Surgical treatment of liver metastases of gastric cancer: is local treatment in a systemic disease worthwhile?

Abstract

Objectives

The prognosis of patients with liver metastases of gastric cancer (LMGC) is dismal, but little is known about prognostic factors in these patients; thus justification for surgical resection is still controversial. The purpose of this study was to review recent outcomes of hepatectomy for LMGC and to determine which patients represent suitable candidates for surgery by assessing surgical results and clinicopathologic features.

Methods

Outcomes in 21 patients with LMGC who underwent hepatectomy between 1998 and 2007 were assessed. Isolated metastases and potential to perform a curative resection were requisite indi-cations for surgery. Surgical outcome and clinicopathologic features of the hepatic metastases were analysed.

Results

Overall 1-, 3- and 5-year survival rates after hepatic resection were 68%, 31% and 19%, respectively; three patients survived for >5 years without recurrence. Univariate analysis revealed a solitary metastasis, negative margin (R0) resection and the presence of a peritumoral fibrous capsule as significant favourable prognostic factors. These characteristics were present in all of the three patients who survived for >5 years.

Conclusions

Solitary metastases from gastric cancer should be treated surgically and confer a better prognosis. Surgical resection should provide microscopically negative margins (R0). A new prognostic factor, the presence of a pseudocapsule, may be associated with improved prognosis.

Introduction

Although the effectiveness of liver resection for metastatic colorectal cancer has already been established,^1–3 reports of hepatic resection for liver metastases of gastric cancer (LMGC) are rare and its efficacy is still controversial.⁴ In fact, a number of studies have reported that the effects and benefits of hepatic resection in either synchronous or metachronous gastric hepatic metastases (GHM) are dubious.⁵ Surgical indications for liver metastases from colorectal cancer have been expanded to include all technically resectable metastases under four or more.⁶ Yet, the surgical indications for LMGC must be carefully determined because of the more severe biologic nature of this disease.⁷

Most patients with gastric cancer with concomitant liver metastases are excluded as candidates for potential curative surgery because of the presence of synchronous distant extrahepatic or locally advanced disease.⁸ In fact, GHM often represent only part of a generalized spread of the primary tumour. Furthermore, very few patients with GHM are good candidates for liver surgery because most patients have multiple, scattered, bilobar lesions.⁹ Patients with isolated metastases are unusual and accounted for 0.5% of patients in a series reported by Linhares et al.¹⁰ Baba et al. have shown that outcomes in patients who undergo non-curative resection for advanced gastric cancer are extremely poor.¹¹ Several authors have reported limited experiences of surgical resection of GHM in selected patients, with 5-year survival rates of 0–38%.^12–14 Therefore, identifying the patients who are most likely to benefit from surgical intervention in the presence of GHM is clinically important. The aim of this study was to review a single-institution experience to determine the benefits of hepatic resection in patients with GHM and to identify important prognostic factors.

Materials and methods

A total of 984 patients with gastric cancer (adenocarcinoma) underwent surgery at the Department of Surgery, San Gerardo Hospital, University of Milan Bicocca, between January 1998 and December 2007. Of these, 36 patients (3.7%) had synchronous liver metastases and 31 patients (3.2%) developed metachronous liver metastases after resection of the primary gastric cancer. Of these 67 patients, 21 (31.3%) consecutive patients underwent hepatic resection with intent to cure and were selected for this study. Outcomes in these patients were retrospectively reviewed.

Criteria for resection of hepatic metastases from gastric cancer were as follows: (i) no local relapse of the primary tumour in metachronous metastases and curative resection of the primary tumour and lymph node involvement proved or thought to be technically possible in synchronous disease; (ii) no evidence of disseminated disease after staging investigations, and (iii) ability to achieve a microscopically margin-negative (R0) hepatic resection. Patients undergoing synchronous en bloc resections of gastric cancer directly invading the liver were not included in this study. Synchronous liver metastases were defined by detection before or during surgery, or within 3 months of primary tumour resection. The following clinicopathologic factors were analysed and subgroups divided according to each variable were compared: age; gender; status of serosal invasion; histologic differentiation of the primary tumour; status of lymph node metastases; temporal relationship of metastases with primary disease (synchronous or metachronous); tumour distribution; size and number of liver metastases; type of hepatic resection; surgical margin and completeness of the resection; presence of a pseudocapsule between the metastasis and the liver parenchyma (defined as a wall of fibrotic tissue between the tumour and the normal parenchyma); histologic differentiation of the GHM, and vascular invasion present within the GHM.

