Surgical outcomes of pulmonary metastasectomy in hepatocellular carcinoma patients according to approach method: thoracoscopic versus open approach

Background The role of surgical intervention as a treatment for pulmonary metastasis (PM) from hepatocellular carcinoma (HCC) has not been established. In this study, we investigated the clinical outcomes of pulmonary metastasectomy. Using propensity score matching (PSM) analysis, we compared the results according to the surgical approach: video-assisted thoracic surgery (VATS) versus the open method. Methods A total of 134 patients (115 men) underwent pulmonary metastasectomy for isolated PM of HCC between January 1998 and December 2010 at Seoul Asan Medical Center. Of these, 84 underwent VATS (VATS group) and 50 underwent thoracotomy or sternotomy (open group). PSM analysis between the groups was used to match them based on the baseline characteristics of the patients. Results During the median follow-up period of 33.4 months (range, 1.8–112.0), 113 patients (84.3%) experienced recurrence, and 100 patients (74.6%) died of disease progression. There were no overall survival rate, disease-free survival rate, and pulmonary-specific disease-free survival rate differences between the VATS and the open groups (p = 0.521, 0.702, and 0.668, respectively). Multivariate analysis revealed local recurrence of HCC, history of liver cirrhosis, and preoperative alpha-fetoprotein level as independent prognostic factors for overall survival (hazard ratio, 1.729/2.495/2.632, 95% confidence interval 1.142–2.619/1.571–3.963/1.554–4.456; p = 0.010/< 0.001/< 0.001, respectively). Conclusions Metastasectomy can be considered a potential alternative for selected patients. VATS metastasectomy had outcomes comparable to those of open metastasectomy.


Introduction
Hepatocellular carcinoma (HCC) is one of the most frequent malignancies in Asia, including Korea [1], and ranks third in cancer deaths worldwide [2]. In the recent years, the postoperative morbidity and mortality rates from HCC have improved; however, long-term outcomes remain poor due to extrahepatic metastasis and intrahepatic recurrence after surgical management. The most frequent site of metastasis and of the first detectable metastasis was reported to be the lungs [3,4]. However, appropriate management for pulmonary metastasis (PM) has not been established, and systemic chemotherapy has been reported to be largely ineffective [5]. The results of the multinational, randomized, placebo-controlled, phase III sorafenib HCC Assessment Randomized Protocol (SHARP) trial demonstrated that sorafenib significantly improved the overall survival in patients with advanced HCC and wellpreserved liver function, and that drug-related adverse events were manageable [6]. However, according to the subanalysis of phase II sorafenib Asia-Pacific trial, the effectiveness of sorafenib in advanced HCC patients remains unclear as the survival gain in patients with PM was only 1.4 months (hazard ratio [HR] 0.87) [7]. Although sorafenib has been a standard regimen for advanced HCC, its effectiveness is marginal in the presence of PM. In most cases, the lesions of extrahepatic metastasis from HCC are not resectable. Therefore, surgical treatment of PM has not been applied as a standard therapy in PM from HCC.
In some reports, pulmonary resection for PM from HCC resulted in long-term survival in highly selected patients [8][9][10]. Usually, wedge resection through open thoracotomy has been performed for superficial lesions, and segmentectomy or lobectomy is needed for deeper lesions. With recent advances in VATS and diagnostic modalities, such as multi-detector row computed tomography (CT), minimally invasive and complete metastasectomy can be provided to chronically ill patients with HCC and PM [1][2][3][4][5][6][7][8][9][10][11][12][13][14]. However, the use of surgical resection and VATS for PM of HCC remains controversial. We believe that an appropriate surgical role is needed for the treatment of patients with advanced HCC. In this study, we investigated the clinical outcomes of pulmonary metastasectomy and the risk factors for survival rate and disease-free survival rate in patients of HCC with PM. Using propensity score matching (PSM) analysis, we compared the results according to surgical approach: VATS versus open thoracotomy or sternotomy methods.

Patients
In this study, we retrospectively reviewed a total of 1085 patients who had undergone metastasectomy due to thoracic metastasis from HCC between January 1998 and December 2010. Of these, 852 cases with different histology and 18 cases with metastatic lesions in the thorax other than the lung were excluded. Sixty-one cases were further excluded due to duplicated patients. Ten patients had concurrent malignancies, and 3 cases were diagnosed with a different histology from HCC. After excluding 7 cases with incomplete medical records, we retrospectively 134 patients (115 men) who had undergone pulmonary metastasectomy for isolated PM of HCC. The enrollment of patients is shown in Fig. 1. The inclusion criteria of our study were as follows: (1) controlled state of primary HCC, (2) no other distant metastasis, and (3) appropriate pulmonary function test for major lung resection. All patients had available follow-up till December 2015.

