Content uploaded by Riccardo Memeo
Author content
All content in this area was uploaded by Riccardo Memeo on Oct 25, 2017
Content may be subject to copyright.
ORIGINAL SCIENTIFIC REPORT
Margin Status is Still an Important Prognostic Factor
in Hepatectomies for Colorectal Liver Metastases: A Propensity
Score Matching Analysis
Riccardo Memeo
1,2,3
•Vito de Blasi
1,2,3
•Rene Adam
4
•Diane Goe
´re
´
5
•
Tullio Piardi
6
•Emilie Lermite
7
•Olivier Turrini
8
•Francis Navarro
9
•
Nicola de’Angelis
10
•Antonio Sa Cunha
4
•Patrick Pessaux
1,2,3,11
•French Colorectal Liver
Metastases Working Group, Association Franc¸aise de Chirurgie (AFC)
ÓSocie
´te
´Internationale de Chirurgie 2017
Abstract
Objective The width of resection margin is still a matter of debate in case of colorectal liver metastasis resec-
tion. The aim of this study was to determine the risk factors for R1 resection. Once risk factors had been identified,
patients were matched according to Fong’s prognostic criteria, in order to evaluate whether R1 resection still
remained a negative prognostic factor impacting overall and disease-free survival.
Methods A total of 1784 hepatectomies were analyzed from a multicentric retrospective cohort of hepatectomies.
Patients were compared before and after a 1:1 propensity score analysis in order to compare R0 versus R1 resections
according to Fong criteria.
Results Primary tumor nodes found positive after colorectal resection (RR =1.20, p=0.02), operative time
([240 min) (RR =1.26, p=0.05), synchronous liver metastasis (RR =1.27, p=0.02), pedicle clamping ([40 min)
(RR =1.52, p=0.001), lesion size larger than 50 mm (RR =1.54, p=0.001), rehepatectomy (RR =1.68,
p=0.001), more than 3 lesions (RR =1.69, p=0.0001), and bilateral lesions (RR =1.74, p=0.0001) were identified
as risk factors in multivariate analysis. After a 1:1 PSM according to Fong criteria, R1 resection still remained a negative
prognosticfactor impacting overall and disease-free survival, with 1-, 3-, 5-year OS at 94,81, and 70% in R0 and 92, 75, and
58% in R1, respectively, (p=0.008), and disease-free survival (DFS) with 1-, 3-, 5-year survival at 64, 41, and 28% in R0
versus 51, 28, and 18% in R1 (p=0.0002), respectively.
Conclusion Even after using PSM as an oncological prognostic criterion, R1 resection still impacts overall and
disease-free survival negatively.
Collaborators of the French Surgical Association (Association
Franc¸aise de Chirurgie) Working Group are listed in
acknowledgements.
&Patrick Pessaux
patrick.pessaux@chru-strasbourg.fr
1
Institut Hospitalo-Universitaire (IHU), Institute for
Minimally Invasive Hybrid Image-Guided Surgery,
Universite
´de Strasbourg, Strasbourg, France
2
Institut de Recherche Contre les Cancers de l’Appareil
Digestif (IRCAD), Strasbourg, France
3
General, Digestive, and Endocrine Surgery, Nouvel Ho
ˆpital
Civil, Strasbourg, France
4
Department of Surgery, Ho
ˆpital Paul Brousse, Villejuif,
France
5
Department of Surgery, Institut Gustave Roussy, Villejuif,
France
6
Department of Surgery, Ho
ˆpital Robert Debre
´, Reims, France
7
Department of Surgery, CHU Angers, Angers, France
8
Department of Surgery, Institut Paoli-Calmettes, Marseilles,
France
9
Department of Digestive Surgery, Universite
´de Montpellier,
Ho
ˆpital Saint-Eloi, Montpellier, France
10
Department of Surgery, Hopital Henri Mondor, Creteil,
France
11
HepatoBiliary and Pancreatic Surgery Department, 1, place
de l’Ho
ˆpital, 67091 Strasbourg, France
123
World J Surg
DOI 10.1007/s00268-017-4229-7
Abbreviations
AFC Association Franc¸aise de Chirurgie
ASA American Society of Anesthesiologists
BMI Body mass index
CEA Carcinoembryonic antigen
CRC Colorectal cancer
CRLM Colorectal liver metastases
DFI Disease-free interval
DFS Disease-free survival
OS Overall survival
PSM Propensity score matching
RR Relative risk
Introduction
The modern treatment of colorectal liver metastasis
(CRLM) is based on a multidisciplinary approach and on
negative resection margin resections in order to achieve the
best long-term results for metastatic patients. Even if the-
oretically large resection margins should be achieved, this
objective could not always be reached due to the hetero-
geneity of the disease, with variations in number, size, and
location of metastases. In addition, primary tumor charac-
teristics could impact the recurrence rate due to cancer
aggressiveness.