In each patient intraoperative liver ultrasound was performed to assess the extent and number of hepatic lesions and their relationships to intrahepatic vascular and biliary structures. Preoperative workup included computed tomography and/or magnetic resonance imaging. Liver surgical procedures were classified as anatomic resection (segmentectomy and hemi-hepatectomy) or limited resection (for all resections less extensive than segmentectomy). Operative death was defined as death occurring within 30 days of the operation; in-hospital death was defined as any death occurring before discharge. Morbidity included any type of complication, whether surgical or non-surgical. Overall survival time was measured from the date of hepatic resection to the date of death from any cause. All patients were observed at intervals of 3–4 months during the first 2 years of the study and every 6 months or 12 months thereafter for 3 years.

Statistical analysis

Data were analysed using spss Version 19 (SPSS, Inc., Chicago, IL, USA) for survival analysis, univariate analysis and Kaplan–Meier curves. Comparisons of categorical and continuous variables were performed using the chi-squared test and the Wilcoxon test, respectively. Survival data were compared with the log rank test. Variables with a P-value of <0.1 in the univariate analysis were entered into a Cox regression analysis. A P-value of <0.05 was considered to indicate statistical significance.

Results

Of the 21 patients, 14 were men. The median age of the patients was 64 years (range: 44–89 years). Clinicopathologic characteristics of the patients and data on primary tumours and metastases are depicted in Table 1. Surgical procedures for primary gastric cancer included 11 partial gastrectomies and 10 total gastrectomies. The primary gastric tumour proved to be well to poorly differentiated adenocarcinoma of the stomach in all patients. Serosal invasion was present in eight patients. The median maximum size of the metastatic tumours was 3.0 cm (range: 1.0–8.5 cm). Tumour locations are reported in Table 1. Hepatic resections included four hemi-hepatectomies, five segmentectomies (anatomical resections) and 12 wedge resections (non-anatomical or limited resections). Blood transfusion was required in five patients.

Table 1.

Clinicopathologic characteristics and their prognostic significance

Variable	Patients, n	Proportion surviving, %		P-value
		1 year	3 years
Gastric carcinoma
Lymph node metastasis
N0–N1	10	60.1	48.3	0.11
N2	11	37.4	12.2
Serosal invasion
Absent	13	78.6	44.5	0.315
Present	8	62.4	27.8
Hepatic metastases
Solitary	12	70.3	56.7	0.044
Multiple	9	29.4	10.1
Lobar distribution
Unilobar	16	64.3	50.0
Bilobar	5	39.8	12.1	0.071
Tumour size
<5 cm	14	81.1	39.4	0.331
≥5 cm	7	50.1	37.5
Histologic differentiation
Good	8	80.2	41.6	0.144
Moderate, poor	13	67.3	30.0
Vascular invasion
Absent	13	75.2	44.1	0.211
Present	8	31.3	23.7
Fibrous pseudocapsule
Present	12	75.5	56.1	0.020
Absent	9	51.7	7.1
Surgery
Timing of hepatectomy
Synchronous	12	56.8	28.3	0.115
Metachronous	9	68.3	34.9
Resection margin
Negative	19	71.1	35.4	0.002
Positive	2	0.0	0.0
Type of surgery
Anatomic	9	75.6	33.2	0.846
Non-anatomic	12	70.6	31.8

In the nine patients who underwent metachronous liver resection, the median interval between gastric resection and the diagnosis of hepatic metastasis [disease-free interval (DFI)] was 10 months (range: 5–15 months). These patients showed no evidence of hepatic involvement at the time of surgery for the primary tumour. In 19 patients the resection was classified as curative (R0), whereas microscopically positive margins were identified in two patients. With regard to pathologic features in the peritumoral section of the metastatic lesion, fibrosis (fibrous pseudocapsule) was found in 12 patients. In-hospital and 30-day postoperative mortality was zero. The overall morbidity rate was 19.0% (four patients); complications comprised transient liver failure, bile leakage, pleural effusion and wound infection in one patient each. Median follow-up was 20 months (range: 6–90 months).

