Prognosis of spontaneously ruptured hepatocellular carcinoma: a propensity score matching study

Abstract

Purpose

Spontaneous rupture is a fatal complication of hepatocellular carcinoma (HCC). This study compared the prognosis of spontaneously ruptured HCC (srHCC) with that of non-ruptured HCC (nrHCC).

Methods

A total of 185 srHCC patients and 1085 nrHCC patients treated by hepatectomy between February 2005 and December 2017 at Zhongshan Hospital were retrospectively reviewed and enrolled. The overall survival (OS) and time to recurrence (TTR) were evaluated. A 1:2 propensity score matching (PSM) analysis was performed using the nearest neighbor matching with a caliper of 0.2.

Results

Before PSM, patients with srHCC who underwent hepatectomy (n = 185) had a poorer prognosis than those with nrHCC (n = 1085; 5-year OS, 39.1% vs 59.2%, P < 0.001; 5-year TTR, 83.8% vs 54.9%, P < 0.001). After PSM, patients with srHCC (n = 156) had higher 5-year TTR (83.2% vs 69.0%, P < 0.001) but comparable 5-year OS with those with nrHCC (n = 312, 44.0% vs 46.0%, P = 0.600). Univariate and multivariate analyses demonstrated spontaneous rupture as an independent risk factor for TTR (hazard ratio [HR], 1.681; 95% confidence interval [CI] 1.326–2.132; P < 0.001), but not for OS (HR: 1.074; 95% CI 0.823–1.401; P = 0.600). Further analysis revealed that srHCC was not appropriate to be assigned as T4 stage in American Joint Committee on Cancer classification.

Conclusion

Spontaneous rupture of HCC is not a risk factor for survival. If resected eventually, srHCC may achieve comparable survival with nrHCC.

Introduction

Hepatocellular carcinoma (HCC) comprises 75–85% of cases of liver cancer, which is the third leading cause of cancer mortality globally, causing 830,000 deaths each year (Sung et al. 2021). Approximately 85% of HCC patients can be attributed to viral hepatitis and liver cirrhosis caused by chronic infection of the hepatitis B virus or hepatitis C virus (Galle et al. 2018; Zhou et al. 2020). Spontaneous rupture is a unique fatal complication of HCC with an incidence of 3–25% and a mortality of 12–56% in the acute phase due to hemorrhage (Yoshida et al. 2016; Wang et al. 2022). Therefore, timely hemostasis and volume resuscitation play a pivotal role in the management of spontaneously ruptured hepatocellular carcinoma (srHCC).

Although spontaneous rupture may result in potential peritoneal dissemination of tumor cells, hepatectomy could provide an improved prognosis than non-surgical treatments (e.g., transarterial embolization [TAE] or chemoembolization [TACE]) (Lai 2006; Lee et al. 2022). Currently, the surgical regimens of srHCC include semi-elective hepatectomy, emergent hepatectomy, and sequential hepatectomy (laparotomy or TACE to achieve hemostasis, followed by hepatectomy) according to the hemodynamic status of patients (Huang et al. 2022). Several previous studies reported that the long-term prognosis of srHCC undergoing hepatectomy eventually was poorer than that of non-ruptured HCC (nrHCC) (Liu et al. 2001; Li et al. 2014; Kwon et al. 2021). However, given the small sample size and the obvious confounding factors between srHCC and nrHCC, whether HCC rupture will affect the efficacy of hepatectomy remains controversial.

In the present study, we performed propensity score matching (PSM) analysis to thoroughly investigate the influence of rupture on the prognosis of HCC treated by hepatectomy. In addition, the impact of spontaneous rupture on the classification of HCC was also evaluated.

Methods

Patient inclusion and data collection

srHCC patients treated by hepatectomy between February 2005 and December 2017 at Zhongshan Hospital were retrospectively reviewed and enrolled. nrHCC patients treated by hepatectomy at the same time were also reviewed and selected. The diagnosis and treatment strategies of srHCC have been discussed in detail in our former study (Huang et al. 2022). Baseline demographics, laboratory test results, pathological characteristics, and liver cirrhosis status were retrieved. Vascular invasion included portal vein tumor thrombus and microvascular invasion. This retrospective study has been censored and approved by the research department of Zhongshan Hospital, Fudan University.

