Patient characteristics
A total of 370 patients with HCC who underwent liver transplantation at our center were enrolled in the analytic cohort, and the enrollment flowchart is presented in Fig. 2. The mean age of all patients was 52.9 ± 9.0 years at the time of liver transplantation, and the majority of patients were male (n = 328, 88.6%). Most patients had hepatitis B (n = 337, 91.1%) and cirrhosis (n = 357, 96.5%). About three-quarters of the patients had no more than three tumors (n = 286, 77.3%), and the largest tumor was usually no more than 5 cm (n = 267, 72.2%). Most patients had compensatory liver function manifested as Child-Pugh class A and B (n = 310, 83.8%). Nearly half of the patients exceeded the Milan criteria (n = 194, 52.4%).
Our study identified 64 patients in the micronecrosis(-) group and 306 patients in the micronecrosis(+) group (Nscore = 1, n = 92; Nscore = 2, n = 95; Nscore = 3, n = 119). The demographic and clinicopathological characteristics of the two groups are shown (Table 1). We followed up the patients with a median of 32.2 months (range: 0.03–85.0 months). After excluding patients who had received preoperative treatment, 80 of 116 HCC patients were present with tumor micronecrosis (Table 2).
Table 1
Demographics and clinicopathological characteristics of patients with HCC receiving liver transplantation
Characteristics | Total (n = 370) | Micronecrosis (-) (n = 64) | Micronecrosis (+) (n = 306) | P-value |
Age, years | 52.9 ± 9.0 | 52.9 ± 9.1 | 52.9 ± 9.0 | 0.993 |
Gender, n (%) | | | | 0.106 |
Male | 328 (88.6%) | 53 (83.8%) | 275 (89.9%) | |
Female | 42 (11.4%) | 11 (16.2%) | 31 (10.1%) | |
Etiology, n (%) | | | | 0.011 |
HBV infection | 337 (91.1%) | 53 (82.8%) | 284 (92.8%) | |
others | 33 (8.9%) | 11 (17.2%) | 22 (7.2%) | |
Cirrhosis, n (%) | 357 (96.5%) | 63 (98.4%) | 294 (96.1%) | 0.707* |
AFP, ng/mL | | | | 0.354 |
AFP ≤ 400 | 278 (75.1%) | 51 (79.7%) | 227 (74.2%) | |
AFP > 400 | 92 (24.9%) | 13 (20.3%) | 79 (25.8%) | |
RDW | 15.3 ± 2.8 | 16.0 ± 3.0 | 15.2 ± 2.7 | 0.035 |
NLR | 4.80 ± 5.48 | 3.81 ± 4.32 | 4.89 ± 5.68 | 0.150 |
INR | 1.28 ± 0.28 | 1.39 ± 0.30 | 1.25 ± 0.28 | 0.001 |
TNM stage, n (%) | | | | < 0.001* |
I | 137 (37.0%) | 39 (60.9%) | 98 (32.0%) | |
II | 128 (34.6%) | 17 (26.6%) | 111 (36.3%) | |
III | 90 (24.3%) | 8 (12.5%) | 82 (26.8%) | |
IV | 15 (4.1%) | 0 (0.0%) | 15 (4.1%) | |
Tumor number, n (%) | | | | 0.070 |
≤ 3 | 286 (77.3%) | 55 (85.9%) | 231 (75.5%) | |
> 3 | 84 (22.7%) | 9 (14.1%) | 75 (24.5%) | |
Size of largest tumor, cm | | | | 0.003 |
≤ 5 | 267 (72.2%) | 56 (87.5%) | 211 (69.0%) | |