Methodology
Eighty-four patients underwent VATS (VATS group), and 50 underwent thoracotomy or sternotomy (open group). Under general anesthesia and double-lumen endotracheal tube insertion for single-lung ventilation, the patients' chest that is where the lesion is located was up on the degree of 90°. Three or four surgical ports were created for VATS, and posterolateral thoracotomy or median sternotomy was performed for open surgery. Surgical approaches such as thoracotomy, sternotomy, and VATS were determined depending on the size, number, location, and laterality of the metastatic nodules. VATS can be applied for lesions of sizes below 3 cm, located in the outer one-third of the lung, having no endobronchial extension. The metastatic lesion located in the periphery of the lung was non-problematically resected by wedge resection. Segmentectomy or lobectomy was needed for deeply located lesions. Mediastinal lymph node dissections were performed only if enlarged lymph nodes were revealed on chest CT. This study was approved by the Asan Medical Center Ethics Committee/Review Board (2019-1166).

Statistical analysis
Time-to-pulmonary recurrence after the initial treatment (TPR) was defined as the interval between control of HCC and the diagnosis of PM. Disease-free survival (DFS) after the first pulmonary metastasectomy was defined as the interval between the first pulmonary metastasectomy and diagnosis of a lesion related to HCC after the first pulmonary metastasectomy. Pulmonary DFS after metastasectomy (PDFS) was defined as the interval between the first pulmonary metastasectomy and the diagnosis of the second PM. Overall survival time was measured by comparing the date of the first pulmonary metastasectomy to the date of death or the last followup. Preoperative alpha-fetoprotein (AFP) was defined as the value of the latest serum AFP level measured before the first metastasectomy. Postoperative AFP was defined as the value of the earliest measured serum AFP level after the first metastasectomy, within 2 months. The calculation of the BCLC stage was based on before lung metastasis was detected. This is because, based on the time after lung metastasis is found, all patients are at BCLC stage c or higher, and other variables of the patient before surgery may not be considered. Categorical variables were expressed as frequencies and percentages and were compared between the two approaches using the chi-square test or Fisher's exact test. Continuous variables were shown as means with standard deviations or median values with range and were compared using a Student's t test. Univariate and multivariate analyses for prognostic factors were performed using Cox's proportional hazard model. Survival rates were estimated using the Kaplan-Meier method, and survival according to the prognostic factors was compared using a log-rank test. Cox proportional hazard models were used to identify the predictors of mortality and recurrence. After excluding the correlated variables, independent variables with p values ≤ 0.05 from the univariate analysis were used for the initial multivariate Cox. The final multivariable model was selected using forward stepwise selection (p value ≤ 0.10 for entering the model and p ≤ 0.05 for staying in the model).
There were several considerations (baseline characteristics, HCC factors, surgical factors) for PSM; 13 variables (age, sex, history of liver cirrhosis, level of initial AFP, etiology of HCC, BCLC stage of HCC, Child-Pugh score, Eastern Cooperative Oncology Group (ECOG) scale of performance status, treatment method for primary HCC, local recurrence or progression of HCC, number of PM, maximum diameter of PM, TPR) were used for PSM. For PSM, observation pairs with equivalent propensity scores were selected with nearestneighbor matching and a caliper width of 0.2 standard deviation. Patients in the VATS group were randomly matched to patients in the open group at a ratio of 1:1. The final sample consisted of 32 matched patients in each group.
Statistical analyses were performed using SPSS version 21.0 (SPSS Inc. Armonk, NY, USA) and R version 3.4.2 (R Project for Statistical Computing, Vienna, Austria). A value of p < 0.05 was considered statistically significant.

Results
The median age was 55.0 years (range, 24-76 years). Patient characteristics are summarized in Table 1.
The patients' characteristics after PSM are summarized in Table 7. No significant differences in baseline and PM characteristics were observed. The open group underwent more extensive surgery and stayed more days than the VATS group (p = 0.025 and p = 0.017, respectively) ( Table 8). There was no significant difference in overall survival rate, DFS rate, and PDFS rate after the first pulmonary metastasectomy between the VATS and open groups (p = 0.764, 0.937, and 0.786, respectively) (Fig. 3).