The aim of this study was to determine the risk factors
for R1 resection. Once these factors were identified,
patients were matched according to Fong’s prognostic
criteria, in order to evaluate if R1 resection still remained a
negative prognostic factor impacting overall and disease-
free survival.
Materials and methods
Data were obtained from a questionnaire-based survey of
patients who underwent surgery for colorectal liver
metastases (CRLM) in 32 French centers from January
2006 to December 2013. This study was performed under
the supervision of the French National Surgical Association
(Association Franc¸aise de Chirurgie—AFC) after institu-
tional approval. Files were submitted by surgeons of each
institution. Demographic data, preoperative, intraoperative,
postoperative data, and oncological results were collected
and evaluated. Each site was responsible for data collection
and entry. Files were then submitted to the AFC. Once
anonymized, all questionnaires were merged to create a
single database. To ensure data completeness, question-
naires were sent back to the institutions in case of missing
data ([10% per variable). Once this step was complete,
patients without long-term follow-up information or with
outlying values were excluded. The authors had complete
access to the final dataset.
Study population
This study was designed to evaluate which factors impact
resection margins when patients’ characteristics and
pathological data were similar in a propensity score
matching analysis, comparing R0 resection and R1 resec-
tion. R0 resection was defined as a microscopic complete
resection. R1 resection was defined as an incomplete
ablation margin with tumor cells involving the resection
margin [1]. Patients with incomplete macroscopic resection
(R2) were excluded from the study. Disease-free interval
(DFI) was defined as the interval from the time of primary
cancer resection and liver metastasis diagnosis. Syn-
chronous metastases were defined as metastases detected
by means of preoperative screening or during primary
tumors resection and occurring within 12 months of the
initial colorectal cancer (CRC) diagnosis. Major hepatec-
tomy was defined as the resection of three and more con-
secutive liver segments. The number of resected segments
was defined by the type of surgical resection according to
the Brisbane classification [2]. In case of multiple resec-
tions, the number of segments was added.
Patients were followed up using a serum tumor marker
(carcinoembryonic antigen) (CEA) and thoraco-ab-
dominopelvic computed tomography every 4 or 6 months
depending on the center. Recurrence was defined as intra-
hepatic or extra-hepatic biopsy-proven recurrent adeno-
carcinoma, or a lesion deemed suspicious on cross-sec-
tional imaging. Overall survival (OS) was analyzed from
the date of liver resection to the date of death, and disease-
free survival (DFS) to the date of recurrence. The indica-
tion for adjuvant chemotherapy was discussed in a multi-
disciplinary meeting.
Data analysis
Quantitative variables were presented as mean. Qualitative
variables were presented as numbers and percentages.
Comparison of quantitative variables was performed using
a Mann–Whitney test. Comparison of qualitative variables
was performed using Pearson’s Chi-squared test or Fisher’s
exact test depending on numbers. A pvalue \0.05 was
considered significant. Overall and disease-free survival
probabilities were calculated using the Kaplan–Meier
method.