Overall 1-, 3- and 5-year survival rates were 68%, 31% and 19%, respectively. Median survival was 11 months. Rates of 1-, 3- and 5-year disease-free survival were 51%, 25% and 14%, respectively. Three patients remain alive and disease-free at >5 years after the surgical procedure. All three of these patients had a solitary metastasis with a fibrous pseudocapsule. In the 19 patients in whom R0 resection was achieved, 1-, 3- and 5-year survival rates were 71%, 35% and 24%, respectively, whereas the margin-positive procedures in two patients resulted in impaired survival and both patients died within 1 year of resection. Recurrence occurred in 14 patients. In 13 of these patients, disease recurred within the liver. No patient underwent a second hepatic resection for recurrent intrahepatic metastasis. The results of univariate analysis identifying poor prognostic factors are depicted in Table 1.

Discussion

In a review of the literature, hepatectomy was found to be indicated in only 0.4–1.0% of gastric cancer patients with liver metastases.⁴ Unfortunately, most hepatic metastases from gastric adenocarcinoma are multiple, bilateral and combined with peritoneal or lymph node metastases, which directly invade adjacent organs, thereby precluding a radical surgical approach.⁴ At specialized treatment centres, the proportion of surgery carried out for LMGC is 7%, whereas that of hepatic resection for all types of hepatic malignancies is 12%.¹⁵ Resectability does not seem to be affected by the presence of metachronous or synchronous metastases.¹⁶

The effectiveness of hepatic resection for GHM has not been well defined. In addition, clinicopathologic factors that affect the prognosis of patients with gastric cancer with hepatic metastases have not been comprehensively identified. Nevertheless, the presence of hepatic metastases is a statistically significant poor prognostic factor for patients with gastric cancer.¹⁷

In recent series, 1-year survival rates have ranged from 42% to 90% and 5-year survival rates from 0% to 38% (Table 2). Most studies concerning this issue come from Japan, where reported longterm survival rates exceed 30% in some series.^14,18 By contrast, Zacherl et al. reported a study in which no patients survived 5 years after resection.¹⁵ Bines et al.¹⁹ reported one longterm survivor in seven patients and other series have shown longterm survival rates of 11–19%. Although they are few, some longterm survivors after hepatic resection have been reported. Therefore, determining the indications for liver resection in the presence of GHM is crucial.

Table 2.

Literature analysis regarding hepatectomy for liver metastases from gastric cancer