Discussion
The survival rate of HCC patients has markedly improved as a result of advances in surgical techniques and perioperative management. Although the lung is the most frequent site of extrahepatic metastasis from HCC, the role of surgery for PM from HCC has not been fully clarified yet, unlike PM from colorectal cancer or osteosarcoma. There have been some reports favoring surgical resection for isolated PM from HCC, and the present study supports those results [8][9][10]. While median survival time and 1-year survival rate of the patients with PM treated with chemotherapy were 4.6-14 months and 20-42%, respectively [5,8], 5-year survival rate of patients who had undergone metastasectomy was reported to be 24-41.4% [15][16][17][18][19][20].
In the present study, the 5-year DFS rate and 5-year survival rate were 20.4% and 37.3%, respectively. More long-term survivors could be found among the patients who underwent metastasectomy than among those who received chemotherapy. However, since most studies were non-randomized and included only operable and resectable lesions in the surgery group, a comparison of outcomes between metastasectomy and chemotherapy needs circumspection.
Cheng et al. reported the overall survival of patients with advanced HCC with or without sorafenib. The median overall survival period was 6.5 and 4.2 months [7]. In our study, the median overall survival period was Surgical intervention for PM can provide excellent local control. Due to its low mortality and morbidity   [15][16][17][18][19][20], metastasectomy should be considered as a viable option to prevent pulmonary complications caused by PM. Several studies suggest that to improve longterm survival, patients with isolated PM and no other distant metastasis should be considered not only for surgical resection but also for repeated resection of PM [21][22][23]. In our experience, patients who underwent pulmonary metastasectomy could be discharged within approximately 7 days without operative mortality or serious complications. The second or third pulmonary metastasectomy could also be safely performed, and repeated metastasectomy could be attempted in patients with localized PM from HCC for better survival (5-year survival rate was 59.1% in repeated metastasectomy patients and 48.1% in non-surgical treatment patients, p = 0.002). Introduction of VATS makes metastasectomy more endurable for patients with chronic diseases such as liver cirrhosis. Less pain and shorter hospitalization can lead to earlier returns to preoperative activities or subsequent treatment [11][12][13]24]. However, the inability to thoroughly palpate the lung by VATS raises concerns about incomplete metastasectomy [25,26] which could result in significantly worse survival rates [27]. In contrast, there are reports suggesting that the outcomes of thoracoscopic surgery are not inferior to those of open thoracotomy [13,28]. This disagreement can partly be due to the possibility that finger palpation of the lung also has limitations in terms of detecting small nodules; in addition, metastasectomy under a thoracotomy is not always complete [29]. Recently, in contrast to the inaccuracy of old-generation CT, the 1-mm-thin section 16-channel multi-detector row CT showed a high detection rate of metastatic pulmonary nodules, especially in patients with non-osteosarcoma. Therefore, it can be a possible substitute for manual palpation [14]. If diagnosis by either imaging or palpation is incomplete, VATS should be considered for metastasectomy. VATS can minimize adhesion formation and render repeated resection more amenable. Given that complete resection is most important in any situation, conversion to thoracotomy should be promptly considered if the lesions detected by CT cannot be identified or resected by VATS. In the present study, all preoperatively detected PMs were completely resected either by VATS or by open surgery. Hospitalization duration and complication rate were significantly reduced in the VATS group. However, because more segmentectomies and lobectomies were adapted in the open group, it is difficult to compare the two methods straightforwardly. To overcome the shortcomings outlined above, in the present study, the propensity score matching analysis was adapted to compare two methods, and relatively favorable results were observed. However, matched analysis needs to drop unmatched data, for which, the main drawback is the shortage of cases. For this reason, further randomized large cohort studies need to be conducted in the future. In the published literature, only several studies have reported risk factors related to long-term survival. Complete resection is the most significant predictor of better survival [18][19][20]27]. Worse survival can be expected for patients with the factors causing a rise in the possibility of incomplete resection including multiple lesions [16,30] and the factors predicting aggressive disease, such as short disease-free interval [18-20, 27, 30-33], multiple PMs, and higher serum AFP levels [19,20]. It was suggested that the size of nodules had an inverse correlation with survival in PM from cancer other than HCC [22]. In the present study, local recurrence or progression of HCC, history of liver cirrhosis, and preoperative AFP were found to be independent prognostic factors for overall survival after the first pulmonary metastasectomy. Local recurrence or progression of HCC and preoperative AFP were also found to be independent prognostic factors for pulmonary-specific recurrence. Therefore, our results suggest that if there is a local recurrence or progression of HCC or if the levels of AFP before surgery are high, decisions on surgical treatments should be made with caution. The impact of all these proposed factors needs to be verified in future largescale studies.

Conclusion
When metastasis confined to the lung with the HCC under control or controllable, metatasectomy can be positively considered for patients with sufficient pulmonary reserve and in a good general condition. Local recurrence or progression of HCC, history of liver cirrhosis, and preoperative AFP were found to be independent prognostic factors. If there is a local recurrence or progression of HCC or if the level of AFP before metastasectomy is high, decisions on surgical treatment should be made with caution. According to the results of the propensity score matching analysis, VATS metastasectomy provided outcomes comparable to those afforded by open metastasectomy. The most important prerequisite for pulmonary metastasectomy is complete resection.  Fig. 3 Survival rate after the first pulmonary metastasectomy in matched group; overall survival rate (a), disease-free survival rate (b), pulmonary specific disease-free survival rate (c)