Resection combined with hyperthermic intraperitoneal chemotherapy in the treatment of spontaneously ruptured hepatocellular carcinoma: a multicenter retrospective study

Abstract

Background:

Spontaneous tumor rupture is a unique and life-threatening presentation of hepatocellular carcinoma (HCC). The optimal postoperative management of patients with spontaneously ruptured HCC (srHCC) remains controversial. Hyperthermic intraperitoneal chemotherapy (HIPEC) has been proposed to reduce peritoneal dissemination, but its clinical benefit in srHCC is uncertain.

Objectives:

This study aimed to evaluate the survival benefit and safety of postoperative HIPEC combined with hepatic resection in patients with srHCC.

Design:

A retrospective multicenter cohort study was conducted, including patients with srHCC who underwent curative hepatectomy with or without postoperative HIPEC between 2018 and 2024.

Methods:

A total of 208 srHCC patients from 4 institutions were enrolled and categorized into the resection group (R) and the resection plus HIPEC group (R-HIPEC). Propensity score matching (PSM) and inverse probability of treatment weighting (IPTW) were applied to minimize baseline differences. The primary endpoint was recurrence-free survival (RFS), and the secondary endpoint was overall survival (OS). Survival outcomes were assessed using Kaplan–Meier analysis, Cox proportional hazards models, and subgroup analysis.

Results:

Across the primary, PSM, and IPTW cohorts, patients in the R-HIPEC group achieved significantly longer median RFS (mRFS) and OS than those in the R group. The median OS was 45.6 versus 26.4 months in the primary cohort (p = 0.025), 48.2 versus 26.4 months in the PSM cohort (p = 0.025), and 42.9 versus 26.5 months in the IPTW cohort (p = 0.012). The mRFS was 15.5 versus 7.7 months (p = 0.002), 18.2 versus 8.3 months (p = 0.002), and 14.7 versus 7.4 months (p = 0.014), respectively. Subgroup analysis indicated that patients with Barcelona Clinic Liver Cancer stage 0/A derived significantly greater RFS benefit from HIPEC than those with stage B/C (interaction p = 0.0264). For OS, a significant interaction was observed with postoperative immunotherapy (interaction p = 0.0054). The R-HIPEC group showed a lower incidence of peritoneal implantation metastasis, without an increase in perioperative complications.

Conclusion:

HIPEC combined with resection for srHCC can effectively prolong survival time. Resection combined with HIPEC and targeted therapy may be a promising strategy for srHCC.

Plain language summary

Surgery combined with heated chemotherapy inside the abdomen may improve survival for people with ruptured hepatocellular carcinoma

Background:

Spontaneous rupture of hepatocellular carcinoma (HCC) is rare but life-threatening. The optimal post-surgical management is unclear. Hyperthermic intraperitoneal chemotherapy (HIPEC), which delivers heated chemotherapy into the abdomen during surgery, may help prevent cancer spread, but its benefit in ruptured HCC is uncertain.

Objective:

To assess whether adding HIPEC to surgery improves survival and is safe in patients with ruptured HCC.

Design:

This multicenter retrospective study (four centers, 2018–2024) included 208 patients who underwent surgery for ruptured HCC. Patients received either surgery alone or surgery plus HIPEC. Survival outcomes were compared between groups.

Methods:

Patients were divided into a resection-only group (R) and a resection plus HIPEC group (R-HIPEC). Statistical methods were used to balance baseline characteristics. Recurrence-free survival (RFS) and overall survival (OS) were analyzed.

Results:

Patients in the R-HIPEC group had significantly better outcomes. Median OS was 45.6 months with HIPEC versus 26.4 months without it; median RFS was 15.5 versus 7.7 months. These advantages persisted after adjustment. Subgroup analysis showed greater RFS benefit from HIPEC in early-stage HCC (BCLC 0/A) than in more advanced stages (B/C). HIPEC also appeared to lower the risk of intraperitoneal spread without increasing perioperative complications.

Conclusion:

Adding HIPEC to surgery for ruptured HCC improves survival without increasing complications. Combined with targeted therapy, it may be a promising treatment strategy for these patients.

Introduction

Hepatocellular carcinoma (HCC) is the most common liver cancer and the fourth most common cause of cancer-related death worldwide. ¹ Spontaneous rupture of primary HCC is a frequently observed and fatal complication. ² It is often accompanied by acute hemorrhagic shock, with an acute-phase mortality rate ranging from 25% to 75%. ³ Secondary dissemination can lead to intra-abdominal organ and peritoneal metastases, resulting in a poorer prognosis. ⁴ Although HCC rupture is a fatal complication, a multimodal approach has contributed to better management of ruptured cases. ⁵

Currently, conservative treatment, emergency surgical resection, transarterial embolization, or hyperthermic intraperitoneal perfusion chemotherapy (HIPEC) may be selected for spontaneously ruptured HCC (srHCC) based on various factors, such as hemodynamic status, underlying liver function, tumor characteristics, and staging.^6,7 Among these treatment methods for srHCC, the value of HIPEC is still unclear. HIPEC, as an adjuvant treatment for abdominal malignant tumors, has achieved promising outcomes in the treatment of advanced gastric cancer peritoneal metastases, recurrent and metastatic colorectal cancer, and primary peritoneal tumors.^8,9 An increasing number of studies have reported the application of this treatment to hepatobiliary and pancreatic malignancies.^8,10 However, this practice remains relatively rare and controversial in srHCC. In one study including 57 patients with srHCC, the HIPEC group exhibited more postoperative complications than the non-HIPEC group, with no significant differences in disease-free survival or overall survival (OS) between the two groups. ¹¹

This study retrospectively analyzed 208 patients with srHCC treated across 4 institutions between January 2018 and December 2023. Clinical and pathological data were reviewed to evaluate the impact of resection with or without HIPEC, providing evidence for its potential role in the management of srHCC.