> 5 | 103 (27.8%) | 8 (12.5%) | 95 (31.0%) | |
Differentiation, n (%) | | | | 0.023* |
Well | 38 (10.3%) | 7 (10.9%) | 31 (10.1%) | |
Moderate | 197 (53.2%) | 41 (64.1%) | 156 (51.0%) | |
Poor | 128 (34.6%) | 13 (20.3%) | 115 (37.6%) | |
NA | 7 (1.9%) | 3 (4.7%) | 4 (1.3%) | |
Macrovascular invasion, n (%) | | | | 0.015 |
Present | 90 (24.3%) | 8 (12.5%) | 82 (26.8%) | |
Absent | 280 (75.7%) | 56 (87.5%) | 224 (73.2%) | |
Child-Pugh class, n (%) | | | | 0.047 |
A (5–6) | 159 (43.0%) | 19 (29.7%) | 140 (45.8%) | |
B (7–9) | 151 (40.8%) | 34 (53.1%) | 117 (38.2%) | |
C (≥ 10) | 60 (16.2%) | 11 (17.2%) | 49 (16.0%) | |
Milan criteria, n (%) | | | | < 0.001 |
Within | 176 (47.6%) | 45 (70.3%) | 131 (42.8%) | |
Out | 194 (52.4%) | 19 (29.7%) | 175 (57.2%) | |
MELD score | 11 (6–40) | 14 (6–28) | 10 (6–40) | 0.011 |
Preoperative treatment, n (%) | | | | < 0.001 |
Yes | 254 (68.6%) | 28 (43.8%) | 226 (73.9%) | |
Liver resection | 18 (4.9%) | 4 (6.3%) | 14 (4.6%) | 0.529* |
TACE | 80 (21.6%) | 9 (14.1%) | 71 (23.2%) | 0.106 |
RFA | 15 (4.1%) | 0 | 15 (4.9%) | 0.048* |
Targeted therapy | 2 (0.5%) | 0 | 2 (0.7%) | 1.000* |
Radiotherapy | 3 (0.8%) | 1 (1.6%) | 2 (0.7%) | 0.528* |
Other | 19 (5.1%) | 2 (3.1%) | 17 (5.6%) | 0.548* |
multiple | 117 (31.6%) | 12 (18.8%) | 105 (34.3%) | 0.015 |
No | 116 (31.4%) | 36 (56.3%) | 80 (26.1%) | |
Recurrence, n (%) | 109 (29.5%) | 11 (17.2%) | 98 (32.0%) | 0.018 |
Intrahepatic | 40 (36.7%) | 1 (9.1%) | 39 (39.8%) | 0.084 |
Thorax | 43 (39.4%) | 2 (18.2) | 41 (41.8%) | 0.115 |
Bone | 19 (17.4%) | 3 (27.3) | 16 (16.3%) | 0.390 |
Abdominal cavity | 40 (36.7%) | 6 (54.4%) | 34 (34.7%) | 0.184 |
Multiple | 29 (26.6%) | 1 (9.1%) | 28 (28.6%) | 0.274 |
Others | 7 (6.4%) | 1 (9.1%) | 6 (6.1%) | |
Death, n (%) | 116 (31.4%) | 17 (26.6%) | 99 (32.4%) | 0.364 |
* Fisher exact test was performed for the comparison. |
AFP, alpha-fetoprotein; RDW, red cell volume distribution width; NLR, neutrophil-to-lymphocyte ratio; INR, international normalized ratio; MELD, model for end-stage liver disease. |
Table 2
Baseline characteristics of HCC patients receiving liver transplantation without preoperative treatment
Characteristics | Total (n = 116) | Micronecrosis (-) (n = 36) | Micronecrosis (+) (n = 80) | P-value |
Age, years | 52.0 ± 9.3 | 52.6 ± 9.7 | 51.7 ± 9.1 | 0.618 |
Gender, n (%) | | | | 0.400 |
Male | 104 (88.6%) | 31 (86.1%) | 73 (91.3%) | |
Female | 12 (10.3%) | 5 (13.9%) | 7 (8.8%) | |
Etiology, n (%) | | | | 0.077 |
HBV infection | 100 (86.2%) | 28 (77.8%) | 72 (90.0%) | |