A propensity score matching (PSM) was calculated to
take into account and reduce selection biases as well as
confusion between the two groups. Considering the Fong
criteria as predictors of recurrence after hepatic resection
World J Surg
123
for CRLM, the two populations of R0 and R1 patients were
matched in order to get two identical populations in pre-
operative oncological data to estimate which factor influ-
ences the feasibility of a margin-free resection. Patients
were matched in a 1:1 analysis with the closest estimated
PS within 0.2 of the PSM standard deviation. For PSM, we
chose variables which are known to potentially affect the
outcome of interest. The propensity score was assessed
using logistic regression including the following variables:
age, gender, comorbidity, body mass index (BMI), ASA
score (American Society of Anesthesiologists), use of
neoadjuvant therapy on primary tumor, primary tumor
resection (yes vs. no), primary tumor localization (colon vs.
rectum), primary tumor lymph node (N0 vs. N?), timing of
metastasis assessment (synchronous vs. metachronous),
neoadjuvant chemotherapy (yes vs. no), liver metastasis
neoadjuvant chemotherapy (yes vs. no), single hepatic
lesion (yes vs. no), largest lesion size [50 mm, bilateral
lesion. CEA was not included due to missing data. The
choice of such variables was based on univariate analysis
results and/or on the known influence of specific factors on
the selection of the intervention type. A 1:1 balance ratio
was used for propensity score matching, based on the
nearest matching PS method. After the matching process,
both groups were compared regarding their initial charac-
teristics in order to re-evaluate the comparability of both
groups. Finally, matched groups could be compared
regarding the different variables of interest in the study.
For univariate analysis, categorical variables were ana-
lyzed using Pearson Chi-squared testing. All values are
expressed as a percentage of the group from which they
were derived (categorical variables). For univariate analy-
sis, p\0.05 was considered significant. Logistic regres-
sion was then performed to identify risk factors for R1
resection. Variables with a p\0.100 in the univariate
analysis were entered into a forward stepwise logistic
regression analysis to estimate the odds ratio (OR) of R1
resection (dependent variables) and the presence or
absence of potential prognostic factors (independent vari-
ables). The odds ratio was defined as the coefficient with
95% confidence intervals (95% CI).
Analyses were performed using the 3.2.0 version R
software (R Core Team, R Foundation for Statistical
Computing, Vienna, Austria).
Results
Among our multicentric data, data concerning surgical
margins in 1784 hepatectomies performed for CRLMs
(1218 R0 resection and 566 R1 resections) were analyzed
in our study. Patients with macroscopic positive resection
margins, non-resectable liver metastasis, extra-hepatic
disease and incomplete data were excluded from the study.
In our population, the median age was of 62 years and
included 60% of male patients.
Comparison of R0 versus R1 hepatectomies
Patients’ characteristics when comparing the two groups
are summarized in Table 1. Patients in the R1 group were
younger (58.8 ±10.7 vs. 61.0 ±11.5, p=0.001), with an
increased presence of comorbidity (43.3 vs. 32.3%,
p=0.0001) and more ASA score 3–4 (17.3 vs. 12.5%,
p=0.01). Concerning primary tumor oncological aspects,
we had more status positive primary nodes in the R1 group
(52.8 vs. 45.2%, p=0.003), more synchronous metastases
(48.2 vs. 39.3%, p=0.0002), higher levels of CEA
(486 ng/mL vs. 195 ng/mL, p=0.007), and less neoad-
juvant chemotherapy (64.7 vs. 70.9%, p=0.008). Con-
cerning characteristics of liver metastasis, there were fewer
single lesions in the R1 group (24.9 vs. 45.1, p=0.0003)
with a higher number of lesions in the R1 group
(4.46 ±4.2 vs. 2.95 ±2.95, p=0.0001) with a larger
size (37.3 vs. 33.7 mm, p=0.0002), and mostly bilateral
(58.1 vs. 40.7, p=0.0002). In surgical data analysis, there
were more portal vein embolizations in the R1 group (20.1
vs. 13.9%, p=0.001), with a higher rate of rehepatec-
tomies (27.6 vs. 18.5%, p=0.0001), a lower rate of
laparoscopy (2.5 vs. 6%, p=0.001), longer operative
times ([240 min) (57.2 vs. 43.1%, p=0.0001), pedicle
clampings ([40 min) (37.5 vs. 22.9%, p=0.007), less
hepatic vein control (31.4 vs. 25.2%, p=0.007), and more
transfusion (21.6 vs. 13.8, p=0.0001).