Patient follow-up and surveillance for tumor recurrence and metastasis

Generally, patients were followed up 1 month after discharge and every 3 months thereafter. The follow‐up protocol included outpatient visits to the clinic, serum liver biochemistry, and imaging examinations (dynamic computed tomography [CT] or abdominal ultrasonography). Tumor progression, remission, recurrence, or metastasis was determined by dynamic CT images, chest films, and bone scanning. The last follow-up was made in December 2020.

PSM and statistical analysis

The PSM was done on R software (R Foundation for Statistical Computing, Vienna, Austria) using the nearest neighbor matching with a caliper of 0.2. Continuous variables were summarized as median (interquartile range) for non-normal distributed data, and means (standard deviation) for normally distributed variables. T-test and Mann Whitney U test were performed to compare the srHCC group and nrHCC group for variables with and without normal distribution, respectively. Categorical variables were analyzed as frequency (percentage) and compared between groups using the Chi-square test or Fisher’s exact test. All tests were two-tailed and a P value less than 0.05 was considered statistically significant. Survival curves were calculated using the Kaplan–Meier method. Univariate analysis was performed via the Cox regression model, and variables significant at P < 0.05 entered as candidate variables into the multivariate analysis.

Results

Patient demographics

A total of 185 srHCC patients who eventually underwent hepatectomy and 1085 nrHCC patients treated by hepatectomy at the same time were retrospectively enrolled. Young patients (P < 0.001) and large tumors (P < 0.001) were associated with higher rupture risk. Compared with nrHCC, srHCC was more advanced, i.e. larger tumor size (P < 0.001), a higher proportion of vascular invasion (P < 0.001), poorer differentiation (P < 0.001), more frequent extrahepatic metastasis (P = 0.001) and with poorer hepatic reserve function (P < 0.001, Table 1).

Table 1.

Demographic and clinicopathologic features of enrolled patients before and after PSM

Variables	Before PSM			After PSM
	srHCC (n = 185)	nrHCC (n = 1085)	P value	srHCC (n = 156)	nrHCC (n = 312)	P value
Gender			0.284			0.962
Female	24 (13.0)	178 (16.4)		21 (13.5)	44 (14.1)
Male	161 (87.0)	907 (83.6)		135 (86.5)	268 (85.9)
Age (year)			< 0.001			0.671
≤ 50	105 (56.8)	438 (40.4)		83 (53.2)	158 (50.6)
> 50	80 (43.2)	647 (59.6)		73 (46.8)	154 (49.4)
HBsAg			0.502			1.000
Negative	25 (13.5)	171 (15.8)		23 (14.7)	46 (14.7)
Positive	160 (86.5)	914 (84.2)		133 (85.3)	266 (85.3)
Cirrhosis			1.000			0.523
No	51 (27.6)	299 (27.6)		43 (27.6)	76 (24.4)
Yes	134 (72.4)	786 (72.4)		113 (72.4)	236 (75.6)
TB (μmol/L)	15.17 ± 8.01	13.46 ± 10.58	0.036	15.24 ± 7.64	13.68 ± 10.76	0.107
ALB (g/L)	38.51 ± 4.52	41.30 ± 3.77	< 0.001	39.38 ± 3.62	39.19 ± 3.21	0.565
PT (s)	12.26 ± 1.10	11.90 ± 0.97	< 0.001	12.15 ± 1.01	12.13 ± 1.04	0.839
ALT (U/L)	54.12 ± 58.82	44.56 ± 63.78	0.057	53.87 ± 60.17	46.51 ± 42.16	0.125
Child–Pugh			< 0.001			1.000
A	176 (95.1)	1076 (99.2)		153 (98.1)	307 (98.4)
B–C	9 (4.9)	9 (0.8)		3 (1.9)	5 (1.6)
AFP (ng/mL)			0.935			1.000
≤ 20	73 (39.5)	435 (40.1)		65 (41.7)	131 (42.0)
> 20	112 (60.5)	650 (59.9)		91 (58.3)	181 (58.0)
Tumor size (cm)	7.00 [5.50, 10.00]	4.50 [2.80, 8.00]	< 0.001	7.00 [5.47, 10.00]	7.00 [4.00, 10.00]	0.234
Tumor number			0.051			0.623
Single	138 (74.6)	880 (81.1)		120 (76.9)	232 (74.4)
Multiple	47 (25.4)	205 (18.9)		36 (23.1)	80 (25.6)
Edmondson stage			< 0.001			0.974
I–II	80 (43.2)	661 (60.9)		71 (45.5)	144 (46.2)
III–IV	105 (56.8)	424 (39.1)		85 (54.5)	168 (53.8)
Vascular invasion			< 0.001			0.869
No	74 (40.0)	655 (60.4)		67 (42.9)	130 (41.7)
Yes	111 (60.0)	430 (39.6)		89 (57.1)	182 (58.3)
Extrahepatic metastasis			0.001			1.000
No	178 (96.2)	1077 (99.3)		154 (98.7)	308 (98.7)
Yes	7 (3.8)	8 (0.7)		2 (1.3)	4 (1.3)
CNLC stage			< 0.001			0.964
I	111 (60.0)	846 (78.0)		104 (66.7)	208 (66.7)
II	37 (20.0)	118 (10.9)		28 (17.9)	54 (17.3)
III	37 (20.0)	121 (11.2)		24 (15.4)	50 (16.0)

