variables, we constructed a robust nomogram (Fig. 3A). Figure 3B depicted the calibration curves of this nomogram, which indicated favorable consistency between the predicted and observed relapse risk in the whole population, and the result of the Hosmer-Lemeshow test suggested an excellent fit (P = 0.519). Table 3 listed the statistical parameters of this predictive model, such as accuracy, specificity and sensitivity.
To emphasize the contribution of TAP in nomogram, we used ROC curve analysis to compare the prognostic performance of model incorporating TAP, MVI, maximun tumor diameter and degree of tumor differentiation with the model that integrated maximun tumor diameter, degree of tumor differentiation and MVI only. With regard to RFS, the area under the ROC curve (AUC) of the model without TAP and the combined model was 0.68 and 0.77 (p = 0.022; Fig. 3C and Table 3), respectively, which indicated that the predictive performance of conjunctive model was more excellent than the model that didn’t incorporate TAP.
Table 3
The predictive performance of prognostic models
Variable | Combined model | Model without TAP |
AUC (95%CI) | 0.77(0.69, 0.84) | 0.68 (0.60, 0.77) |
Sensitivity, % | 64.1 | 47.4 |
Specificity, % | 80 | 83.3 |
Accuracy, % | 72.6 | 66.7 |
Positive predictive value, % | 73.5 | 71.2 |
Negative predictive value, % | 72 | 64.7 |
Positive likelihood ratio | 3.21 | 2.85 |
Negative likelihood ratio | 0.45 | 0.63 |
Diagnostic odd ratio | 7.14 | 4.51 |
AUC area under the curve, CI confidence interval |
The clinical value of this nomogram was evaluated by decision curve analysis (Fig. 3D). When the high-risk threshold was greater than 25% in the clinical decision, compared with none patient relapse pattern or all patients relapse pattern, both models is more benefit in predicting RFS. Furthermore, as for predicting RFS, the nomogram incorporating TAP, MVI, degree of tumor differentiation and maximun tumor diameter was more benefit than the model that only incorporated MVI, degree of tumor differentiation and maximun tumor diameter in this range.