Patients and methods

Patients

Clinical data from srHCC patients admitted to Hunan Province People’s Hospital, Zhujiang Hospital of Southern Medical University, People’s Hospital of Guangxi Zhuang Autonomous Region, and South China University of the Second Affiliated Hospital were retrospectively collected for analysis were enrolled for retrospective analysis. This retrospective study was conducted in accordance with the ethical principles of the Declaration of Helsinki and the STROBE guidelines ¹² (Supplemental Table 1). The study protocol was approved by the Ethics Committee of Hunan Provincial People’s Hospital ([2024]-348) and the Ethics Committee of Zhujiang Hospital of Southern Medical University (2023-KY-078-02). According to institutional regulations, the other two participating hospitals were exempted from separate ethical review once approval from the primary center had been obtained, and the study was registered at those institutions accordingly. As this was a retrospective study, the requirement for informed consent was waived by all participating ethics committees. The inclusion criteria were as follows: (1) preoperative diagnosis of srHCC by clinical manifestations, laboratory findings, and imaging; (2) postoperative diagnosis of HCC by pathological analysis; (3) aged 18–80 years; (4) no interventional, surgical, or systemic treatments for the tumor before initial surgery; (5) completed and signed the informed consent form for surgery; and (6) without absolute contraindications to surgery.

The exclusion criteria were as follows: (1) No surgical treatment was performed and (2) incomplete follow-up data.

Methods

All patients were treated to ensure that there were no absolute contraindications prior to treatment. Antishock agents and oxygen were administered, and corrective measures were implemented for acid-base and water-electrolyte imbalances and maintenance of vital signs.

Surgical procedure

All surgical procedures were performed under general endotracheal anesthesia. Surgical access was achieved through either a J-shaped laparotomy incision or laparoscopic exploration. Depending on the location of the tumor, the volume of the remaining liver, and the conditions at the time of the operation, the appropriate radical surgical resection of liver cancer was selected. Tumor margins were required to be greater than 1 cm. The patient’s blood pressure and central venous pressure were closely monitored; if necessary, the patient’s blood pressure was elevated, and the central venous pressure was lowered to reduce blood seepage from the incision for the benefit of the operation. Delicate handling of liver tissue was used to avoid damage to the organ tissue and blood vessels. The hepatic portal blocking technique was reasonably applied, and the blocking time was not more than 15 min/time. After surgery, the abdominal cavity was rinsed with 3 L of warm sterile saline. The patients in the surgery group received one to two peritoneal drainage tubes in the liver section, while those in the resection with HIPEC (R-HIPEC) group received four abdominal heat perfusion tubes placed at the hepatic apex, splenic fossa, and left and right pelvic cavities.

R-HIPEC procedure

After the operation, when the patient’s vital signs were stable, HIPEC was performed 1–3 times, depending on the patient’s condition. The machine used was BR-TRG-II (Guangzhou Poly Medical Technology Co., Ltd., Guangzhou, Guangdong, China). An external BR-TRG-II continuous intraperitoneal hyperthermic perfusion chemotherapy device was connected. The perfusion temperature was set to 43 ± 0.5°C. Perfusion with cisplatin was started at a flow rate of 400 mL/min to cause the entire abdomen to be filled with the perfusate. The total dose of cisplatin was 75 mg/m², administered in three separate infusions. The HIPEC lasted for 60 min each time. ⁸ The patients received intraperitoneal hyperthermic perfusion chemotherapy 3 times after the abdomen was closed, on days 0, 2, and 4 after the operation.

Other treatments

The doctor administered interventional therapy, targeted therapy, immunotherapy, or other traditional Chinese patent drugs (Huaier Granules, etc.) according to the needs of the patients and condition of each patient. There is no restriction on the types of drugs used according to the actual situation, including Lenvatinib: 8 mg, po, Qd; Donafenib: 200 mg, po, Bid; and Surufatinib: 200 mg, po, Qd. There is no restriction on the types of immunotherapy drugs used according to the actual situation, including Tislelizumab, 200 mg, iv, once every 3 weeks (Q3W), Camrelizumab, 200 mg, iv, Q3W, Sintilimab, 200 mg, iv, Q3W, and Atezolizumab, 1200 mg, iv, Q3W.

Follow-up

Postoperative discharge follow-up was conducted through outpatient clinic visits and telephone calls. Follow-up began at the time of review in the first postoperative month, after which patients were followed every 1–3 months in the first year and then every 3–6 months from the second year onward. Follow-up was completed after the patient’s death or on May 15, 2024. During each review, serum levels of alpha-fetoprotein (AFP) were examined, and abdominal ultrasonography and chest and abdomen CT or MRI were performed. Recurrence and metastasis were determined based on the identification of de novo nodules or inhomogeneous enhancement foci in the CT or MRI findings. The primary endpoint of this study was recurrence-free survival (RFS), and the secondary endpoint was OS. RFS was defined as the time from surgery to the first recurrence or the last follow-up without recurrence. OS was defined as the time from surgery to the date of death, the date of the last follow-up, or the cutoff date, whichever came first. Postoperative deaths were not excluded from the survival analysis.