Categorical variables were summarized as n (%); continuous variables were presented as mean ± standard deviation or median [interquartile range]

HBsAg hepatitis B surface antigen, TB total bilirubin, ALB albumin, PT prothrombin time, ALT alanine aminotransferase, AFP alpha-fetoprotein, CNLC China liver cancer

Prognosis of srHCC treated by hepatectomy before PSM

In general, srHCC could be safely treated by hepatectomy. There are no differences in the 30- (0.5% vs 1.2%; P = 0.680) and 90-day (3.8% vs 2.8%, P = 0.450) mortalities between srHCC and nrHCC patients. The 1-, 3-, and 5-year OS of srHCC patients were 76.3%, 47.9%, and 39.1% respectively, compared with 87.0%, 69.8%, and 59.2% of nrHCC patients (Fig. 1A; median, 34.9 vs 98.9 months; P < 0.001). The 1-, 3-, and 5-year TTR of srHCC patients were 54.4%, 76.0%, and 83.8% respectively, and 22.3%, 46.2%, and 54.9% of nrHCC patients (Fig. 1B; median, 11.1 vs 47.2 months; P < 0.001). srHCC was associated with higher peritoneal metastasis (PM) (17.8% vs 0.5%, P < 0.001) and intrahepatic recurrence (IM) (64.3% vs 44.7%, P < 0.001).

Fig. 1.

The prognosis of srHCC and nrHCC patients treated by hepatectomy. The 5-year OS (A) and TTR (B) before PSM and the 5-year OS (C) and TTR (D) after PSM

Prognosis of srHCC treated by hepatectomy after PSM

To reveal the true influence of tumor rupture on survival, PSM at a ratio of 1:2 was then performed to preclude the confounding bias factors. Totally, 15 clinicopathological variables which might impact prognosis were selected for matching and were well-balanced after PSM (Table 1). After PSM, the 1-, 3-, and 5-year OS of srHCC patients (n = 156) were 78.5%, 52.2%, and 44.0% respectively, compared with 81.5%, 59.0%, and 46.0% of nrHCC patients (n = 312) (Fig. 1C; median, 39.6 vs 49.5 months; P = 0.600). The 1-, 3-, and 5-year TTR of srHCC patients were 54.6%, 74.6%, and 83.2% respectively and in the nrHCC group were 30.5%, 57.7%, and 69.0% (Fig. 1D; median, 11.3 vs 25.2 months; P < 0.001). The PM (18.6% vs 1.0%, P < 0.001) and IM (65.4% vs 53.5%, P = 0.019) were still higher in srHCC patients after PSM.