others | 16 (13.8%) | 8 (22.2%) | 8 (10.0%) | |
Cirrhosis, n (%) | 113 (97.4%) | 35 (97.2%) | 78 (97.5%) | 1.000* |
AFP, ng/mL, n (%) | | | | 0.178 |
AFP ≤ 400 | 91 (78.4%) | 31 (86.1%) | 60 (75.0%) | |
AFP > 400 | 25 (21.6%) | 5 (13.9%) | 20 (25.0%) | |
RDW | 15.8 ± 3.1 | 16.2 ± 3.1 | 15.6 ± 3.1 | 0.312 |
NLR | 4.39 ± 5.53 | 4.60 ± 6.03 | 4.89 ± 5.68 | 0.529 |
INR | 1.41 ± 0.33 | 1.46 ± 0.29 | 1.39 ± 0.35 | 0.305 |
TNM stage, n (%) | | | | < 0.001* |
I | 81 (69.8%) | 32 (88.9%) | 49 (61.3%) | |
II | 2 (1.7%) | 2 (5.6%) | 0 (0.0%) | |
III | 33 (28.4%) | 2 (5.6%) | 31 (38.8%) | |
Tumor number, n (%) | | | | 0.011 |
≤ 3 | 93 (80.2%) | 34 (94.4%) | 59 (73.8%) | |
> 3 | 23 (19.8%) | 2 (5.6%) | 21 (26.3%) | |
Size of largest tumor, cm | | | | 0.059 |
≤ 5 | 83 (71.6%) | 30 (87.5%) | 53 (66.3%) | |
> 5 | 33 (28.4%) | 6 (12.5%) | 27 (33.8%) | |
Differentiation, n (%) | | | | 0.020* |
Well | 5 (4.3%) | 4 (11.1%) | 1 (1.3%) | |
Moderate | 71 (61.2%) | 26 (72.2%) | 45 (56.3%) | |
Poor | 36 (31.0%) | 5 (13.9%) | 31 (38.8%) | |
NA | 4 (3.4%) | 1 (2.8%) | 3 (3.8%) | |
Macrovascular invasion, n (%) | | | | 0.003 |
Present | 31 (26.7%) | 3 (8.3%) | 28 (35.0%) | |
Absent | 85 (73.3%) | 33 (91.7%) | 52 (65.0%) | |
Child-Pugh class, n (%) | | | | 0.987 |
A (5–6) | 30 (25.9%) | 9 (25.0%) | 21 (26.3%) | |
B (7–9) | 60 (51.7%) | 19 (52.8%) | 41 (51.2%) | |
C (≥ 10) | 26 (22.4%) | 8 (22.2%) | 18 (22.5%) | |
Milan criteria, n (%) | | | | < 0.001 |
Within | 61 (52.6%) | 28 (77.8%) | 33 (41.3%) | |
Out | 55 (47.4%) | 8 (22.2%) | 47 (58.8%) | |
MELD score | 11 (6–40) | 14 (6–28) | 10 (6–40) | 0.758 |
Recurrence, n (%) | 41 (35.3%) | 6 (16.7%) | 35 (43.8%) | 0.005 |
Death, n (%) | 32 (27.6%) | 5 (13.9%) | 27 (33.8%) | 0.027 |
Baseline comparison of micronecrosis(+) and micronecrosis(-) groups
Patients with tumor micronecrosis were more likely to have advanced tumor-node-metastasis (TNM) stage (P < 0.001), larger tumor size (P = 0.003), poorer differentiation (P = 0.023), and more frequent macrovascular invasion (P = 0.015). More patients in the micronecrosis(+) group exceeded Milan criteria and received preoperative treatments (P < 0.001 for both) than patients in the micronecrosis(-) group, and the MELD score was lower in the micronecrosis(+) group than in the micronecrosis(-) group (P = 0.011). Patients in the micronecrosis(+) group developed hepatitis B virus infection more often than those in the micronecrosis(-) group (P = 0.011), had lower red cell volume distribution width (RDW) and international normalized ratio (INR) (P = 0.035 and 0.001, respectively) (Table 1).
