Abstract
Objective
To evaluate the safety and efficacy of laparoscopic enucleation versus laparoscopic hepatectomy for hepatic hemangioma using propensity score matching to adjust for confounding factors.
Methods
This retrospective cohort study used data from consecutive patients who underwent laparoscopic surgery for pathologically confirmed hepatic hemangioma at our hospital from January 2019 to December 2023. Patients were divided into laparoscopic hepatectomy and laparoscopic enucleation groups. Propensity score matching (1:1 nearest neighbor matching) was performed to balance the confounding variables. Balance was assessed using standardized mean differences. Perioperative indicators of the matched cohorts were statistically analyzed.
Results
In total, 100 patients (40 with laparoscopic enucleation and 60 with laparoscopic hepatectomy) were initially included, with 34 pairs (68 patients) successfully matched. Laparoscopic enucleation was associated with superior outcomes in terms of intraoperative blood loss (median (interquartile range): 100.0 (50.0–200.0) mL vs. 400.0 (100.0–600.0) mL, p = 0.001), blood transfusion rate (11.8% vs. 47.1%, p = 0.003), and postoperative hospital stay (5.0 (4.0–6.0) days vs. 7.0 (5.0–8.0) days, p = 0.004). There were no significant differences in the operative time, hepatic vascular occlusion time, and postoperative complication rate between the groups. No postoperative complications occurred in either group after propensity score matching, precluding meaningful comparison of safety profiles. No mortality was observed within 30 days postoperatively.
Conclusion
After propensity score matching, laparoscopic enucleation for hepatic hemangioma was associated with lower intraoperative blood loss, reduced transfusion requirement, and shorter hospital stay compared with laparoscopic hepatectomy, with similar complication rates and short-term mortality. These findings should be interpreted cautiously due to the retrospective study design and limitations associated with propensity score matching.
Introduction
Hepatic hemangiomas (HHs) are the common benign hepatic tumors, with a prevalence of approximately 0.4%–20% in the general population; they usually develop in women aged 30–50 years.1,2 When the diameter of the hemangioma is small, there may be no obvious symptoms. As the diameter increases, gastrointestinal symptoms (such as nausea and vomiting) may occur, and in more serious cases, jaundice and ascites may appear. 3 Currently, observation with follow-up remains the standard management approach for hemangiomas, especially for asymptomatic patients. Several therapeutic options are available for hemangiomas that require treatment, including hemangioma enucleation, hepatectomy, hepatic artery interventional embolization, and radiofrequency ablation.4,5 For the surgical treatment of HHs, either hepatectomy or enucleation is considered the preferred surgical approach.6–8 Hepatectomy is the traditional procedure for the treatment of HH, which may cause greater trauma due to excessive removal of normal liver tissue, inaccurate handling of liver sections, and massive blood leakage.7,9 Hemangioma enucleation, which involves blunt dissection of the hemangioma using the well-defined fibrous membrane formed by the compression of normal liver tissue by HH, can preserve as much normal liver tissue as possible, reducing surgical trauma and hepatobiliary injury.6,10–12
The current choice between the two surgical procedures is determined primarily based on tumor characteristics, including location, depth, and proximity to major vascular structures, rather than surgeon preference alone. Most previous studies comparing the two surgical approaches have been conducted in the context of traditional open surgery.6,7,11–17 However, with the development of minimally invasive surgery, laparoscopic surgery is no longer considered a forbidden approach in liver surgery. 18 Although multiple meta-analyses have compared enucleation and hepatectomy, few studies have focused on laparoscopic procedures with rigorous confounding adjustment. This retrospective cohort study was designed to bridge this gap by using propensity score matching (PSM) to balance baseline differences and provide more reliable evidence for surgical decision making in the laparoscopic era.
Patients and methods
Patients
From January 2019 to December 2023, in total, 100 patients with HH underwent laparoscopic surgical treatment at Yijishan Hospital. This was a retrospective cohort study, and consecutive patients were enrolled.
Inclusion criteria
The following inclusion criteria were applied: (a) postoperative pathology was confirmed as HH; (b) surgical approach was laparoscopic; (c) no preoperative treatment (e.g. interventional embolization and radiofrequency ablation) was received; and (d) complete clinical data (missing data were defined as lack of key indicators, including tumor size, operative time, blood loss, or postoperative follow-up information). In total, six patients were excluded before PSM due to incomplete data.