Tumor rupture is an independent risk factor for recurrence

Univariate and multivariate analyses were then made to investigate the potential factors affecting the prognosis. Interestingly, spontaneous rupture was not a risk factor for OS (hazard ratio [HR], 1.074; 95% confidence interval [CI] 0.823–1.401; P = 0.600), while univariate analysis (HR, 1.696; 95% CI 1.343–2.140; P < 0.001) and multivariate analysis (HR, 1.681; 95% CI 1.326–2.132; P < 0.001) identified it as an independent risk factor for poor TTR. Other risk factors for poor OS included AFP level > 20 ng/mL (HR, 1.676; 95% CI 1.273–2.207; P < 0.001), tumor size > 5 cm (HR, 1.81; 95% CI 1.351–2.425; P < 0.001), multiple tumors (HR, 1.555; 95% CI 1.181–2.048; P < 0.001), vascular invasion (HR, 1.739; 95% CI 1.334–2.267; P < 0.001), and extrahepatic metastasis (HR, 10.034; 95% CI 4.291–23.466; P < 0.001) (Table 2).

Table 2.

Univariate and multivariate Cox regression analyses of clinicopathologic characteristics in HCC patients

Variables	OS		TTR
	HR (95% CI)	P value	HR (95% CI)	P value
Univariate analysis
Sex (female vs. male)	0.698 (0.474–1.029)	0.070	0.825 (0.595–1.143)	0.248
Age (year) (> 50 vs. ≤ 50)	0.949 (0.738–1.219)	0.679	0.873 (0.696–1.094)	0.238
HBsAg (positive vs. negative)	1.186 (0.813–1.728)	0.376	1.539 (1.076–2.199)	0.018
Liver cirrhosis (yes vs. no)	1.089 (0.814–1.456)	0.568	1.173 (0.903–1.523)	0.232
TB (μmol/L), continuous	1.001 (0.99–1.012)	0.873	1.003 (0.992–1.013)	0.634
ALB (g/L), continuous	0.972 (0.937–1.009)	0.132	0.994 (0.961–1.027)	0.715
PT (s), continuous	1.001 (0.998–1.003)	0.517	1 (0.998–1.003)	0.706
ALT (U/L), continuous	1.074 (0.953–1.21)	0.242	1.026 (0.915–1.15)	0.666
Child–Pugh class (A vs. B–C)	0.846 (0.315–2.273)	0.740	1.2 (0.535–2.693)	0.658
AFP (ng/mL) (> 20 vs. ≤ 20)	1.983 (1.515–2.595)	< 0.001	1.805 (1.425–2.287)	< 0.001
Tumor size (cm) (> 5 vs. ≤ 5)	1.914(1.433–2.555)	< 0.001	1.644(1.282–2.109)	< 0.001
Tumor number (multiple vs. single)	1.605 (1.222–2.109)	< 0.001	1.69 (1.319–2.166)	< 0.001
Edmondson stage (III–IV vs. I–II)	1.405 (1.09–1.811)	< 0.009	1.22 (0.972–1.531)	0.086
Vascular invasion (yes vs. no)	1.881(1.448–2.443)	< 0.001	1.532(1.217–1.929)	< 0.001
Extrahepatic metastasis (yes vs. no)	9.921 (4.347–22.643)	< 0.001	2.557 (0.814–8.03)	0.108
Rupture (yes vs. no)	1.074 (0.823–1.401)	0.600	1.696 (1.343–2.14)	< 0.001
Multivariate analysis
HBsAg (positive vs. negative)	NA	NA	1.447 (1.011–2.072)	0.043
AFP (ng/mL) (> 20 vs. ≤ 20)	1.676 (1.273–2.207)	< 0.001	1.627 (1.28–2.068)	< 0.001
Tumor size (cm) (> 5 vs. ≤ 5)	1.81 (1.351–2.425)	< 0.001	1.381 (1.069–1.784)	0.013
Tumor number (multiple vs. single)	1.555 (1.181–2.048)	0.002	1.73 (1.347–2.223)	< 0.001
Edmondson stage (III–IV vs. I–II)	1.295 (0.999–1.68)	0.051	NA	NA
Vascular invasion (yes vs. no)	1.739 (1.334–2.267)	< 0.001	1.525 (1.207–1.926)	< 0.001
Extrahepatic metastasis (yes vs. no)	10.034 (4.291–23.466)	< 0.001	NA	NA
Rupture (yes vs. no)	NA	NA	1.681 (1.326–2.132)	< 0.001

