Available online at www.sciencedirect.

com

ScienceDirect
EJSO 42 (2016) 1229e1235 www.ejso.com

Factors influencing survival after hepatectomy for

metastases from gastric cancer
a, a b c
G.A.M. Tiberio
d
*, S. Ministrini
e
, A. Gardini
f
, D. Marrellig ,
A. Marchet , aC. Cipollari , L.
b
Graziosi , C.
c
Pedrazzanif
,
G.L. Baiocchi , G. La Barba , F. Roviello
e
, A. Donini , G.
de Manzoni ,
on behalf of the Italian Research Group for Gastric Cancer
a
Surgical Clinic, Department of Clinical and Experimental Sciences. University of Brescia, Italy
b
Department of General Surgery, Morgagni Hospital, Forl, Italy
c
Surgical Oncology, Department of Human Pathology and Oncology, University of Siena, Italy
d
Department of Oncological and Surgical Sciences, University of Padova, Italy
e
Division of General Surgery, University of Verona, Italy
f
General Surgery, Department of Surgical Sciences, Radiology and Dentistry, University of Perugia, Italy
g
Division of General and Hepatobiliary Surgery, University of Verona, Italy
Accepted 31 March 2016
Available online 19 April 2016

Abstract

Purpose: To investigate clinical factors influencing the prognosis of patients submitted to hepatectomy for metastases from gastric cancer
and their clinical role.
Methods: Retrospective multi-center chart review. We evaluated how survival from surgery was influenced by patient-related, gastric
cancer-related, metastasis-related and treatment-related candidate prognostic factors.
Results: One hundred and five patients submitted to hepatectomy for metastases from gastric cancer, in the synchronous and metachronous
setting of the disease. In 89 cases a R0 resection was achieved, while in 16 a R hepatic resection was performed. Adjuvant chemotherapy
was administered to 29 patients.
Surgical mortality was 1% and morbidity 13.3%.
Median disease-free survival was 10 months, median overall survival was 14.6 months. Overall 1, 3, and 5-year survival rates were
58.2%, 20.3%, and 13.1%, respectively. Survival was influenced independently by the factor T of the gastric primary (p < 0.001), by the
curativity of surgical procedure (p 0.001), by the timing of hepatic involvement (p < 0.001) and by adjuvant chemotherapy (p < 0.001).
T4 gastric cancer, R resection, synchronous metastases, and abstention from adjuvant chemotherapy were associated with a worse
prognosis; T4 gastric cancer and R resections displayed a cumulative effect (p < 0.001).
Conclusions: Our data show that R0 resection must be pursued whenever possible. Furthermore, in the synchronous setting, the
coexistence of T4 gastric primaries and R resections suggests prudence and probably abstention from hepatectomy. Finally, a multimodal
treatment associating surgery and chemotherapy offers the best survival results.
2016 Elsevier Ltd. All rights reserved.

Keywords: Gastric cancer; Hepatic metastasis; Prognostic factors; Hepatectomy

Introduction

* Corresponding author. Surgical Clinic, Department of Clinical and In recent years surgical treatment of hepatic metastases
Experimental Sciences, University of Brescia, 1, Piazzale Spedali Civili, from gastric cancer has been discussed more and more. We
25100 Brescia, Italy. Tel.: 39 3358159298; fax: 39 0303397476. participated in the debate through previous publications,
E-mail address: guido.tiberio@unibs.it (G.A.M. Tiberio).

http://dx.doi.org/10.1016/j.ejso.2016.03.030 0748-
7983/ 2016 Elsevier Ltd. All rights reserved.
1230 G.A.M. Tiberio et al. / EJSO 42 (2016) 1229e1235