Statistical analysis

Measurements in accordance with a normal distribution are presented as mean ± standard deviation (SD; $\overline{X}\pm S$ ). Skewness was described as the median and interquartile range. Comparisons between groups were made using the independent samples t test or the two independent samples rank sum test. Count data are presented as frequencies (n) or proportions (%), and comparisons between groups were made using the χ² test or Fisher’s exact probability method. Potential imbalances in baseline characteristics between groups were estimated using the standardized mean difference (SMD). First, we conducted a variance inflation factor (VIF) analysis to assess multicollinearity among the variables. Variables with VIF >5 (including “CNLC stage,” “Hepatitis,” and “Targeted therapy plus immunotherapy”) were considered to have severe collinearity and were excluded from subsequent model analyses. To eliminate possible differences in baseline characteristics due to selection bias between the 4-center R group and R-HIPEC patients, we used propensity score matching (PSM) and stabilized inverse probability of treatment weighting (IPTW). Variables included in the propensity model include age, sex, AFP, Hepatitis, alanine aminotransferase (ALT), aspartate aminotransferase (AST), total bilirubin (TBIL), albumin (ALB), prothrombin time (PT), hemoglobin (HB), platelet count (PLT), tumor number, tumor diameter, ascites, liver cirrhosis, Child-Pugh classification, Barcelona Clinic Liver Cancer (BCLC) classification, and China Liver Cancer Staging (CNLC) classification. PSM analysis adopted a 1:1 nearest neighbor matching algorithm, with calipers were 0.03, which were not replaced. For IPTW, patients in both groups were weighted by the inverse tendency score and the inverse 1-tendency score, respectively. The SMD was calculated to evaluate the balance between the two groups at baseline, and an SMD of ⩽0.1 indicated the best balance. The cumulative curves of RFS and OS were plotted using Kaplan–Meier, and comparisons between groups were made using the logarithmic rank test. Independent risk factors for relapse were analyzed using the Cox regression model. After eliminating collinear variables through VIF screening, the remaining variables were included in the univariate Cox regression analysis. Variables with p < 0.1 were further incorporated into the multivariate Cox regression model. Similarly, RFS and OS were substituted into the Cox regression model for multivariate analysis of risk ratios (hazard ratio, HR) and 95% confidence intervals (CI), respectively. In addition, we conducted subgroup analyses based on the variables selected after VIF screening and used the interaction p value to assess whether there was heterogeneity in the treatment effects of HIPEC on RFS and OS in different subgroups. In this study, the statistical significance was established at p < 0.05. Data were processed using SPSS software (version 27.0; IBM Corp., Armonk, NY, USA) and Python software (version 3.13; Python Software Foundation, Wilmington, DE, USA).

Results

Baseline characteristics of patients

A total of 208 srHCC patients from 4 institutions were enrolled in this study as the Primary cohort. Based on the treatment strategies, all patients were divided into the resection group (R) or R-HIPEC group. To ensure model stability and avoid multicollinearity in subsequent survival analyses, we performed VIF analysis to assess correlations among baseline covariates. Variables with VIF >5 were sequentially removed. The initial analysis identified high VIF values for hepatitis (VIF = 11.96), CNLC stage (VIF = 7.66), and targeted therapy plus immunotherapy (VIF = 7.34; Supplemental Figures 1 and 2). After stepwise elimination, all remaining variables showed VIF <5, indicating acceptable multicollinearity levels. The final model included AGE, SEX, AFP, ALT, AST, TBIL, ALB, PT, HB, PLT, Tumor_number, Diameter, Ascites, Liver_cirrhosis, Child_Pugh, BCLC stage, Targeted Therapy, and Immunotherapy. Subsequently, PSM and IPTW were applied to balance baseline characteristics between groups. The PSM created 78 pairs of patients who underwent resection and R-HIEPC, while IPTW created 92.7 standardized patients who underwent resection and 112.8 who underwent R-HIPEC assessment (Figure 1). The clinical characteristics of the study population are presented in Table 1. Based on the propensity score, we found that the covariate distribution between the R group and R-HIPEC group was balanced, with no statistically significant difference in SMD between the two groups in the PSM and IPTW cohorts (all SMD < 0.1).

Figure 1.

Participant flow in the R-HIPEC trial.

R-HIPEC, resection combined with hyperthermic intraperitoneal chemotherapy.

Table 1.

Baseline clinical characteristics of patients with resection or R-HIPEC.