HBsAg hepatitis B surface antigen, TB total bilirubin, ALB albumin, PT prothrombin time, ALT alanine aminotransferase, AFP alpha-fetoprotein

Classification of srHCC

Currently, neither Barcelona Clinic Liver Cancer (BCLC) (Reig et al. 2022) nor China liver cancer (CNLC) (Zhou et al. 2020) staging system has taken rupture into classification except the American Joint Committee on Cancer (AJCC)/International Union Against Cancer (UICC) (Amin et al. 2017) staging system, which classifies tumor with perforation of visceral peritoneum as T4 stage. To figure out whether its impact has been overestimated, we first grouped patients with srHCC without using rupture as T4 classification factor and found srHCC at various stages displayed different 5-year OS (P < 0.001; Fig. 2A) and TTR (P = 0.005; Fig. 2B). Next, we compared nrHCC patients bearing T4 stage tumors (n = 113) with all srHCC patients (n = 185) and found srHCC patients had better 5-year OS (median, 39.6 vs 16.3 months; 39.1% vs 15.1%, P < 0.001; Fig. 2C) than and comparable 5-year TTR (median, 6.3 vs 11.3 months; 83.8% vs 86.7%, P = 0.470; Fig. 2D) with the patients with T4 stage nrHCC.

Fig. 2.

The influence of TNM classification on srHCC. The 5-year OS (A) and TTR (B) of srHCC patients stratified by TNM staging without using rupture as a T4 factor; the 5-year OS (C) and TTR (D) of srHCC patients and nrHCC patients with T4 stage tumors

We then evaluated another two classification systems. In BCLC, patients in stage A and B had similar survival (P > 0.05; Fig. 3A). While in CNLC, patients in II and III had comparable 5-year OS (P > 0.05; Fig. 3B). Notably, when using a tumor size of 5 cm as the cut-off value, which classified Ia and Ib HCC in CNLC, srHCC patients with tumor size ≤ 5 cm (n = 41) had poorer 5-year OS than nrHCC patients of the same kind (n = 632) (40.6% vs 72.3%, P < 0.001; Fig. 3C). While for tumor size > 5 cm, srHCC patients (n = 138) had comparable 5-year OS with nrHCC patients (n = 453) (38.7% vs 42.0%, P = 0.350; Fig. 3D). We questioned such difference by investigating the association between tumor size and prognosis within each group and found for srHCC, tumor size did not impact OS (≤ 5 cm vs > 5 cm: median, 39.6 vs 31.2 months; P = 0.330). Consistently, cox regression analysis revealed that tumor size (univariate analysis: HR, 1.272; 95% CI 0.784–2.062; P = 0.330) was not a risk factor for OS of srHCC.

Fig. 3.

The BCLC and CNLC classifications and subgroup analysis of srHCC patients. The 5-year OS of srHCC patients stratified by BCLC (A) and CNLC (B); the 5-year OS of srHCC and nrHCC patients with tumor size ≤ 5 cm (C) and > 5 cm (D)

Discussion

Spontaneous rupture of HCC is a potentially life-threatening complication caused by overpressure inside the tumor, especially in the cirrhotic liver, with the symptom of sudden-onset abdominal pain and the sign of hemodynamic instability (Huang et al. 2022). Accumulating studies have reported that srHCC patients, unless not eligible for surgery, should receive hepatectomy to achieve long-term survival (Huang et al. 2022; Lai 2006; Park et al. 2022). The prognosis of srHCC patients has steadily ameliorated along with the improvement of diagnosis, surgical technique, and perioperative management (Ong and Taw 1972; Moris et al. 2018).