Collected indicators
Preoperative data (patient characteristics, hemangioma characteristics, and laboratory variables), intraoperative information (surgical method, operative time, hepatic vascular occlusion (HVO) time, estimated blood loss, and intraoperative blood transfusion), and postoperative data (length of hospital stay, postoperative complications, and mortality) were collected.
Preoperative management
All patients completed the relevant laboratory tests and electrocardiography as required for surgery and anesthesia. Contrast-enhanced computed tomography (CT) or magnetic resonance imaging (MRI) was the preferred diagnostic methods for hemangioma. Preoperative three-dimensional CT reconstruction was performed if the tumor was adjacent to major vascular structures such as the inferior vena cava (IVC), portal vein, or hepatic vein. This study was conducted in accordance with the Declaration of Helsinki (2024 revision). 19 All patient data were fully deidentified to protect privacy. This study was approved by the Ethics Committee of The First Affiliated Hospital of Wannan Medical College (Yijishan Hospital, Approval Number: 2023-194). Due to the retrospective nature of the study, requirement of written informed consent was waived by the ethics committee. The reporting of this study conforms to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines. 20 All resected specimens were examined pathologically.
Surgical techniques
Terms related to anatomical liver resection were consistent with the Brisbane 2000 Classification System. Anatomic resection involved removal of the hemangioma as well as the associated portal vein and its hepatic area, including hemihepatectomy (right or left), sectionectomy, and segmentectomy (resection of Couinaud’s segments). Non-anatomic hepatectomy refers to resection of hemangiomas without consideration of segmental, sectional, or hemiliver anatomies. Enucleation refers to blunt dissection of the hemangioma using the well-defined fibrous membrane formed by the compression of normal liver tissue by HH, without loss of normal hepatic parenchyma. Surgical selection was guided mainly by tumor anatomy, location, size, and proximity to major vessels.
Follow-up
All patients were under standardized postoperative surveillance. Abdominal ultrasound or CT scans were performed 3 days after the surgery to exclude fluid collections or vascular complications. Laboratory tests were conducted every 3 days and discontinued if indicators normalized. Complications were prospectively recorded and graded according to the criteria proposed by the International Study Group of Liver Surgery (ISGLS). Long-term follow-up included clinical review, liver function tests, and abdominal ultrasound at 6 and 12 months and yearly subsequently. Symptom evolution was classified as (a) complete resolution; (b) partial amelioration; (c) unchanged persistence; or (d) worsening.
Statistical analyses
PSM analysis
Propensity scores were calculated using a logistic regression model with surgical method (laparoscopic enucleation (LE) vs. laparoscopic hepatectomy (LH)) as the dependent variable and confounding variables (age, sex, surgical indication (asymptomatic/symptomatic/rapid growth), tumor size, tumor location, tumor number, hemoglobin level, platelet count, and alanine aminotransferase (ALT) level) as independent variables. Furthermore, 1:1 nearest neighbor matching without replacement was performed with a caliper value of 0.2 × SD of the propensity score. Balance after PSM was assessed using standardized mean differences (SMDs); |SMD| <0.10 was defined as excellent balance. Given the relatively small matched sample (34 pairs), PSM may have reduced the statistical power of the analysis and excluded complex cases, thereby limiting the representativeness of the results.
Descriptive statistics
Normally distributed continuous variables were expressed as mean ± SD, and non-normally distributed continuous variables were presented as median (interquartile range (IQR)). Categorical variables were expressed as counts (%).
Comparative tests
Normally distributed continuous variables were compared using unpaired independent samples t-test; non-normally distributed continuous variables were compared using the Mann–Whitney U test; categorical variables were analyzed using Pearson chi-square test or Fisher’s exact test as appropriate.
Multiple comparison adjustment
Bonferroni correction was applied for multiple comparisons, with a corrected significance level of p <0.05/15 ≈ 0.0033.
Sample size calculation
Based on the primary outcome indicator (intraoperative blood loss), a preliminary sample size calculation was performed using G*Power 3.1. Assuming a mean difference of 200 mL in blood loss between the groups, an SD of 300 mL, α of 0.05, and power of 0.8, the required sample size was 58 (29 per group). The matched sample size (34 per group) met the statistical power requirement.