Variable	Before VIF filtering				After VIF filtering
	Primary cohort				1:1 PSM cohort				IPTW cohort
	R group	R-HIPEC group	SMD	p	R-group	R-HIPEC group	SMD	p	R group	R-HIPEC group	SMD	p
	(n = 94)	(n = 114)			(n = 78)	(n = 78)			(n = 92.7)	(n = 112.8)
Age, years
Mean (SD)	54.7(12.8)	55(12.6)	0.062	0.761	56.8(12.4)	55.9(12.7)	−0.072	0.655	54.9(22.7)	55.3(23.4)	0.017	0.912
Sex, %			−0.230	0.111			0.085	0.792			−0.056	0.103
Male	85	94			69	71			80.9	96.4
Female	9	20			9	7			11.7	16.5
Hepatitis			−0.054	0.796			–	–			–	–
Yes	86	106			–	–			–	–
No	8	8			–	–			–	–
AFP, ng/mL			−0.107	0.533			−0.021	>0.999			−0.007	0.443
<400	57	75			50	50			58.3	71.3
⩾400	37	39			28	28			34.4	41.5
ALT, U/L			−0.067	0.748			0.058	0.859			−0.012	0.630
<50	69	87			58	56			68.1	83.5
⩾50	25	27			20	22			24.6	29.4
AST, U/L			−0.158	0.318			−0.077	0.749			−0.035	0.256
<40	47	66			36	39			48.2	60.6
⩾40	47	48			42	39			44.5	52.2
TBIL, µmol/L			0.033	0.935			0.028	>0.999			0.019	0.813
<20	69	82			56	55			67.4	81.1
⩾20	25	32			22	23			25.3	31.7
ALB, g/L			−0.098	0.571			−0.013	0.512			−0.007	0.479
<35	40	43			45	50			36.6	44.1
⩾35	54	71			33	28			56.1	68.7
PT, s			−0.134	0.523			−0.012	0.681			−0.068	0.342
<13.5	6	4			4	2			4.7	4.2
⩾13.5	88	114			74	76			88.0	108.7
HB, g/L			−0.054	0.805			−0.052	0.871			0.005	0.699
<90	52	60			32	34			49.8	60.9
⩾90	42	54			46	44			42.9	51.9
PLT, ×109/L			−0.024	>0.999			0.035	>0.999			−0.014	0.861
<100	14	16			12	13			14.0	16.5
⩾100	80	98			66	65			78.7	96.4
Tumor number			−0.004	>0.999			0.079	0.741			−0.01	0.976
Single	60	73			50	47			58.0	71.2
Multiple	34	41			28	31			34.7	41.6
Diameter, cm			−0.199	0.201			0.026	>0.999			−0.023	0.155
<10	58	81			50	50			60.9	75.4
⩾10	36	33			26	26			31.7	37.4
Ascites			0.016	>0.999			0.062	0.846			0.011	0.909
Yes	20	25			16	18			20.3	25.2
No	74	89			62	60			72.4	87.6
Liver cirrhosis			−0.026	0.964			0	>0.999			0.003	0.850
Yes	35	41			26	26			33.2	40.6
No	59	73			52	52			59.5	72.2
Child-Pugh			0.089	0.607			0.018	0.744			0.017	0.522
A	88	104			72	74			85.5	103.6
B	6	10			6	4			7.2	9.2
BCLC			0.021	0.989			0	>0.999			0.018	0.879
0/A	48	57			38	38			46.3	55.4
B/C	46	57			40	40			46.3	57.4
CNLC			0.123	0.452			–	–			–	–
I	52	56			–	–			–	–
II/III	42	58			–	–			–	–

AFP, serum alpha-fetoprotein; ALB, albumin; ALT, alanine aminotransferase; AST, aspartate aminotransferase; BCLC, Barcelona Clinic Liver Cancer Staging System; Child-Pugh, Child-Pugh Score; CNLC, China liver cancer staging; HB, hemoglobin; IPTW, inverse probability of treatment weighting; PLT, platelet count; PSM, propensity score matching; PT, prothrombin time; R, resection; R-HIPEC, resection combined with hyperthermic intraperitoneal chemotherapy; SMD, standardized mean difference; TBIL, total bilirubin; VIF, variance inflation factor.

RFS analysis between the R and R-HIPEC groups

HIPEC combined with resection could significantly prolong the RFS of srHCC patients compared to resection alone. The RFS curves for the three cohorts are shown in Figure 2. In the three cohorts, the median RFS (mRFS) was significantly longer in the R-HIPEC group than in the R group (in the primary cohort, 7.7 vs 15.5, p = 0.002; PSM cohort: 8.3 vs 18.2, p = 0.002; IPTW cohort: 7.4 vs 14.5, p = 0.014). The 1-, 2-, and 3-year RFS rates were 56.9%, 40.1%, and 36.5% in the R-HIPEC group in the cohort and 43.0%, 20.5%, and 20.5% in the R group, respectively. The PSM cohort was 61.7%, 44.1%, 39.8% and 44.0%, 20.9%, 20.9%. 57.3%, 39.4%, 32.3% and 37.5%, 20.5%, 18.5% in IPTW cohort. Univariate and multivariate regression analyses of RFS-related prognostic factors for primary cohort, PSM cohort, and IPTW cohort are shown in Table 2 and Supplemental Tables 2 and 3. Multivariate analysis showed that AFP ⩾400 ng/mL (in the primary cohort, OR = 1.827, 95% CI: 1.272–2.623, p = 0.001; PSM cohort, OR = 1.532, 95% CI: 1.023–2.295, p = 0.039; IPTW cohort, OR = 1.911, 95% CI: 1.327–2.752, p = 0.001) and tumor diameter ⩾10 cm (in the primary cohort, OR = 1.926, 95% CI: 1.415–2.833, p < 0.001; PSM cohort, OR = 2.056, 95% CI: 1.356–3.119, p = 0.001; IPTW cohort, OR = 1.981, 95% CI: 1.389–2.823, p < 0.001) were the independent risk factors for RFS. In contrast, HIPEC (in the primary cohort, OR = 0.608, 95% CI: 0.427–0.865, p = 0.006; PSM cohort, OR = 0.527, 95% CI: 0.352–0.789, p = 0.002; IPTW cohort, OR = 0.629, 95% CI: 0.445–0.890, p = 0.009) were associated with poor RFS.

Figure 2.

Kaplan–Meier curves for OS and PFS. Primary cohort (a and b), 1:1 PSM cohort (c and d), and IPTW cohort (e and f).

IPTW, inverse probability of treatment weights; OS, overall survival; PFS, progression-free survival; PSM, propensity score matching.