Prognostic factors for R1 resection
After multivariate analysis (Table 1), the independent
factors for R1 resection were primary tumor nodes positive
after colorectal resection (OR =1.20, p=0.02), operative
time ([240 min) (RR =1.26, p=0.05), synchronous
liver metastasis (OR =1.27, p=0.02), pedicle clamping
([40 min) (OR =1.52, p=0.001), lesion size larger than
50 mm (OR =1.54, p=0.001), rehepatectomy
(OR =1.68, p=0.001), more than three lesions
(RR =1.69, p=0.0001), and bilateral lesions
(RR =1.74, p=0.0001).
Disease-free and overall survival
There was a difference in terms of oncological results, with
1-, 3-, 5-year OS at 92, 82, and 74% in the R0 group and
92, 71, and 56% in the R1 group, respectively (Fig. 1).
DFS was similar with 1-, 3-, 5-year survival at 71, 42, and
33% in the R0 group versus 60, 25, and 21% in the R1
group, respectively (Fig. 2).
World J Surg
123
Comparison of R0 versus R1 hepatectomies
after propensity score matching
Once R0 and R1 group were compared and risk factors for
R1 resection were identified using a multivariate analysis,
we decided to compare R0 and R1 resection with a
propensity score matching regarding patients’ characteris-
tics and Fong’s criteria prognostic factors. After propensity
score matching, we analyzed 517 patients in the R0 group
and 517 patients in the R1 groups (Table 2). By comparing
the matched groups, differences were found in surgical
data, with more rehepatectomies (29 vs. 17%, p=0.0001),
a higher rate of two-stage hepatectomies (21 vs. 16%,
p=0.05), a lower rate of laparoscopy (3 vs. 6%,
p=0.004), longer operative times ([240 min) (56 vs.
49%, p=0.02), and pedicle clampings ([40 min) (37 vs.
29%, p=0.004) in the R1 group.
Prognostic factors for overall and disease-free
survival after propensity score matching
Multivariate analysis demonstrated that R1 resection
(OR =1.38, p=0.05), pedicle clamping [40 min
(OR =1.70, p=0.001), primary N?resection (OR =1.6,
p=0.004), and bilateral lesions (OR =1.6, p=0.0003)
were considered major risk factors impacting overall sur-
vival (Table 3). Multivariate analysis demonstrated that R1
resection (OR =1.50, p=0.0002), neoadjuvant
chemotherapy for liver metastasis (OR =1.70, p=0.002),
primary N?resection (OR =1.3, p=0.01) were
Table 1 Population data before PSM
All patients (n=1784) R0 resection (n=1218) R1 resection (n=566) p
Period of resection, 2010–2013 970 (68.3) 680 (55.8) 290 (51.8) 0.07
Age (year) 60 ±11.2 61.0 ±11.5 58.8 ±10.7 0.001
Gender, male, n(%) 1068 (59.9) 716 (58.7) 352 (61.2) 0.17
ASA score 3–4, n(%) 282 (15.8) 211 (17.3) 71 (12.5) 0.01
Comorbidity, yes, n(%) 711 (39.9) 528 (43.3) 183 (32.3) 0.0001
BMI (kg/m
2
) [median, (IQR)] 25.6 ±3.2 25.8 ±3.9 25.3 ±4.2 0.71
Primary resected, yes, n(%) 1532 (85.9) 1055 (86.6) 477 (84.3) 0.18
Primary rectum, yes, n(%) 477 (26.7) 329 (27) 148 (26.1) 0.70
Primary N?, yes, n(%) 849 (47.6) 550 (45.2) 299 (52.8) 0.003
Primary neoadjuvant chemotherapy, yes (%) 271 (15.2) 193 (15.8) 78 (13.8) 0.25
Disease-free interval [12 months, n(%) 478 (26.2) 325 (26) 153 (27) 0.87
Synchronous liver metastasis, yes, n(%) 753 (42.2) 479 (39.3) 274 (48.4) 0.0002