Statistical analyses were performed using the Statistical Package for Social Sciences (SPSS) software (IBM, SPSS Statistics for Windows, version 22.0, SPSS Inc.; Chicago, IL, USA) and Python 3.9 (scikit-learn, SciPy, pandas packages for PSM). A two-tailed p-value <0.05 was considered statistically significant.
Results
Patient and hemangioma characteristics (after PSM)
After PSM, 34 pairs of patients (68 total) were included in the final analyses, with a matching success rate of 85.0% and an average propensity score difference of 7.01%. All baseline variables were balanced between the two groups (all p > 0.05, Table 1). The median age of the patients was 53 (range, 30–76) years. The male-to-female ratio was 14/20 (LH group) and 13/21 (LE group). In total, 41 patients (60.3%) were asymptomatic (incidental finding during health checkup), 16 (23.5%) had upper abdominal discomfort and indigestion, 9 (13.2%) had right or left sided abdominal pain, and 2 (2.9%) exhibited rapid tumor growth. The median size of the largest tumor was 7.5 (range, 4.19–12.91) cm in the LH group and 7.4 (range, 5.0–12.0) cm in the LE group. Tumor location included the left liver (22 cases), right liver (31 cases), and bilateral liver (15 cases). There were 40 patients (58.8%) with a single mass and 28 (41.2%) with multiple masses. Preoperative laboratory indicators (hemoglobin level, platelet count, prothrombin time, ALT level, aspartate aminotransferase (AST) level, and total bilirubin level) were within normal ranges in all the patients (Table 1).
Baseline characteristics of the two patient groups.
Continuous variables are presented as median (IQR) or mean ± SD; categorical variables are presented as n (%). Two-tailed p-value <0.05 was considered statistically significant; Bonferroni-corrected p-value <0.0033 was considered statistically significant for multiple comparisons. An absolute SMD <0.10 indicated excellent baseline balance after propensity score matching.
LE: laparoscopic enucleation; LH: laparoscopic hepatectomy; IQR: interquartile range; SMD: standardized mean difference; PSM: propensity score matching.
Comparison of perioperative outcomes (after PSM)
Intraoperative outcomes
The median operative time was 191.5 (139.25–339.5) min in the LH group and 166.5 (114.5–221.75) min in the LE group (p = 0.112). HVO was performed for 21 patients (61.8%) in the LH group (median duration: 30.0 (0.0–75.0) min) and 8 patients (23.5%) in the LE group (median duration: 11.25 (0.0–30.0) min) (p = 0.105). The median intraoperative blood loss was significantly lower in the LE group (100.0 mL; IQR, 50.0–200.0) than in the LH group (400.0 mL; IQR, 100.0–600.0) (p = 0.001). The blood transfusion rate was 47.1% (16/34) in the LH group and 11.8% (4/34) in the LE group (p = 0.003) (Table 2).
Intraoperative and postoperative indicators of the two patient groups.
Continuous variables are presented as median (IQR) or mean ± SD; categorical variables are presented as n (%). Postoperative complications were graded according to the ISGLS criteria. Two-tailed p-value <0.05 was considered statistically significant; Bonferroni-corrected p-value <0.0033 was considered statistically significant for multiple comparisons.
LE: laparoscopic enucleation; LH: laparoscopic hepatectomy; IQR: interquartile range; ISGLS: International Study Group of Liver Surgery; PSM: propensity score matching.
Postoperative outcomes
The main postoperative complications included pleural effusion, pneumonia, and bile leak, with no significant difference in the complication rate between the two groups (0.0% vs. 0.0%, p = 1.000). No perioperative deaths occurred in either group. The median postoperative hospital stay was significantly shorter in the LE group (5.0 days; IQR, 4.0–6.0) than in the LH group (7.0 days; IQR, 5.0–8.0) (p = 0.004) (Table 2).
Follow-up
The median follow-up period was 50 (range, 18–78) months. Among the 25 patients who were symptomatic before the surgery, 92.0% (23/25) achieved complete resolution or apparent improvement of symptoms; 2 patients with a long history of erosive gastritis and duodenal ulcer continued to experience recurrent epigastric discomfort. Liver function tests of all patients showed no significant abnormalities. CT or ultrasound examinations revealed no new hemangiomas in other parts of the liver.