focused on unselected populations affected by gastric metastasis-related, and treatment-related candidate prog-
cancer and hepatic metastases. We showed that simple nostic factors.
clinical ele-ments may effectively help in the selection of Patient-related factors included gender and age (the me-
good candi-dates for surgery from those who will not dian age of 68 years was the cut-off). Gastric cancer-related
benefit from an aggressive therapeutic choice, both in the factors included tumor location, factors T and N of the
1 2
metachronous and in the synchronous setting of the TNM system, grading and histological type (Lauren classi-
disease. Moreover, we showed that surgical resection of the fication). Metastasis-related factors included factors H and
tumor bulk is a major favorable prognostic determinant. P (peritoneal metastases were sometimes detected at surgi-
The present work focuses on the subgroup of patients cal exploration). Treatment-related factors included radical-
submitted to surgical management of hepatic metastases, ity (R0/R), type of gastrectomy, extension of
following the mainstream of existing literature. In this sub- lymphectomy and adjuvant chemotherapy. The above
group of patients multiple factors, mainly related to the mentioned variables are detailed in Table 1.
gastric primary and to the hepatic metastases, had been re-
ported as prognostic determinants. However, most studies Statistical analysis
consider very small cohorts of patients and report discor-
dant data regarding the candidate prognostic factors, con-
Descriptive statistics are presented as median and inter-
founding their clinical relevance and practical value.
quartile range (IQR, 25%e75%) or confidence interval
Considering that the magnitude of cohorts may represent a (CI). Overall survival (OS) was measured from the date of
major structural bias, we designed this study to investi-gate resection to the date of death or last follow-up. Recur-
prognostic factors after hepatectomy for metastases from rence-free survival (RFS) was defined as the time from the
gastric cancer and their clinical role, with the strength of the date of resection to the time of recurrence. Survival curves
largest Western cohort of patients, gathered through the
were generated by the KaplaneMeier method, and
network of the Italian Gastric Cancer Research Group.
statistical significance was determined using the log-rank
test. Only variables resulted statistically significant (p <
Materials and methods 0.05) at univariate analysis were considered for multi-
variate analysis with Cox proportional hazards model.
Retrospectively, we reviewed gastric cancer patients
submitted to hepatectomy for metastases, both in the syn- Table 1
chronous and metachronous setting of the disease, during Clinical characteristics of the population.
the period 1990e2013.
Characteristics Synchronous Metachronous
Data were extrapolated from a prospectively collected n 74 n 31
multicentric database, shared by 7 institutions members of Sex Male 50 21
the Italian Research Group on Gastric Cancer. Data were Female 24 10
managed according to institutional rules, with patient Median age 68 years
consent. Gastric tumor location Upper third 30 8
Middle third 26 3
Preoperative work-up systematically included a
Distal 10 4
Computed Tomography, while hepatic MRI and staging Linitis 1 0
laparoscopy were not routinely employed. Gastric stump 2 0
Patients with preoperative evidence of concomitant x 5 16
extra-hepatic metastases or direct infiltration of the hepatic Histology Intestinal 47 12
parenchyma from the gastric primary were not considered. Diffuse 11 2
Other 6 1
Pathologic data concerning the gastric primary were
x 10 16
collected as suggested by the General Rules of the IGCA Grading G1e2 14 6
and classified following the 7th AICC-TNM system. The G3 25 5
extent of hepatic involvement was classified according to x 35 20
3
the JGCA H grading of liver involvement ; this T T<4 24 14
information was obtained from preoperative imaging and T4 45 1
from surgical reports. x 5 16
4 N N0e1 27 9
Follow-up was structured as already described and N2 6 5
stopped on April 30, 2015. N3 28 1
Post-operative adjuvant chemotherapy was not routinely x 13 16
employed. Protocols varied among the centers and during H H1 43 27
the study period, with combinations based on 5- H2 17 4
fluorouracil or taxane. H3 14 0
P P0 69 31
We evaluated how survival from diagnosis was influ- P1 5 0
enced by patient-related, gastric cancer-related,
x: unknown.
 G.A.M. Tiberio et al. / EJSO 42 (2016) 1229e1235 1231