Liver metastasis CEA (ng/mL) [median, (IQR)] 32 (1–27,000) 195 ±596 486 ±2130 0.007
Liver neoadjuvant chemotherapy, yes, n(%) 1230 (68.9) 864 (70.9) 366 (64.7) 0.008
Portal vein embolization 283 (15.9) 169 (13.9) 114 (20.1) 0.001
Single lesion, n(%) 690 (38.7) 549 (45.1) 141 (24.9) 0.0003
Lesion size C50 mm, n(%) 346 (19.4) 204 (16.7) 142 (25.1) 0.0001
Bilateral lesions 825 (46.2) 496 (40.7) 329 (58.1) 0.0002
No. of resected segments, [median, (IQR)] 2 (0–6) 2 ±1.7 2.2 ±1.8 0.71
Major resection 696 (39) 464 (38.1) 232 (41) 0.24
Rehepatectomy 381 (21.4) 225 (18.5) 156 (27.6) 0.0001
Two-stage hepatectomy 291 (16.3) 171 (14) 120 (41) 0.001
Laparoscopy, yes, n(%) 87 (4.9) 73 (6) 14 (2.5) 0.001
Operative time [240, n(year) 849 (47.5) 525 (43.1) 324 (57.2) 0.0001
Pedicle clamping, yes, n(%) 1156 (64.8) 777 (63.8) 379 (67) 0.19
Pedicle clamping [40 min, n(%) 491 (27.5) 279 (22.9) 212 (37.5) 0.0002
Hepatic vein control, n(%) 526 (29.5) 383 (31.4) 143 (25.2) 0.007
Vascular resection, n(%) 43 (2.4) 27 (2.2) 16 (2.8) 0.43
Associated radiofrequency, n(%) 315 (17.7) 224 (18.4) 91 (16.1) 0.23
Transfusion, yes, n(%) 290 (16.3) 168 (13.8) 122 (21.6) 0.0001
World J Surg
123
considered major risk factors impacting disease-free sur-
vival (Table 4).
Disease-free and overall survival after propensity
score matching
There was a difference in terms of oncological results, with
1-, 3-, 5-year OS at 94, 81, and 70% in R0 and 92, 75, and
58% in R1, respectively (p=0.008) (Fig. 3). DFS was
different with 1-, 3-, 5-year survival at 64, 41, and 28% in
R0 versus 51, 28, and 18% in R1 (p=0.0002) (Fig. 4).
Margin width and overall and disease-free survival
Patients were divided into three groups (0, 1–5, [5 mm)
in order to assess the impact of margin width on overall and
disease-free survival. Differences were evidenced in terms
of 3- and 5-year OS overall survival in favor of [5mm
margins (86, 80%) as compared to 1–5 mm margins (83,
76%) and 0 mm margins (70, 58%) before (p=0.0001)
(Fig. 5). After PSM, differences were evidenced in terms
of 3- and 5-year OS in favor of [5 mm margins (83, 76%)
as compared to 1–5 mm margins (82, 75%) and 0 mm
margins (70, 62%) (p=0.02) (Fig. 6).
Discussion
To our knowledge, this study represents the first PSM-
based comparison for patients with same preoperative
characteristics (demographic data, primary and metastatic
cancer), allowing to understand which elements influence
the feasibility of R0 resection and how R1 resection
impacts overall and disease-free survival. In our popula-
tion, once preoperative characteristics are comparable, the
technical difficulties of hepatectomies (rehepatectomy,
long operative time and pedicle clamping, hepatic vein
control, and perioperative transfusion) represent major risk
factors to obtain an incomplete resection margin. However,
R1 resection still impacts overall and disease-free survival
even when negative prognostic are balanced with a PSM.
Finally, the achievement of at least 1 mm of free resection
margins is mandatory to obtain a significant improvement
recurrence-free and overall survival.