Table 2.

Univariate and multivariate Cox regression analysis for RFS in the primary cohort after VIF-based variable selection.

Variable	Univariate		Multivariate
	OR (95% CI)	p-Value	OR (95% CI)	p-Value
Age (years, ⩾60 vs <60)	0.866(0.598–1.255)	0.449
Sex (male vs female)	0.951(0.583–1.550)	0.840
AFP (μg/L, ⩾400 vs <400)	2.011(1.414–2.859)	<0.001	1.827(1.272–2.623)	0.001
ALT (U/L ⩾50 vs <50)	1.407(0.962–2.059)	0.078	1.245(0.843–1.839)	0.271
AST (U/L ⩾40 vs <40)	1.337(0.943–1.895)	0.103
TBIL (µmol/L ⩾20 vs <20)	1.269(0.872–1.846)	0.214
ALB (g/L, <35 vs ⩾35)	1.153(0.803–1.655)	0.440
PT (s, <13.5 vs ⩾13.5)	0.748(0.321–1.747)	0.503
HB (g/L, <90 vs ⩾90)	1.128(0.794–1.601)	0.501
PLT (×109/L, <100 vs ⩾100)	0.802(0.478–1.345)	0.403
Tumor number (MULT vs SGL)	0.748(0.513–1.091)	0.131
Diameter (cm, ⩾10 vs <10)	1.905(1.330–2.730)	<0.001	1.926(1.415–2.833)	<0.001
Ascites (yes vs no)	1.331(0.891–1.990)	0.163
Liver cirrhosis (yes vs no)	1.043(0.727–1.496)	0.818
Targeted therapy (yes vs no)	0.997(0.702–1.416)	0.986
Immunotherapy (yes vs no)	1.398(0.980–1.995)	0.065	1.248(0.872–1.785)	0.225
Child-Pugh (B vs A)	1.287(0.722–2.296)	0.393
BCLC (B/C vs 0/A)	0.951(0.670–1.351)	0.779
HIPEC (yes vs no)	0.582(0.410–0.826)	0.002	0.608(0.427–0.865)	0.006

AFP, serum alpha-fetoprotein; ALB, albumin; ALT, alanine aminotransferase; AST, aspartate aminotransferase; BCLC, Barcelona Clinic Liver Cancer Staging System; Child-Pugh, Child-Pugh Score; CI, confidence interval; HB, hemoglobin; HIPEC, hyperthermic intraperitoneal chemotherapy; OR, odds ratio; PLT, platelet count; PT, prothrombin time; RFS, recurrence-free survival; TBIL, total bilirubin; VIF, variance inflation factor.

Bold values in the Univariate analysis indicate p < 0.1 (criteria for inclusion in the multivariate analysis), while bold values in the Multivariate analysis indicate statistical significance (p < 0.05).

A deeper analysis of the postoperative recurrence sites showed a significant reduction in peritoneal metastasis with HIPEC, compared to surgery alone, in each of the three cohorts (Table 3).

Table 3.

Comparison of postoperative recurrence between the two groups of patients.

Variable	Primary cohort			PSM cohort			IPTW cohort
	R group	R-HIPEC group	p	R-group	R-HIPEC group	p	R group	R-HIPEC group	p
	(n = 94)	(n = 114)		(n = 78)	(n = 78)		(n = 92.7)	(n = 112.8)
Intrahepatic recurrence	22 (23.4)	24 (21.1)	0.738	17 (21.7)	15 (19.2)	0.711	20.8 (22.4)	22.2 (19.7)	0.628
Peritoneal implantation metastasis	18 (19.1)	10 (8.8)	0.023	13 (16.7)	4(5.1)	0.047	18.5 (20.0)	9.3 (8.2)	0.015
Lung metastasis	14 (14.8)	17 (13.7)	0.846	8 (10.2)	7 (8.9)	>0.999	13.4 (14.1)	15.6 (13.8)	>0.999
Bone metastasis	1 (1.1)	2 (1.8)	>0.999	0	1 (1.2)	>0.999	0.7 (0.7)	2.1(1.8)	0.496

IPTW, inverse probability of treatment weighting; PSM, propensity score matching; R, resection; R-HIPEC, resection combined with hyperthermic intraperitoneal chemotherapy.

Bold values indicate statistical significance (p < 0.05).