Author(s) (year)	n	Period	Resection criteria	Resectability rate, %	S/M	TG/STG	Major/minor liver surgery	Solitary	Multiple uni-/bilobar	R1, %	Overall survival at 1, 3, 5 years	Longterm survivors, % (n)	Recurrence	Follow-up, months	Prognostic factors for poor survival
Ochiai et al. (1994)20	21		No extrahepatic lesions	NA	13/8	NA		14	7	0	19%	19% (4)			Primary serosal invasion
			No carcinosis												Primary venous invasion
			R0 resection												Primary lymphatic invasion
Miyazaki et al. (1997)25	21		No extrahepatic lesions	NA	11/10		5/16	7	14	NA	45%, 19%, 11%	24% (5)	76.1%	NA	Primary serosal invasion
			No carcinosis						11/3						Primary venous invasion
			R0 resection												Primary lymphatic invasion
															Surgical margin of <10 mm
Imamura et al. (2001)12	17	1990–1997	No extrahepatic lesions	NA	7/10	NA	6/11	8	9	18%	47%, 22%, 0%	0	76%	NA	Primary serosal invasion
			No carcinosis						4/5						Primary venous invasion
			R0 resection												Primary lymphatic invasion
															Primary lymphatic invasion
															Extrahepatic disease
Ambiru et al. (2001)5	40	1975–1999	No extrahepatic lesions	NA	18/22	19/21	21/19	19	21	0	70%, 28%, 18%	15% (6)	75%	88	Resection margin of <10 mm
			No carcinosis						5/16						Synchronous
			R0 resection
Fujii et al. (2001)28	10	1979–1999	NA	NA	3/7	3/7	6/4	6	4	NA	60%, 20%, 20%	10% (1)	80%	10–240	DFI <1 year on metachronous
									2/2						Diameter >5 cm
															Absence of TILs
Zacherl et al. (2002)15	15	1980–1999	No extrahepatic lesions	NA	10/5	9/6	3/12	8	7	33	35.7%, 14.3%, 0%	0	90%	NA	Primary site (proximal)
			No carcinosis						2/5						Synchronous
			R0 resection												Multiple bilobar metastases
															R1 resection
Saiura et al. (2002)24	10	1981–1998	No extrahepatic lesions	15.6%	7/3	NA	6/4	5	5	40%	65%, 38%, 20%	20% (2)	80%	1–68	Primary lymphatic invasion
			≤3 segments						4/1
Okano et al. (2002)18	19	1986–1999	No extrahepatic lesions	17%	13/6	NA	7/12	10	9	0	77%, 34%, 34%	14% (3)	74%	NA	Multiple
			No carcinosis						2/7						Synchronous
			R0 resection												Absence of pseudocapsule
Sakamoto et al. (2003)14	22	1985–2001	No extrahepatic lesions	8%	12/10	10/12	3/19	16	6	0	73%, 38%, 38%	20% (5)	68%	NA	Multiple
			No carcinosis						1/5						Bilobar
			R0 resection												Metastases >5 cm
Shirabe et al. (2003)21	36	1979–2001	NA	NA	16/20	17/19	10/16	NA	NA	0	64%, 43%, 26%	11% (4)	83.3%	NA	Lymphatic invasion
															Venous invasion
															Number of metastases
Koga et al. (2007)16	42	1985–2005	No extrahepatic lesions	17%	20/22	NA	7/35	29	13	0	76%, 48%, 42%	20% (8)	67%	1–86	Multiple
			No carcinosis												Serosal invasion
			R0 resection
Sakamoto et al. (2007)23	37	1990–2005	No extrahepatic lesions	12%	16/21	10/27	5/32	21	16	14%	60%, 27%, 11%	6% (2)	81%	NA	Bilobar
			R0 resection						9/7						Metastases >4 cm
			No carcinosis
Thelen et al. (2008)13	24	1988–2002	No carcinosis	NA	15/9	NA	8/16	13	11	25%	38%, 16%, 10%	8% (2)	65%	1–67	Resection margin
			R0 resection						5/6
Cheon et al. (2008)30	22	1995–2005	No extrahepatic lesions	7.5%	18/4	7/15	3/19	18	4	NA	77%, 30.4%, 23%	15% (3)	63.6%	1–106	Multiple metastases
			No carcinosis						3/1
			R0 resection
			Hepatic function
Nomura et al. (2009)26	17	1991–2005	No extrahepatic lesions	NA	9/8	NA	3/14	NA	NA	0	30.8%	25% (4)	70.5%	1–117	Presence of micrometastases
			No carcinosis
			≤5 lesions
			R0 resection
Tsujimoto et al. (2010)22	17	1980–2007	No extrahepatic lesions	NA	9/8	NA	6/11	13	4	NA	31%	30% (5)	70%	NA	Primary size >6 cm
			No carcinosis												D2 resection
			Unilobar
			R0 resection
Present study (2010)	21	1998–2007	No extrahepatic lesions	31%	12/9	10/11	4/17	12	9	10%	68%, 31%, 19%	14.2% (3)	66%	6–90	Multiple
			No carcinosis						4/5						Resection margin positive
			R0 resection												Absence of pseudocapsule

S, synchronous; M, metachronous; TG, total gastrectomy; STG, subtotal gastrectomy; NA, data not available; DFI, disease-free interval; TILs, tumour-infiltrating lymphocytes.

The 5-year survival rate identified in this series is 19%. Three longterm survival patients were identified, by contrast with findings in another report from a Western country.¹⁵

None of the clinicopathologic features related to the primary gastric cancer were predictors of survival. Ochiai et al.²⁰ reported that hepatic resection should only be attempted in patients with synchronous or metachronous metastases in the absence of serosal or microscopic lymphovascular invasion of the primary tumour.²¹ Zacherl et al.¹⁵ reported that tumour localization in primary gastric cancer was a marginally predictive negative factor for the overall survival of patients. Tsujimoto et al.²² identified gastric cancer tumour size as a predictor of poor survival.