In patients without preoperative treatment, patients with micronecrosis had more tumor nodules (P = 0.011), poorer differentiation (P = 0.020) and more frequent macrovascular invasion (P = 0.003) than patients without tumor micronecrosis. More patients in the micronecrosis(+) group exceeded Milan criteria and had a higher TNM stage than the micronecrosis(-) group (both P < 0.001) (Table 2). Moreover, the presence of tumor micronecrosis was associated with TNM stage (r = 0.312, P = 0.001, Pearson’s chi-squared test), tumor nodules (r = 0.240, P = 0.009), tumor differentiation (r = 0.223, P = 0.018), macrovascular invasion (r = 0.279, P = 0.002), and Milan criteria (r = 0.338, P < 0.001).
Survival outcomes of micronecrosis(+) and micronecrosis(-) groups
The median follow-up was 32.2 months. During the follow-up, tumors were found in 11 patients (17.2%) from the micronecrosis(-) group and 98 patients (32.0%) from the micronecrosis(+) group (P = 0.018). The 1-, 3-, and 5-year RFS rates in the micronecrosis(-) group were 77.0%, 70.5%, and 64.4%, respectively, which were higher than those in the micronecrosis(+) group, 62.4%, 54.7%, and 49.3%, respectively (P = 0.037) (Fig. 3A). Seventeen patients (26.6%) in the micronecrosis (-) group and 99 patients (32.4%) in the micronecrosis(+) group died (P = 0.364). No significant difference in the 1-, 3-, and 5-year OS rate was observed between patients with tumor micronecrosis (78.7%, 64.4%, and 57.5%) and those without tumor micronecrosis (83.6%, 75.3%, and 61.7%, P = 0.307) (Fig. 3B). Tumors recurred less frequently in intrahepatic sites in the micronecrosis(-) patients; local recurrence was found in one patient (9.1%) in the micronecrosis(-) group and 39 patients (39.8%) in the micronecrosis(+) group (P = 0.84) (Fig. 3C).
We observed even more significant differences in prognosis between patients with and without tumor micronecrosis among untreated patients in our classified population-based study depending on whether they had preoperative antitumor treatment. The 1-, 3-, and 5-year RFS rates in patients with tumor micronecrosis were 62.1%, 56.3%, and 50.6%, respectively, which were much lower than 82.3%, 82.3%, and 82.3%, respectively, in patients without tumor micronecrosis (P = 0.016). The 1-, 3-, and 5-year OS rates in the micronecrosis(+) group (76.6%, 67.8%, and 56.5%) were poorer than those in the micronecrosis(-) group (84.8%, 84.8%, and 84.8%, P = 0.040). The degree of tumor micronecrosis was also associated with RFS (Nscore = 0 vs. Nscore = 1 vs. Nscore = 2 and 3, P = 0.002) and OS (Nscore = 0 vs. Nscore = 1 vs. Nscore = 2 and 3, P = 0.007) (Fig. 4A and B). The prognosis of high-score necrosis was poorer than the low-score necrosis.
However, there was no difference in the prognosis between patients with and without tumor micronecrosis who received preoperative treatments. The 1-, 3-, and 5-year RFS rates in the micronecrosis(-) group were 70.1%, 60.7%, and 43.3%, respectively, which were comparable to 63.5%, 54.8%, and 49.5%, respectively, in the micronecrosis(+) group (P = 0.666). The 1-, 3-, and 5-year OS rates in the micronecrosis(-) group (81.6%, 63.3%, and 43.5%, respectively) were similar to those in the micronecrosis(+) group (80.1%, 63.4%, and 58.5%, respectively; P = 0.686) (Fig. 4C and D).