Discussion
With rapid advancements in minimally invasive techniques, interventional techniques are commonly used for treating HHs, such as hepatic artery embolization and radiofrequency ablation.4,5 However, surgery, including hepatectomy and enucleation, remains the most effective treatment for HH. 6 In particular, it is important to emphasize that no studies or guidelines recommend either of the two surgical options as the preferred option. In clinical practice, the choice between the two procedures is primarily driven by tumor characteristics such as location, depth, and proximity to major vascular structures, rather than depending solely on the surgeon’s personal preference. Many previous studies have compared the two surgical approaches. However, most of these comparisons have been made in the setting of open surgery.6,7,13 With the development of laparoscopic techniques in liver surgery, both LH and LE can be performed at almost any large liver center. 18 In the era of laparoscopic surgery, the question of which of the two surgical approaches is more appropriate for the treatment of HHs is one that needs to be explored and cannot depend solely on surgeon preference. To further minimize selection bias and enhance the reliability of conclusions, this study supplemented PSM analysis; after 1:1 nearest neighbor matching (without replacement) with a caliper value of 0.2 × SD of propensity score, 34 pairs of patients (34 in each group) were finally included, with a matching success rate of 85.0% and favorable baseline balance (all p > 0.05 for confounders including age, sex, tumor characteristics, and laboratory parameters), which allowed a more objective comparison of perioperative outcomes between the two groups.
In this study, we compared the perioperative period of the LE and LH groups with respect to HH, and the PSM-adjusted results suggested potential advantages of LE. In terms of operative time, LE was shorter than LH (median: 166.5 (114.5–221.75) min vs. 191.5 (139.25–339.5) min); however, the difference did not reach statistical significance after PSM (p = 0.112), a result possibly attributed to reduced sample size or balanced surgical complexity across matched groups. This trend, however, is consistent with previous studies 21 as LH causes more damage to the blood vessels and bile ducts than LE, requiring surgeons to spend more time on ligation and hemostasis and disrupting the continuity of the procedure. 22 To the best of our knowledge, the operative time is usually related to the difficulty of performing the surgery, smoothness of the operation, size of the tumor, and skill level of the surgeon in charge. We also compared the size of the tumors between the two groups, and there was no significant difference in tumor size after PSM (p = 0.713).
In terms of HVO, the present study found significant differences between the two groups before PSM matching; enucleation required portal triad clamping in only 32.5% (13/40) of the patients, with a median occlusion time of 15.0 (8.0–25.0) min, whereas hepatectomy necessitated HVO in 76.7% (46/60) of the procedures and a longer median ischemia interval (28.0 (18.0–40.0) min, p = 0.003). After PSM adjustment, the proportion of patients requiring HVO in the LE group remained significantly lower than that in the LH group (23.5% (8/34) vs. 61.8% (21/34), p = 0.001), and LE showed a trend of shorter median occlusion time (0.0 (0.0–11.25) min vs. 0.0 (0.0–30.0) min), though the difference was not statistically significant (p = 0.105). The HVO technique is an important skill in modern liver surgery. Most scholars believe that an appropriate technique can reduce bleeding during liver surgery. The selective hepatic vascular exclusion and the Pringle maneuver are two of the most commonly used methods in liver surgery. 23 In liver surgery, the porta hepatis occlusion technique can reduce intraoperative bleeding and maintain a clear anatomical plane, especially when the tumor is adjacent to important intrahepatic structures. However, due to the unique characteristics of HHs—most of which exhibit expansile growth, compressing the surrounding liver tissue, blood vessels, and bile ducts to form a visible fibrous capsule—this special capsular structure serves as the anatomical basis for performing hemangioma enucleation. Enucleating the hemangioma completely along the edge of the capsule not only ensures the thorough removal of the lesion but also minimizes damage to the normal intrahepatic structures. 10 Therefore, under the conditions of minimal bleeding from the surgical wound and a clear operative field, the authors believe that porta hepatis occlusion can be avoided as much as possible, which helps reduce ischemia-reperfusion injury caused by porta hepatis occlusion.