OS analysis between the R and R-HIPEC groups

The OS curves for the three cohorts are shown in Figure 2. The median OS (mOS) in the R-HIPEC group was significantly longer than that in the R group in all three cohorts (26.4 vs 45.6 in the Primary cohort, p = 0.025; PSM cohort: 26.4 vs 48.2, p = 0.025; IPTW group: 26.5 vs 42.9, p = 0.012). In the Primary cohort, the 1-, 3-, and 5-year OS rates were 85.0%, 62.1%, and 38.7%, respectively, in the R-HIPEC group and 72.6%, 43.5%, and 32.6%, respectively, in the R group. In the PSM group, they were 88.7%, 77.2%, and 68.0% and 72.5%, 57.1%, and 45.9%, respectively. 81.9%, 56.9%, and 40.6% and 70.7%, 33.3%, and 33.3% in IPTW group, respectively. Univariate and multivariate regression analyses of prognostic factors related to OS in the Primary cohort, the PSM cohort, and the IPTW cohort were presented in Table 4 and Supplemental Tables 4 and 5. Multivariate analysis showed that AFP ⩾400 ng/mL (in the primary cohort, OR = 2.638, 95% CI: 1.657–4.199, p < 0.001; PSM cohort: OR = 2.350, 95% CI: 1.373–4.022, p = 0.002; IPTW cohort: OR = 2.628, 95% CI: 1.654–4.178, p < 0.001) and tumor diameter ⩾10 cm (primary cohort: OR = 1.559, 95% CI: 1.006–2.416, p = 0.047; PSM cohort: OR = 1.976, 95% CI: 1.196–3.264, p = 0.008; IPTW cohort, OR = 1.539, 95% CI: 1.073–2.435, p = 0.045) were independent risk factors for OS. By contrast, HIPEC (in the Primary cohort, OR = 0.685, 95% CI: 0.443–0.923, p = 0.040; PSM cohort: OR = 0.531, 95% CI: 0.322–0.876, p = 0.013; IPTW cohort: OR = 0.695, 95% CI: 0.399–0.824, p = 0.037) and targeted therapy (in the primary cohort, OR = 0.633, 95% CI: 0.398–0.910, p = 0.043; PSM cohort: OR = 0.523, 95% CI: 0.303–0.903, p = 0.020; IPTW cohort: OR = 0.599, 95% CI: 0.367–0.976, p = 0.040) were independently associated with better OS. In addition, ALB <35 g/L was suggested to be a possible risk factor for OS in the primary cohort (OR = 1.830, 95% CI: 1.132–2.958, p = 0.014).

Table 4.

Univariate and multivariate Cox regression analyses for OS in the primary cohort after VIF-based variable selection.

Variable	Univariate		Multivariate
	OR (95% CI)	p-Value	OR (95% CI)	p-Value
Age (years ⩾60 vs <60)	1.132(0.731–1.755)	0.578
Sex (male vs female)	0.694(0.390–1.236)	0.215
AFP (μg/L, ⩾400 vs <400)	2.303(1.499–3.538)	<0.001	2.638(1.657–4.199)	<0.001
ALT (U/L ⩾50 vs <50)	1.470(0.934–2.315)	0.096	1.054(0.632–1.756)	0.841
AST (U/L ⩾40 vs <40)	1.658(1.079–2.548)	0.021	1.306(0.802–2.126)	0.283
TBIL (µmol/L ⩾20 vs <20)	1.382(0.880–2.169)	0.160
ALB (g/L, <35 vs ⩾35)	1.657(1.074–2.556)	0.022	1.830(1.132–2.958)	0.014
PT (s, <13.5 vs ⩾13.5)	1.358(0.578–3.188)	0.482
HB (g/L, <90 vs ⩾90)	1.392(0.899–2.155)	0.138
PLT (×109/L, <100 vs ⩾100)	0.969(0.544–1.724)	0.914
Tumor number (MULT vs SGL)	0.929(0.597–1.447)	0.745
Diameter (cm, ⩾10 vs <10)	1.645(1.067–2.537)	0.024	1.559(1.006–2.416)	0.047
Ascites (yes vs no)	1.321(0.822–2.123)	0.251
Liver cirrhosis (yes vs no)	1.111(0.718–1.718)	0.637
Targeted therapy (yes vs no)	0.644(0.412–1.007)	0.054	0.633(0.398–0.910)	0.043
Immunotherapy (yes vs no)	0.793(0.495–1.268)	0.332
Child-Pugh (B vs A)	1.412(0.723–2.758)	0.313
BCLC (B/C vs 0/A)	1.023(0.668–1.567)	0.917
HIPEC (yes vs no)	0.649(0.422–0.998)	0.049	0.685(0.443–0.923)	0.040

AFP, serum alpha-fetoprotein; ALB, albumin; ALT, alanine aminotransferase; AST, aspartate aminotransferase; BCLC, Barcelona Clinic Liver Cancer Staging System; Child-Pugh, Child-Pugh Score; CI, confidence interval; HB, hemoglobin; HIPEC, hyperthermic intraperitoneal chemotherapy; OR, odds ratio; OS, overall survival; PLT, platelet count; PT, prothrombin time; TBIL, total bilirubin; VIF, variance inflation factor.

Bold values in the Univariate analysis indicate p < 0.1 (criteria for inclusion in the multivariate analysis), while bold values in the Multivariate analysis indicate statistical significance (p < 0.05).

Postoperative complications

In this study, a total of 15 patients experienced grade 3 or 4 postoperative adverse events, but there were no significant differences between the two groups. Complications such as bleeding, bile leakage, abdominal infection, hepatic failure, and perioperative death occurred in both the R and HIPEC groups, but there were no significant differences between the two groups. Among the remaining perioperative variables we counted, the hospitalization cost was higher in the R-HIPEC group than in the R group (p < 0.001), whereas the duration of drain retention and hospitalization were not significantly different compared with the R group (Table 5).

Table 5.

Comparison of perioperative variables between the two groups of patients.