Results In the synchronous setting we discovered intraopera-

In this study we consider 105 patients, 71 males and 34
females. Median age was 68 years (IQR: 57.5e74.5 years).
Seventy-four patients presented synchronous hepatic
metastases at the diagnosis of gastric cancer, while 31
developed metachronous hepatic metastases after curative
surgery for gastric cancer; in this case the median RFS was
21.6 month (IQR: 8.3e31.6 month).
Gastric cancer was treated in all patients by radical sur-
gery, associated in most cases to D2 (79.8%) or D3
(10.5%) lymphectomy (Table 2).
Hepatic resection was achieved by metastasectomy in
73 cases (69.5%), segmetectomy/bisegmentectomy in 25
cases (23.8%) and by major hepatic resection in the
remaining 11 (10.5%). Twenty-seven patients received
more than 1 meta-stasectomy (25.7%).
An R0 resection was obtained in 89 cases (84.6%);
indeed, in 16 cases (15.4%) the hepatic resection was
macroscopically (R2, 9 patients) or microscopically (R1, 7
patients) not curative, mainly due to the position of the
metastases or to unexpected intraoperative up-staging of
the factor H.
tively that 5 patients presented unexpected limited perito-
neal metastases (P): in all cases peritonectomy was
performed.
Post-operative complication rate was 13.3% and only 1
patient (0.9%) died during the post-operative period.
Adjuvant chemotherapy was administered to 29 patients
(27.6%).
Thirteen patients (12.4%) were alive at the time we
stopped follow-up; ten of them had no tumor relapse, while
3 had intrahepatic recurrence. Two patients died due to a
cerebrovascular accident and one owing to colorectal can-
cer; 3 patients were lost at follow-up. The majority of pa-
tients (85, 82%) died because of gastric cancer recurrence.
This was systemic in 7 cases, peritoneal in 25 and
exclusively intrahepatic in 22; details of recurrence are
unknown for 31 cases. In case of T4 gastric cancer the
peritoneal pattern of recurrence was observed in 17/ 31
cases (55%). Five patients with intrahepatic recurrence
were resubmitted to surgical resection, associated to sys-
temic chemotherapy in 2 cases and to PEI in 1 case.
One-year, 3-year, and 5-year DFS rates after surgery
were 48%, 20.2%, and 8.6% respectively, with a median

Table 2
Treatment details and post-operative results.
Treatment and operative results Synchronous mets Metachronous mets
n 74 % n 31 %
Gastric primary Subtotal gastrectomy 27 36.5 3 9.7
Total gastrectomy 40 50.1 12 38.7
Re-gastrectomy 2 2.7 0 0
X 5 6.7 16 51.6
Associated resections Splenectomy 6 8 0
Colonic resection 3 4 0
Other 3 4 0
Lymphectomy D1 8 10.9 0
D2eD3 64 68.5 31 100
x 2 2.7 0
METASTASES H Segmentectomy 8 10.8 9 29.0
Bisegmentectomy 4 5.4 2 6.4
1 metastasectomy 39 52.7 7 22.6
2 metastasectomies 14 18.9 3 9.7
>2 metastasectomies 8 10.8 2 6.4
Major resection 3 4 8 25.9
P No treatment 0 0
Peritonectomy 5 6.7 0
Curativity R0 58 78.4 31 100
R 16 21.6 0
Post-operative complications Morbidity Total 12 16.2 2 6.4
Anastomotic leak 5 6.7
Bleeding 1 1.3
Liver resection-related 2 2.7 2 6.4
Cardiac 2 2.7
Respiratory 2 2.7
Mortality a 1.3 0
1
Adjuvant chemotherapy Fluorouracil 14 3
Taxane 10 2
a
Patient submitted to R2 hepatic resection.
1232 G.A.M. Tiberio et al. / EJSO 42 (2016) 1229e1235

DFS of 10 months (CI 5e14.9). At similar time points OS Role of R0 resection

rates after surgery were 58.2%, 20.3%, and 13.1% respec-
tively (Fig. 1), with a median OS of 14.6 months (CI
11.7e17.4).
Considering the entire cohort of 105 patients, statistical
analysis showed that variables T, timing of hepatic involve-
ment, curativity of surgical resection and postoperative
chemotherapy displayed a significant, independent relation-
ship with survival (Table 3). T4 gastric cancer, R resec-tion,
synchronous hepatic involvement and abstention from
postoperative chemotherapy were associated with a worse
prognosis (p 0.001), the first two also displayed a clear (p <
0.001) cumulative effect (Fig. 2).
Discussion
Our work produced several findings, expected and
unex-pected, in accordance and in contrast to the existing
litera-ture, which deserve attention and merit discussion.
We were surprised by the absence, once excluded the fac-
tor T of the gastric primary, of other gastric cancer or
metastasis-related prognostic variables emerging from our
data. The same factors that we suggested consider in order to
select the therapeutic approach at diagnosis, the factor H in the
2
synchronous setting and the factors N and G of the gastric
1
primary in the metachronous, did not display prog-nostic
value in the subpopulation of patients submitted to sur-gical
treatment. Indeed, this enhances the surgeons role in the
management of these particular cases, as our data show that
R0 resection of the tumor bulk is a major prognostic fac-tor
and suggests that no effort must be spared to achieve it, also
5e7
referring to non surgical ablative techniques. Median
survival exceeds 16 months after curative surgery and drops to
6 months in case of R resection. In our hands 6 months is the
median survival of patients submitted to palliative non
2
ablative surgery in the synchronous setting or to supporting
treatments in the metachronous.
1