In terms of intraoperative bleeding and blood transfusion, significant advantages of laparoscopic hemangioma enucleation over LH were observed both before and after PSM. Before matching, the median intraoperative blood loss in the LE group was 150.0 (50.0–300.0) mL, which was significantly lower than 550.0 (300.0–800.0) mL in the LH group (p < 0.001); the transfusion rate in the LE group was 10.0% (4/40), which was significantly lower than 46.7% (28/60) in the LH group (p < 0.001). PSM analysis further confirmed this advantage: the median intraoperative blood loss in the LE group was 100.0 (IQR: 50.0–200.0) mL, which was significantly lower than 400.0 (IQR: 100.0–600.0) mL in the LH group (p = 0.001); the transfusion rate in the LE group was 11.8% (4/34), which was significantly lower than 47.1% (16/34) in the LH group (p = 0.003). The transfusion rate in the LH group (47.1%) appeared relatively high in relation to the reported blood loss; this may be explained by the transfusion criteria used at our institution (hemoglobin <7 g/dL or hemodynamic instability) and some cases of prophylactic transfusion due to preoperative anemia or anticipated major bleeding. The main causes of hemorrhage in liver surgery are summarized as follows:
21
Poor operative field exposure and unexpected injury to large vessels. When the hemangioma is large, it can block the surgical field of view, resulting in limited surgical space, making it easy to puncture or sever large blood vessels. At this time, the tumor cannot be removed immediately for a while, which caused bleeding that is difficult to control. Additionally, the surgical field is even less clear because of bleeding, and it is easy to damage the large blood vessels that are located at the bottom of the tumor, resulting in more serious hemorrhage. This is often observed in cases where the hemangioma is located in the second hepatic hilar, deeper part of the middle liver, paravalvular vein, and caudal lobe. Excessive traction on the liver ruptures large blood vessels. During resection of the right half of the liver or a huge tumor in the right posterior lobe, the short hepatic vein, right hepatic vein, and right adrenal vein are easily torn during the process of turning the liver to the left after freeing the ligament, and massive bleeding occurs. The narrow surgical field of view and challenges associated with hemostatic operation increase the risk of serious intraoperative complications. Severe adhesions in the operative field make the operation difficult. In cases of tumor on top of the diaphragm or repeated liver surgery, when separating the adhesions, it is easy to cause the liver tissue to break or the tumor to rupture, thus causing massive bleeding. At this time, if the hemorrhage is stopped by blind clamping, it may lead to progressive enlargement of the tumor rupture surface, making bleeding difficult to control. This situation is mostly observed in patients with hemangioma located at the top of the hepatic diaphragm, bare liver area, and right posterior lobe. Unfamiliarity with an intrahepatic anatomical structure and error of judgment. Due to pushing of the huge tumor, the large blood vessels in the liver are displaced and the vascular alignment is changed, resulting in judgment error for the surgeon, which causes large vessel injury and bleeding. For example, right hepatectomy injures the posterior hepatic vena cava or left inner lobe; alternatively, the left hepatectomy accidentally injures the root of the middle hepatic vein or the root of the left hepatic vein. In contrast, especially in liver segments with deep tumor locations and difficult exposure, hepatectomy requires greater involvement of the large blood vessels and bile ducts of the liver, whereas hemangioma debulking only requires separation of the tumor from the normal liver tissue along a clear demarcation line, usually with relatively few or small bile ducts or blood vessels between the tumor and normal liver tissue.
In terms of complications, the present cohort showed lower overall morbidity after enucleation (5.0%, 2/40) than after hepatectomy (11.7%, 7/60) before PSM, although this difference did not reach statistical significance (p = 0.283). The spectrum of complications was also qualitatively different; the two events observed in the LE group consisted of bile leak (ISGLS grade A) in one case and pleural effusion in one case that resolved with conservative drainage, whereas the seven events in the LH group included pneumonia in two cases, diaphragmatic injury in one case, postoperative hemorrhage requiring angioembolization in one case, bile leak (grade B) in one case, and pleural effusion in two cases. After PSM adjustment, no postoperative complications occurred in either group (p = 1.000). The absence of complications after PSM does not allow a meaningful comparison of safety between the two techniques; therefore, statements suggesting reduced complication rates with enucleation should be interpreted with caution. This trend is consistent with the recently published meta-analysis by Jiang et al. 21 who reported a pooled odds ratio of 0.57 (95% CI: 0.44–0.74) in favor of enucleation over hepatic resection for hemangioma. According to the published meta-analysis discussed above, compared with liver resection, the blood vessels and bile ducts involved in HH resection are smaller, and the loss of normal liver tissue volume is also reduced, which can reduce postoperative bleeding and bile leakage. 21 The fact that no patient in either arm developed liver failure or exhibited 30-day mortality (0.0% in both groups before and after PSM) supports the view that enucleation does not compromise functional reserve even when giant lesions (>10 cm) abut major vascular pedicles. Taken together, our data suggest that enucleation reduces both incidence and severity of perioperative complications compared with formal hepatectomy and maintained equivalent safety in terms of liver function and survival.