Although previous retrospective studies almost unexceptionally reported significantly poorer survival of srHCC patients compared with nrHCC patients treated by hepatectomy (Liu et al. 2001; Li et al. 2014), it should be cautiously interpreted since srHCC are usually associated with poorer performance status, more damaged liver function, and more aggressive tumor characteristics. Studies with small case number recorded similar long-term results with elective hepatectomy (Vergara et al. 2000; Joliat et al. 2018). Recently, several studies performed PSM to balance the significant differences in patient characteristics, whereas the results were still controversial: two small sample size studies (Lee et al. 2014; Sada et al. 2016), a multicenter analysis study (Tanaka et al. 2016), and another with a 1:2 PSM design (Chua et al. 2019) reported comparable long-term survival. However, the largest study heretofore which employed 1:1 PSM with 89 patients in each group identified spontaneous rupture as an independent risk factor for OS (Zhu et al. 2019). Herein, with large sample size and 1:2 PSM ratio, we found that comparable OS was displayed between srHCC and nrHCC patients, while TTR in the srHCC group was significantly shorter than that in the nrHCC group. This indicated more intensive surveillance of tumor recurrence should be performed for srHCC patients during follow-up.

The AJCC/UICC TNM classification is the only staging system that incorporates tumor rupture as a factor, with T4 defined as tumors with direct invasion of adjacent organs other than the gallbladder, or perforation of visceral peritoneum. However, it may be problematic to categorize all srHCC into T4 stage since we found srHCC patients showed significantly superior survival than nrHCC patients with T4 stage tumor, which was in line with previous studies (Mizuno et al. 2004; Yoshida et al. 2008; Chan et al. 2016a, b). After the exclusion of rupture as a T4 factor, we also noted different prognosis for srHCC patients at various stages (Hiraoka et al. 2015). A future revision of the AJCC/UICC TNM staging of HCC may need to consider removing rupture as a T4 factor. Another factor affecting prognosis is the tumor size. Previous studies reported that rupture affected the outcomes of hepatectomy for those with tumor size ≤ 10 cm (Chan et al. 2016a). Nevertheless, through in-depth analysis, it was identified that srHCC had significantly dismal outcomes compared with nrHCC in the subgroup of patients with tumor size ≤ 5 cm. It indicated that even though rupture didn’t affect HCC patients in terms of survival on the whole, it still had an influence on the survival of patients with small tumors.

There were several limitations of the current study. Firstly, although we conducted PSM analysis to minimize the influence of confounding factors on results, the retrospective single-center design inevitably led to unintentional biases. Considering the characteristics of the acute onset of srHCC, it is not practical and unethical to perform a prospective clinical trial. Thus, a multicenter study with an adequate sample size is needed to validate the results. Secondly, the data of patients treated by hepatectomy between February 2005 and December 2017 was collected. During this period, the surgical technique and treatment strategies for HCC developed rapidly. Therefore, there were deviations among the factors of different treatments for each patient. In addition, the predominant etiology of our cohort was hepatitis B virus infection, which is significantly different from that of western countries. Further studies should also enroll HCC patients with other etiologies.

Conclusions

Herein, we provided a thorough analysis of the long-term survival of srHCC. Spontaneous rupture of HCC is not a risk factor for OS and a comparable prognosis with nrHCC could be achieved by hepatectomy. Furthermore, it may be not appropriate to routinely categorize ruptured HCC into the T4 stage.

Author contributions

AH, JZ and S-YZ conceptualized this study and wrote the manuscript. S-YZ and D-ZG conducted data analysis. XZ collected the clinical information of patients. AH, JZ and JF interpreted the data. All authors edited the manuscript. All authors contributed to the article and approved the submitted version.

Funding

This study was jointly supported by the National Key R&D Program of China (2019YFC1315800, 2019YFC1315802), National Natural Science Foundation of China (No.81830102, 81772578, 81802991), and Shanghai Municipal Key Clinical Specialty.

Data availability

The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Conflict of interest

The authors have no relevant financial or non-financial interests to disclose.

Ethical approval

This retrospective study has been censored and approved by the scientific research department of Zhongshan Hospital, Fudan University.

Consent to participate

Informed consent was obtained from all individual participants included in the study.

Consent to publish

Not applicable.