Figure 1. Survival curves: overall, T < 4, T4, synchronous and metachronous metastases. Median overall survival was 14.6 months. Median survival for T <
4 was 28.6 months (CI 8e45) vs 8 months (CI 6.2e9.8) for T4 (p < 0.001). Median survival for metachronous presentation was 31 months (CI 23e39) vs 11
months (CI 7.2e14.7) for synchronous presentation (p < 0.001).
 G.A.M. Tiberio et al. / EJSO 42 (2016) 1229e1235 1233
Table 3

Prognostic factors: univariate and multivariate analysis.

Prognostic variables n Median survival 95% CI Univariate p-value Multivariate p-value Odds ratio ( 95% C.I.)
Age <68 years old 61 19.6 10.5e28.7 <0.001 n.s.
68 years old 44 10.2 5.9e14.5
T <4 38 26.8 8.7e44.9 <0.001 <0.001 6.8 (2.5e18.3)
4 46 8.0 6.2e9.8
N 0e1 36 13.9 6.6e21.2 0.021 n.s.
2 11 24.0 2.9e45.1
3 29 11.2 5.1e17.2
H 1 70 15.9 10.6e21.2 0.042 n.s.
2 21 13.0 4.6e21.4
3 14 8.2 3.7e12.7
P 0 100 15.2 12.1e18.3 <0.001 n.s.
5 3.1 2.0e4.2
Timing Synchronous 74 11.0 7.2e14.7 <0.001 <0.001 6.0 (2.3e15.1)
Metachronous 31 31.0 23.0e39.0
Curativity R0 89 16.1 11.4e20.8 <0.001 0.001 2.9 (1.5e5.5)
R 16 6.1 5.1e7.1
Chemotherapy Yes 29 24.4 10.1e38.7 <0.001 <0.001 3.9 (2.2e7.0)
No 50 8.2 3.9e12.5
Only significant results at univariate analysis are shown.

On this point, our data confirm those produced by some
9e11
European authors but are in clear contrast with those
5 biology, not considered in our study. Potentially, they may
reported by Cheon et al., who observed similar survival
performances after R0 or R resection. To our eyes, and in
accordance to other findings of our work, this can be
probably explained by the positive impact of
chemotherapy, administered to a higher percentage of
patients (88% vs 27.6%) in the Korean series.
In the synchronous setting R0 resection must be
achieved both on the hepatic metastases and on the gastric
primary. In our institutions gastrectomy for advanced
gastric cancer is routinely associated to standard D 2
lymphectomy. We consider that this attitude should be
encouraged, recommended and promoted. We are
conscious that our data do not support this attitude, but we
speculate that it would be grotesque to record a lymphatic
persistency after a complex procedure combining
gastrectomy and hepatectomy.
We would like to insist on the fact that once the indica-
tion to surgery is given and a R0 resection can be achieved,
the extent of hepatic involvement no longer influences the
prognosis. This finding is in contrast with data from some
5e10,12,13
of the most numerous cohorts published and
must be considered with full attention. From the clinical
point of view, to assume that patients can obtain good sur-
vival performances even in presence of multiple scattered
metastases in both lobes of the liver (H3), if all of them can
be removed safely, pushes the tight limits in which the
surgical indication is restricted in this particular field. More
importantly and from a speculative point of view, this
enforces the idea that hepatic metastases may still be
included in the concept of regional disease, which may
benefit from regional surgery. This is not a new concept
and it is validated for metastases from colorectal cancer,
but it is a breakthrough for metastases from gastric cancer.
Discussing the role of R0 resection we should also
consider other variables, such as the tumor burden and
offer different perspectives for the discussion of our results:
it is clear that R0 resection has an inverse relation-ship
with the tumor burden and that tumor biology influ-ences
survival but, unfortunately, our data did not allow an
insight into these elements.
The other prognostic factors
The other prognostic elements emerging from our data
offer little argument for discussion; they were expected and
documented in literature.
It is admitted that the late presentation of hepatic metas-
9,10,14e17
tases is associated with better survival. The
different chronology of hepatic involvement may charac-
terize tumors with different biology but, probably, witness
a different selection of candidates for surgery, metachro-
nous lesions being super-selected. It is probably for this
reason that in the metachronous setting we never recorded
R resections.
The prognostic role of the factor T of the gastric
primary, negative when rated T4, is widely
17e20
recognized. Indeed, the involvement of the serosa is a
watershed between a neoplasm still theoretically limited
into the gastric wall and one with diffusive potential into
the entire peritoneal cavity.
The beneficial role of adjuvant chemotherapy was easily
foreseeable, as hepatic metastases are expression of haema-
togenous diffusion that may at best reap advantage from
systemic treatment completing surgical debulking. A multi-
disciplinary approach is needed for the treatment of meta-
5,10,12
static patients. In our cohort neoadjuvant
chemotherapy was never administered as during the study
1234 G.A.M. Tiberio et al. / EJSO 42 (2016) 1229e1235