In terms of post-surgical hospital stay, enucleation was associated with a significantly shorter length of stay than hepatectomy both before and after PSM. Before matching, the median hospital stay in the LE group was 5.0 (4.0–7.0) days, which was significantly shorter than that in the LH group (8.0 (6.0–10.0) days, p < 0.001). PSM analysis further verified this advantage; the median hospital stay in the LE group was 5.0 (IQR: 4.0–6.0) days, which was significantly shorter than that in the LH group (7.0 (IQR: 5.0–8.0) days, p = 0.004). This 22% reduction is clinically relevant and mirrors the results of a recently published meta-analysis by Jiang et al. who reported a pooled mean difference of −2.33 days (95% CI: −3.22 to −1.44) in favor of enucleation. 21 The quicker hospital discharge can be attributed to the limited parenchymal transection plane, lower blood loss (215 vs. 620 mL), and reduced need for transfusion (10% vs. 47%), all of which attenuate the systemic inflammatory response and accelerate resumption of oral intake.
Although this study compared the perioperative outcomes of the two surgical approaches and used PSM analysis to reduce selection bias, it has certain limitations. First, tumor location was categorized broadly (left, right, and bilateral), which may not adequately reflect surgical complexity; more detailed anatomical information (e.g. segmental location or proximity to major vascular structures) is required for fully accounting for between-groups differences. Second, the inclusion of “surgical indication” as a matching variable warranted further specification, and some heterogeneity in indications remains. Third, the relatively small matched cohort (34 pairs) reduced the statistical power of the analysis and may have excluded more complex cases, potentially affecting representativeness. Fourth, the absence of postoperative complications after PSM precludes meaningful comparison of safety profiles. Finally, as surgeon preference determines the proficiency of the surgeon in his or her surgical approach, it will inevitably exert a non-excludable impact on the indicators associated with the procedure.
Conclusion
To the best of our knowledge, this study is among the first to compare enucleation and hepatectomy for HHs in the field of laparoscopic liver surgery and incorporate PSM analysis to adjust for confounding factors. The results suggest that LE is associated with favorable perioperative outcomes, including reduced intraoperative blood loss, lower transfusion requirement, and shorter hospital stay. However, given the observational design, relatively small matched sample size, and absence of postoperative complications after PSM, these findings should be interpreted cautiously and considered hypothesis‑generating rather than definitive. Nonetheless, this study may inform laparoscopic surgeons when choosing between enucleation and hepatectomy for hemangioma treatment, encouraging a decision that considers tumor characteristics and patient factors rather than relying solely on personal preference.
Footnotes
Acknowledgments
We thank all patients involved in the study and the medical staff of the Department of Hepato-Biliary-Pancreatic Surgery for their support.
Authors’ contributions
X.S. Fang and B. Jiang were responsible for data collection and study conceptualization; S.H. Xi and J. Liu performed statistical analyses (including PSM); X.M. Wang and X.S. Fang interpreted the results and revised the manuscript; B. Jiang and X.F. Yan drafted the manuscript. All authors approved the final version.
Availability of data and materials
The raw data and statistical code used in this study are available from the corresponding author upon reasonable request (including deidentified patient data and PSM analysis scripts).
Consent for publication
All authors provided consent for publication.
Declaration of conflicting interests
The authors declare no competing financial or non-financial interests.
Ethics approval and consent to participate
All data acquisition was approved by the Ethics Committee of The First Affiliated Hospital of Wannan Medical College (Yijishan Hospital, No. 2023-194). Informed consent was waived due to the retrospective nature of the study.
Funding
No funding was received for this study.