Several perioperative factors influence survival after
liver resection for colorectal liver metastasis. Even if the
width of resection margins is still a matter of debate, a
1 mm margin has been defined as the minimum margin to
obtain oncological results after hepatectomies [1,3–6]. A
wider resection margin ([10 mm) was initially designed
as the ideal distance due to the presence of microsatellites
[5] and recently evoked by a meta-analysis by Liu et al. [7],
but this distance could not be always achieved due to the
Fig. 1 Overall survival before PSM
Fig. 2 Disease-free survival before PSM
World J Surg
123
Table 2 Population analysis after PSM
All patients (n=1034) R0 resection (n=517) R1 resection (n=517) p
Period of resection, 2010–2013 536 (51.8) 272 (53) 264 (51) 0.61
Age, yr 60.2 ±10.2 60.4 ±11.4 59.2 ±10.5 0.31
Gender, male, n(%) 642 (62.1) 317 (61) 325 (63) 0.60
ASA score 3–4, n(%) 158 (15.3) 88 (17) 70 (14) 0.12
Comorbidity, yes, n(%) 366 (35.4) 189 (37) 177 (34) 0.43
BMI (kg/m
2
) [median, (IQR)] 25.3 ±3.0 25 ±3.5 25.3 ±4.3 0.71
Primary resected, yes, n(%) 878 (84.9) 447 (86) 431 (83) 0.16
Primary rectum, yes, n(%) 272 (26.3) 138 (27) 134 (26) 0.77
Primary N?, yes, n(%) 530 (51.3) 267 (52) 263 (51) 0.80
Primary neoadjuvant ctx, yes (%) 154 (14.9) 84 (16) 70 (14) 0.21
Disease-free interval [12 months, n(%) 273 (26.4) 126 (24) 147 (28) 0.13
Liver metastasis synchronous, yes, n(%) 465 (45) 232 (45) 233 (45) 0.95
Liver metastasis CEA (ng/mL) [median, (IQR)] 270 ±390 224 ±632 332 ±1082 0.19
Liver metastasis neoadjuvant yes, n(%) 692 (66.9) 350 (68) 342 (66) 0.59
Portal vein embolization 192 (18.6) 91 (18) 101 (20) 0.42
Single lesion, n(%) 282 (27.3) 141 (27) 141 (27) 0.99
Lesion size C50 mm, n(%) 246 (23.8) 125 (24) 121 (23) 0.77
Bilateral lesions 567 (54.8) 284 (55) 283 (55) 0.95
Number of resected segments, [median, (IQR)] 2 (0–6) 2.2 ±1.8 2.3 ±1.7 0.46
Major resection 422 (40.8) 212 (41) 210 (41) 0.89
Rehepatectomy 238 (23) 90 (17) 148 (29) 0.0001
Two-stage hepatectomy 195 (18.9) 85 (16) 110 (21) 0.05
Laparoscopy, yes, n(%) 45 (4.4) 32 (6) 13 (3) 0.004
Operative time [240, n(y) 547 (52.9) 255 (49) 292 (56) 0.02
Pedicle clamping, yes, n(%) 694 (67.1) 343 (66) 351 (68) 0.59
Pedicle clamping [40 min, n(%) 340 (32.9) 148 (29) 192 (37) 0.004
Hepatic vein control, n(%) 302 (29.2) 164 (32) 138 (27) 0.07
Vascular resection, n(%) 27 (2.6) 11 (2) 16 (3) 0.32
Associated radiofrequency, n(%) 191 (18.5) 100 (19) 91 (18) 0.47
Transfusion, yes, n(%) 191 (18.5) 85 (16) 106 (21) 0.09
Table 3 Univariate and multivariate model for overall survival after PSM
Univariate analysis Multivariate analysis
Non survivor pvalue OR (95% CI) p
Yes
R1 resection 117 0.01 1.38 (1.1–2.0) 0.05
Disease free interval [12 months 44 0.08
ASA score 3–4 22 0.04
Comorbidity 59 0.03
Primary rectal cancer 43 0.05
Primary N?124 0.002 1.6 (1.2–2.2) 0.004
Bilateral liver metastasis 134 0.001 1.6 (1.2–2.3) 0.003
Operative time [240 min 122 0.02
Pedicle clamping [40 min 91 0.002 1.7 (1.2–2.3) 0.001
World J Surg
123
aggressiveness of the disease. These results are confirmed
in our series, with an increased risk of mortality and
recurrence in case of even more narrow margins even after
propensity score matching. Sadot et al. [3] hypothetized
that the necessity of a submillimeter margin was due to a
surrogate of the tumor’s biological behavior rather than a
surgical technique, demonstrating that the same surgical
technique could achieve different histological margins, and
concluding that liver resection could not be precluded to
patients who were suspected to undergo a resection with
narrow margins.