Figure 2. Survival according to the number of the negative prognostic factors (T4 and R) in the synchronous setting. Median survival of patients without
risk factors was 35 months (CI 28e42). Median survival of patients with 1 risk factor was 13 months (CI 8.5e17.5). Median survival of patients with 2 risk
factors was 8 months (CI 6.8e9.1).

period in our institutions neoadjuvant protocols did not
include metastatic patients; furthermore, only 29 patients
received adjuvant chemotherapy due to inhomogeneous at-
titudes in the different centers. Notwithstanding, chemo-
therapy displayed a clear survival benefit (24.4 months vs
8.2 months of those who did not receive it), especially
appreciable in the medium and long-term (3/4 patients that
survived more than 5 years had received chemo-therapy). It
is of note that in our study the candidate prog-nostic
variable was chemotherapy, and not response to
chemotherapy, as is normally the case. This underlines the
still archaic setting of this particular subject and en-lightens
the direction of future work.
Clinical value of prognostic factors
The association of T4 primaries, R resection and syn-
chronous hepatic metastases has negative effects on sur-vival,
whose median drops to 10.2 months if the first 2 coexist and
to 6.2 months if the 3 are all present, question-ing the rationale
for an aggressive attitude. Considering that in the
metachronous setting we never recorded R resec-tions, these
prognostic factors display a clinical role only in the
synchronous setting, when indication for surgery is
forced (as in our cases) by symptomatic primaries that in-
crease surgical risks. It is always difficult to extrapolate a
therapeutic indication from retrospective data, but we are
confident enough to suggest extreme prudence if a com-
plete removal of hepatic metastases cannot be achieved.
These patients should be considered on a case-by-case
basis with a certain propensity toward abstention from
hepatec-tomy, and surgery e if ineludible-should be limited
to the symptomatic gastric primary. The low procedural
morbidity and mortality rate recorded in our and others
series is no argument in favor of surgery. It is recognized
that if per-formed in the respect of established indications,
in tertiary institutions radical surgery is a low-risk
procedure but, in order to improve results maximally and
expand this partic-ular indication, which is at present
marginal, the surgeon must avoid adventures and adhere to
solid principles of common-sense based surgery.
Conclusion
In conclusion, our data show that in this
particular subset of patients the surgeon plays
a fundamental role when and if it is possible to
achieve R0 resection, which is a major
favorable prognostic determinant.
Furthermore, in the
 G.A.M. Tiberio et al. / EJSO 42 (2016) 1229e1235
synchronous setting of the disease, T4 gastric primaries and
R resections display a clinical role as their coexistence
suggests prudence and probably the simple palliation of
symptomatic gastric primaries. Finally, a multimodal treat-
ment associating surgery and chemotherapy offers the best
survival results.
Conflict of interest statement
The authors declare no conflict of interest.
Acknowledgements
Authors are grateful to Mr. Richard Humphries (B.A.
Hons. English Literature) for his friendly and precious help
in revising the English language of the manuscript.
References
1. Tiberio GA, Coniglio A, Marchet A, et al. Metachronous hepatic me-
tastases from gastric carcinoma: a multicentric survey. Eur J Surg On-