Variable	Primary cohort			1:1 PSM cohort			IPTW cohort
	R group	R-HIPEC group	p	R-group	R-HIPEC group	p	R group	R-HIPEC group	p
	(n = 94)	(n = 114)		(n = 78)	(n = 78)		(n = 92.7)	(n = 112.8)
Drainage tube indwelling time (d, $\overline{\mathit{x}}\pm \mathit{s}$ )	5.9 ± 1.4	6.4 ± 1.2	0.140	6.3 ± 1.4	6.4 ± 1.4	0.650	5.9 ± 1.5	6.2 ± 1.4	0.142
Length of stay (d, $\overline{\mathit{x}}\pm \mathit{s}$ )	13.4 ± 3.6	14.7 ± 4.1	0.069	13.6 ± 2.4	14.2 ± 3.7	0.232	13.1 ± 2.8	14.3 ± 3.7	0.083
Hospitalization expenses (¥, $\overline{\mathit{x}}\pm \mathit{s}$ )	53,184 ± 6713.2	76,282 ± 8343.5	<0.001	51,689 ± 4288.4	72,436 ± 7634.1	<0.001	54,716 ± 4279.5	74,122 ± 5793.4	<0.001
Clavien–Dindo classification (n (%))			0.804			0.739			0.932
I	75 (79.8)	93 (81.6)		65 (83.3)	65 (83.3)		75.8	94.8
II	11 (11.7)	14 (12.3)		7 (9.0)	8 (10.3)		9.9	12.6
III–V	8 (8.5)	7 (6.1)		5 (6.4)	3 (3.8)		5.7	5.7
Perioperative complications (n (%))
Bleeding	4 (4.3)	5 (4.4)	0.963	2 (2.6)	2 (2.6)	>0.999	4.3	4.7	>0.999
Biliary leaks	5 (5.3)	7 (6.1)	0.801	4 (5.1)	4 (5.1)	>0.999	4.0	6.1	0.715
Abdominal infection	4 (4.3)	1 (0.9)	0.114	3 (3.8)	1 (1.3)	0.618	4.4	1.3	0.118
Liver failure	1 (1.1)	1 (0.9)	0.891	1 (1.3)	1 (1.3)	>0.999	0.8	0.8	>0.999
Acute kidney injury stage			0.605			0.143			0.182
Stage 1	27 (28.7)	35 (30.7)		21 (27.0)	24 (30.8)		27.1	35.8
Stage 2	5 (5.3)	8 (7.0)		2 (2.6)	6 (7.7)		4.7	7.5
Stage 3	3 (3.1)	2 (1.8)		1 (1.3)	2 (2.6)		2.6	2.4
Pulmonary complications	19(20.2)	24(21.1)	0.882	14 (17.9)	17 (21.8)	0.670	17.8	24.1	0.702
Death	3(3.2)	4(3.5)	>0.999	3(3.8)	2(2.6)	>0.999	2.8	3.8	0.886

IPTW, inverse probability of treatment weighting; PSM, propensity score matching; R, resection; R-HIPEC, resection combined with hyperthermic intraperitoneal chemotherapy.

Bold values indicate statistical significance (p < 0.05).

Subgroup analyses of HIPEC efficacy

To identify patient subgroups who may benefit most from HIPEC, we performed stratified Kaplan–Meier analyses and interaction tests across baseline characteristics and postoperative treatments (Supplemental Figures 3 and 4). To identify patient subgroups who may benefit most from HIPEC, we performed interaction tests across baseline characteristics and postoperative treatments (Supplemental Figure 4). Stratified Kaplan–Meier analyses by systemic therapy type (Supplemental Figure 3) showed consistent numerical survival advantages for R-HIPEC across treatment modalities. Forest plot analyses (Supplemental Figure 4) revealed significant effect modification for two subgroups. For RFS, patients with BCLC stage 0/A derived significantly greater benefit from HIPEC compared with stage B/C patients (interaction p = 0.0264). For OS, a significant interaction was observed with postoperative immunotherapy (interaction p = 0.0054): patients receiving immunotherapy after R-HIPEC showed markedly improved survival compared with those receiving immunotherapy after resection alone. No significant interactions were detected for age, sex, Child-Pugh class, tumor number, tumor size, or targeted therapy, indicating consistent HIPEC efficacy across most patient subgroups.

Discussion

Once diagnosed with srHCC, timely and effective hemostasis and antishock treatment are key to reducing early mortality. Interventional therapy such as TACE demonstrates good hemostatic efficacy in srHCC patients with hemodynamic instability, poor liver function, and persistent bleeding, with hemostatic rates ranging from 53% to 100%. ¹³ However, tumor resection remains necessary for achieving long-term survival benefits. This study enrolled 208 srHCC patients who underwent resection from 4 institutions to analyze the value of HIPEC in srHCC.

Currently, no guidelines exist for prognostic indicators in srHCC patients.^14,15 Previous studies have identified various prognostic factors, including cirrhosis, multifocality, macroscopic vascular invasion, and spontaneous tumor rupture in large HCC (⩾10 cm). ¹⁶ A meta-analysis of 1876 srHCC patients identified AFP level, liver function, tumor size, and surgical margin as prognostic factors for OS following hepatectomy. ¹⁷ Another study highlighted hypertension, liver cirrhosis, and tumor size as significant predictors. ¹⁸ Our study confirmed that AFP level, tumor size, targeted therapy, and HIPEC were prognostic factors for OS in srHCC patients after hepatectomy, with large ruptured HCC demonstrating particularly poor prognosis.

The treatment of srHCC remains controversial, with varying institutional strategies. While some studies focus on interventional therapy ¹⁹ and others on surgical timing,^20,21 resection has proven valuable for srHCC. Evidence demonstrates that srHCC can achieve comparable survival with non-ruptured HCC when eventually resected. ²² A single-center experience in 239 srHCC patients showed that hepatectomy provided a better prognosis than TACE. ²³ In a Korean cohort, 1- and 3-year disease-free survival rates following hepatectomy were 48.2% and 31.7%. ²⁴ Our study found that the resection alone group had lower disease-free survival rates, whereas resection combined with HIPEC achieved 55.2% and 36.5% for 1- and 3-year disease-free survival, respectively, indicating improved outcomes.