To explain this phenomenon, we analyzed the preoperative treatment conditions of patients (Table 1 and Fig. 4E). We did not find significant difference in therapies of liver resection, transcatheter arterial chemoembolization (TACE), radiofrequency ablation (RFA), targeted therapy and radiotherapy alone in patients with tumor micronecrosis; however, more patients in the micronecrosis(+) group received comprehensive treatment (P = 0.015). We also evaluated the relationship between preoperative treatment and tumor micronecrosis, and correlation analysis showed that preoperative antitumor treatment was positively correlated with tumor micronecrosis (r = 0.245, P < 0.001, Pearson correlation). The result indicated that multiple bridging treatments may affect necrosis scores, and we should analyze patients differently according to their treatment situation.
Independent risk factors for the prognosis of patients without preoperative treatment
We carefully chose the variables to establish the Cox regression model of RFS. After univariate analysis, variables with P < 0.1 of serum AFP level, neutrophil-to-lymphocyte ratio, tumor number, size of the largest tumor, tumor differentiation, Child-Pugh class, and tumor micronecrosis were included in the multivariate Cox regression model. In terms of RFS, we identified tumor number ≥ 3 (HR, 2.905; 95% CI, 1.227–6.875; P = 0.015) and advanced Child-Pugh class (HR, 6.692; 95% CI, 2.012–22.262; P = 0.004) as independent prognostic factors in patients with HCC after liver transplantation. Even though tumor micronecrosis increased the possibility of poor prognosis, it was not an independent prognostic factor after liver transplantation (Table 3).
Table 3
Univariate and multivariate analyses of prognostic factors of RFS in patients without preoperative anti-tumor treatment
| Univariate analysis | Multivariate analysis |
Variables | HR (95% CI) | P-value | HR (95% CI) | P-value |
Age | 1.005 (0.972–1.039) | 0.776 | | |
Gender | 2.440 (0.589–10.116) | 0.219 | | |
HBV infection | 1.659 (0.591–4.656) | 0.336 | | |
AFP > 400 ng/mL | 1.909 (0.986–3.694) | 0.055 | 1.032 (0.451–2.363) | 0.941 |
RDW | 1.067 (0.973–1.170) | 0.166 | | |
NLR | 1.052 (1.019–1.086) | 0.002 | 1.022 (0.983–1.062) | 0.273 |
INR | 0.906 (0.340–2.410) | 0.843 | | |
Size of the largest tumor | 2.885 (1.556–5.347) | 0.001 | 2.115 (0.996–4.495) | 0.051 |
Tumor number | 4.809 (2.558–9.040) | < 0.001 | 2.905 (1.227–6.875) | 0.015 |
Poor differentiation | 4.032 (2.121–7.665) | < 0.001 | 1.894 (0.895–4.009) | 0.095 |
Macrovascular invasion | 4.521 (2.429–8.415) | < 0.001 | 2.360 (0.919–6.058) | 0.074 |
Child-Pugh class | | 0.039 | | 0.004 |
A (5–6) | Reference | | Reference | |
B (7–9) | 2.606 (0.990–6.861) | 0.052 | 6.692 (2.012–22.262) | 0.002 |
C (≥ 10) | 3.834 (1.364–10.776) | 0.011 | 8.160 (2.279–29.212) | 0.001 |
Micronecrosis | 3.021 (1.270–7.184) | 0.012 | 1.482 (0.588–3.735) | 0.404 |
Tumor micronecrosis adjusted to prognosis models
We used tumor micronecrosis to illustrate the potential incremental value above the frequently used survival prediction models of liver transplantation. Models incorporating micronecrosis displayed better predictive accuracy in predicting 5-year RFS in patients with HCC after liver transplantation than models based on the Milan criteria, MELD score, and AFP level alone. The AUC values of Milan criteria, MELD score, and AFP level were 0.705, 0.682, and 0.566, respectively. For adjusted Cox regression models, tumor micronecrosis improved the AUC value of Milan criteria to 0.738, the AUC value of MELD score to 0.738, and the AUC value of AFP level to 0.664. The former two models had moderate accuracy and were considered clinically useful (Fig. 5).