In the current study, the presence of a solitary metastasis was a positive predictor for survival. The number of metastatic nodules in the liver has been reported to be an important prognostic factor in previous studies. Okano et al.¹⁸ reported 3-year survival rates of 56% in patients with a single metastasis and 0% in patients with multiple metastases. Koga et al.¹⁶ and Shirabe et al.²¹ reported no longterm survivors in patients with multiple gastric liver metastases. In a more recent study, Sakamoto et al.²³ also reported the value of a solitary lesion, adding to the evidence in support of unilobar distribution as a good predictive factor for survival.

However, Saiura et al.²⁴ reported two longterm survivors with more than three metastases and concluded that, if curative resection (R0) can be achieved, hepatic resection should not be abandoned even in patients with multiple liver metastases. In fact, even in this series R0 resection proved key to achieving good survival and every effort should be made to achieve radical resection with an adequate margin because this is the only factor on which the surgeon can have any direct influence. Miyazaki et al.²⁵ demonstrated significant differences in the size of the tumour-free resection margin (<10 mm or >10 mm) in long- and short-term survivors. Thelen et al.¹³ reported that a positive resection margin should be considered a powerful determinant of poor outcome. Nomura et al.²⁶ showed that the recurrence rate in the remnant liver was higher in patients with a surgical margin of <5 mm. Isono et al.²⁷ reported that micrometastases around the macroscopic tumour were found more frequently in hepatic metastases from gastric cancer than in those from colorectal cancer. According to the pattern of recurrence, relapse developed most commonly in the liver (median: 70%; range: 63.6–83.3%), indicating that relapse monitoring should focus on the remaining liver. A sensible strategy for improving survival would involve close observation for a second relapse in the liver and the provision of adjuvant chemotherapies after surgery. In this series, only one patient received a postoperative hepatic arterial infusion; this patient survived 38 months. In terms of adjuvant therapy, however, no conclusions can be based on these data.

The timing of hepatic resection has been reported to be a significant prognostic factor and some papers have reported synchronous hepatectomy as a significant poor prognostic factor.^5,19 Nevertheless, an analysis of the data reported in the recent literature showed that 22 of 48 5-year survivors underwent synchronous hepatectomy.²³ Thus, synchronous hepatectomy should not be a contraindication for hepatic resection. The data reported in the present study show comparable median survival rates in patients with synchronous and metachronous lesions. With regard to perioperative morbidity, the current results confirm the observation of Bines et al.¹⁹ that synchronous liver resection carries a higher risk. Moreover, the DFI between gastric and hepatic resection has been reported to represent a prognostic factor. Fujii et al.²⁸ showed that a DFI of >1 year had a significant survival advantage as a result of the slow-growing nature of this tumour. However, this trend was not confirmed in the present study.

The aggregation of lymphocytes enclosing the metastatic tumour was reported as a good prognostic factor by Fujii et al.²⁸ This may be explained by the favourable action of tumour-infiltrating lymphocytes (TILs) in preventing tumour extension in gastric cancer patients.²⁹ Okano et al.¹⁸ demonstrated that the presence of fibrous pseudocapsules around liver metastases is a promising indicator of a better prognosis and is closely associated with patient survival. The formation of a pseudocapsule should be considered as a protective immunoinflammatory reaction against the metastastic nodule and as representing the host defence reaction of creating a wall to stop tumour diffusion.

Conclusions

Although some authors hold the view that metastatic gastric cancer represents a systemic disease and indicates a diffuse cancer, and that surgery has no role in its treatment, a considerable proportion of patients in the present as well as former series were found to be tumour-free at >5 years after liver resection. The present authors believe that surgery may provide a benefit and should be part of a multidisciplinary approach in patients with LMGC.³¹ Solitary metastases, the ability to achieve R0 resection, and the presence of a pseudocapsule are determinants of improved survival.

Conflicts of interest

None declared.