Evidence for HIPEC in srHCC remains limited. The combination of cytoreductive surgery and HIPEC is considered standard care for isolated resectable colorectal peritoneal metastases and primary peritoneal malignancies.^25,26 A multicenter retrospective study reported that cytoreductive surgery and HIPEC were safe and effective for HCC peritoneal metastasis, with mOS of 46.7 months and a 5-year OS of 49.4%. ²⁷ A prospective study exploring prophylactic mitomycin C-based HIPEC in seven srHCC patients showed a positive tendency to prevent peritoneal metastasis. ²⁸ A previous study of 56 srHCC patients found no statistically significant differences in Disease-Free Survival (DFS) (p = 0.28) or OS (p = 0.56) between HIPEC (n = 30) and non-HIPEC groups (n = 27), possibly due to insufficient sample size. ¹¹ Our larger cohort of 208 consecutive, non-selective patients demonstrated significant advantages in both OS and DFS for the R-HIPEC group compared to resection alone, with PSM employed to control for selection bias.

Regarding safety, HIPEC-related symptoms, including fever, nausea, fatigue, sweating, and abdominal distention, were alleviated with symptomatic treatment. No serious complications such as chemical peritonitis, adhesive intestinal obstruction, or intestinal perforation occurred, confirming HIPEC’s safety profile. Although hospitalization costs were higher in the HIPEC group (76,282 ± 8343.5 vs 53,184 ± 6713.2, p < 0.001), this reflected the cost of the procedure itself rather than complication management, as postoperative complication rates were similar between groups. Our previous research also demonstrated HIPEC’s efficacy and safety for stage III gallbladder cancer. ⁸

All ruptured HCC are classified as stage T4 according to the International Union for Cancer Control and Japan staging systems,^29,30 indicating advanced stage and poor prognosis. For this reason, postoperative adjuvant therapy is reasonable to reduce recurrence and metastasis risk. Patients receiving postoperative adjuvant therapy, including oral tyrosine kinase inhibitors, immunotherapy, or combination treatments, demonstrated longer survival compared to observation alone. Notably, combined tyrosine kinase inhibitor treatment with resection achieved similar RFS to combined HIPEC with resection, suggesting both modalities reduce tumor recurrence. The combination of resection, HIPEC, and tyrosine kinase inhibitors achieved the longest OS in this study, providing a valuable management strategy for srHCC.

Several important limitations warrant consideration. First, the retrospective, multicenter design introduces inevitable selection bias, as treatment allocation was determined by surgeons based on intraoperative findings, institutional protocols, and equipment availability. Although PSM and IPTW balanced baseline characteristics, unmeasured confounders—including surgical technique variations, intraoperative blood loss, and surgeon experience—could not be fully accounted for.^31,32 Heterogeneity in postoperative systemic therapies represents another confounding factor. While stratified analyses demonstrated consistent HIPEC benefits across treatment subgroups, the independent effect of HIPEC cannot be completely isolated from subsequent therapies. Interaction analysis suggested potential synergy between HIPEC and immunotherapy (interaction p < 0.05 for OS), warranting mechanistic investigation.^33,34

Second, multicollinearity among covariates necessitated excluding clinically relevant variables such as hepatitis status and CNLC staging from multivariable models. While VIF-based screening improved model stability, this may have omitted factors independently influencing ruptured HCC outcomes. The correlation between CNLC and BCLC staging systems reflects overlapping prognostic information, retaining only BCLC may not fully capture tumor burden complexity. ³⁵ Furthermore, the current follow-up duration may not adequately capture long-term recurrence patterns and late survival differences, particularly as peritoneal metastasis can occur years post-treatment.

Third, generalizability may be limited by population characteristics. This Chinese cohort consisted predominantly of hepatitis B-related HCC, potentially limiting applicability to Western populations where hepatitis C and non-alcoholic fatty liver disease predominate. Different etiologies may influence tumor biology and treatment response. While Western studies demonstrate promising results for cytoreductive surgery plus HIPEC in HCC peritoneal metastasis,^{10,27,36–38} prophylactic HIPEC’s role in ruptured HCC requires validation in non-Asian populations. Despite these limitations, this multicenter study of 208 patients represents the largest cohort examining HIPEC in surgically resectable ruptured HCC. An ongoing prospective randomized controlled trial (NCT 05546619) investigating HIPEC combined with tislelizumab and targeted therapy following R0 resection will provide definitive evidence with standardized protocols and comprehensive biomarker analyses. Future research should investigate optimal HIPEC regimens (drug selection, perfusion temperature, duration), patient selection criteria, and synergistic mechanisms with immunotherapy to maximize therapeutic efficacy while maintaining safety.

Conclusion

This multicenter retrospective and prospective cohort study demonstrates that resection combined with HIPEC represents a safe and effective comprehensive treatment strategy for patients with srHCC. The addition of HIPEC significantly improved both recurrence-free and OS compared with surgery alone, particularly by reducing the risk of peritoneal metastasis. Multivariate analyses identified tumor burden indicators such as elevated AFP and large tumor size as adverse prognostic factors, whereas HIPEC and postoperative targeted therapy served as independent protective factors. Notably, patients in early BCLC stages and those receiving adjuvant immunotherapy derived greater benefit from HIPEC, underscoring the value of individualized treatment approaches. Moreover, HIPEC did not increase the incidence of severe postoperative complications, confirming its favorable safety profile under appropriate perioperative management. Collectively, these findings support the integration of HIPEC into multimodal strategies for srHCC to enhance long-term survival outcomes.