col 2009 May;35(5):48691.
2. Tiberio GA, Baiocchi GL, Morgagni P, et al. Gastric cancer and syn-
chronous hepatic metastases: is it possible to recognize candidates to
R0 resection? Ann Surg Oncol 2015 Feb;22(2):58996.
3. Japanese Gastric Cancer Association. Japanese classification of
gastric carcinoma: 3rd English ed. Gastric Cancer 2011; 14:10112.
4. Marrelli D, De Stefano A, de Manzoni G, et al. Prediction of recur-
rence after radical surgery for gastric cancer: a scoring system ob-
tained from a prospective multicenter study. Ann Surg 2005; 241:
24755.
5. Cheon SH, Rha SY, Jeung H-C, et al. Survival benefit of combined
curative resection of the stomach (D2 resection) and liver in gastric
cancer patients with liver metastases. Ann Oncol 2008;19:114653.
6. Ueda K, Iwahashi M, Nakamori M, et al. Analysis of the prognostic
factors and evaluation of surgical treatment for synchronous liver
metastases from gastric cancer. Langenbecks Arch Surg 2009;394:
64753.
1235
7. Oki E, Tokunaga S, Emi Y, et al. Kyushu Study Group of Clinical
Can-cer. Surgical treatment of liver metastasis of gastric cancer: a
retro-spective multicenter cohort study (KSCC1302). Gastric Cancer
2015 Aug 11. http://dx.doi.org/10.1007/s10120-015-0530-z [Paper
published ahead of print].
8. Makino H, Kunisaki C, Izumisawa Y, et al. Indication for hepatic
resection in the treatment of liver metastasis from gastric cancer. Anti-
cancer Res 2010;30:236776.
9. Chiche L, Ducreux M, Lebreton G, et al. Metastases hepatiques des
cancers de lestomac. In: Adam R, Chiche L, editors. Chirurgie des
metastases hepatiques de cancers non colorectaux non endocrine.
Monographies de lassociation Francaise de Chirurgie. Paris: Arnette;
2005;, p. 459.
10. Schildberg CW, Croner R, Merkel S, et al. Outcome of operative ther-
apy of hepatic metastatic stomach carcinoma: a retrospective analysis.
World J Surg 2012;36:8728.
11. Thelen A, Jonas S, Benckert C, et al. Liver resection for metastatic
gastric cancer. Eur J Surg Oncol 2008;34:132834.
12. Koga R, Junji Y, Shigekazu O, et al. Liver resection for metastatic
gastric cancer: experience with 42 patients including eight long-term
survivors. JJCO 2007;37:83642.
13. Kinoshita T, Kinoshita T, Saiura A, Esaki M, Sakamoto H, Yamanaka
T. Multicentre analysis of long-term outcome after surgical resection
for gastric cancer liver metastases. Br J Surg 2015 Jan; 102(1):1027.
14. Imamura H, Matsuyama Y, Shimada R, et al. A study of factors influ-
encing prognosis after resection of hepatic metastases from colorectal
and gastric carcinoma. Am J Gastroenterol 2001;96:317884.
15. Okano K, Maeba T, Ishimura K, et al. Hepatic resection for metastatic
tumors from gastric cancer. Ann Surg 2002;235:8691.
16. Zacherl J, Zacherl M, Scheuba C, et al. Analysis of hepatic resection
of metastasis originating from gastric adenocarcinoma. J Gastrointest
Surg 2002 Sep-Oct;6(5):6829.
17. Ambiru S, Miyazaki M, Ito H, et al. Benefits and limits of hepatic
resection for gastric metastases. Am J Surg 2001;181:27983.
18. Ochiai T, Sasako M, Mizuno S, et al. Hepatic resection for metastatic
tumours from gastric cancer: analysis of prognostic factors. Br J Surg
1994;81:11758.
19. Sakamoto Y, Sano T, Shimada K, et al. Favorable indications for hep-
atectomy in patients with liver metastasis from gastric cancer. J Surg
Oncol 2007;95:5349.
20. Takemura N, Saiura A, Koga R, et al. Long-term outcomes after
sur-gical resection for gastric cancer liver metastasis: an
analysis of 64 macroscopically complete resections.
Langenbecks Arch Surg 2012; 397:9517.