Table 4 Univariate and multivariate model for recurrence after PSM
Univariate analysis Multivariate analysis
Recurrence pvalue OR (95% CI) p
Yes
R1 resection 327 0.00002 1.5 (1.2–2.1) 0.0002
Disease free interval [12 months 143 0.04
Liver metastasis synchronous 286 0.02
Comorbidity 223 0.10
Liver metastasis neoadjuvant 317 0.0002 1.7 (1.2–2.2) 0.002
Primary N?, 327 0.006 1.3 (1.1–1.7) 0.01
Bilateral lesion 346 0.01
Rehepatectomy 149 0.07
Pedicle clamping [40 min 211 0.04
Associated RFA 127 0.006
More than one lesion 146 0.02
Fig. 3 Overall survival after PSM
Fig. 4 Disease-free survival after PSM
World J Surg
123
Starting from this consideration, we tried to compare
patients with a similar biological behavior, in order to
comprehend which factors influenced the performance of
an incomplete resection.
We decided to perform a PSM in order to reduce at a
minimum the confounding elements due to patient selec-
tion and oncological data. Age, gender, ASA score,
comorbidity, and BMI were compared, in order to obtain
comparable oncological data. The Fong [8] prognostic
criteria were chosen to achieve comparable populations in
terms of oncological results, mainly focusing on tumor
biology such as metastatic burden (size, number, and
synchronicity) and nodes status of primary tumor. In the
same way, perioperative chemotherapy was chosen as
parameter for matching, considering that the effect of
chemotherapy was established by Nordlinger et al. [9].
The results of our study evidenced the strong relation-
ship between difficult hepatectomies and incomplete
resection margins. As previously demonstrated, the
necessity to perform a major hepatectomy extended
resection, in case of large, bilateral, and centrally located
tumors, increase the risk to obtain a R1 resection [10–12].
As confirmed by Elias et al. [6], patients requiring complex
hepatectomies were more exposed to incomplete resection,
due to the complexity of achieving a radical clearance.
Another subject of debate is the comparison between
anatomical and non-anatomical resections. In the era of an
even more increasing parenchymal-sparing attitude, this
represents a major key point. Initially considered a major
risk factor for R1 resection [13], with a lower rate of
positive margins (2 vs. 16%) [12], non-anatomical resec-
tion is now considered equivalent to anatomical resections
[14,15], and the primary goal has been changed to obtain a
resection with clear margins, independent of a surgical
technique. Considering this, even in the current literature,
we consider laparoscopic resection to be safe and feasible
[16–18] in terms of resection margin and survival. In our
series, the role of laparoscopic surgery still remains unclear
due to the lower frequencies of laparoscopic resection in
R1 resection, probably due to the less frequent use of the
minimally invasive access in case of R1 resection high
risks.
As demonstrated by our study, R1 resection still repre-
sents a risk factor impacting overall and disease-free sur-
vival even once the recurrence risk factor has been
balanced. As previously evidenced by Welsh et al. [10], R1
resection represents a significant disadvantage in terms of
survival for patients. In our series, these data are confirmed,
considering that positive nodes in colorectal cancer resec-
tion, the presence of bilateral liver metastasis, and long
pedicle clamping impact survival negatively. There is a
necessity to perform neoadjuvant chemotherapy due to
primary tumor load which impacts disease-free survival.
Considering the high possibility of relapse, repeat hep-
atectomy currently represents the best treatment for
recurrence [19], even though it is still recognized to be a
risk factor of incomplete resection [10] due to the necessity
of performing parenchymal-sparing resection in order to
preserve the reduced quantity of hepatic parenchyma.
In conclusion, even if retrospectivity and multicentricity
represent major biases, this study confirms that difficult
Fig. 5 Overall survival by margin width after PSM
Fig. 6 Overall survival by margin width after PSM
World J Surg
123
hepatectomies represent a major factor which impacts
recurrence and patient survival. A prevision of ‘‘difficult
hepatectomies’’ could well represent a factor to analyze in
a multidisciplinary meeting, in order to improve the pos-
sibility to obtain margin-free hepatectomies. Even if cur-
rently everything is done to achieve R0 resection, margins
strongly impact overall and disease-free survival even
when ‘‘tumor load’’ is well balanced by a PSM, underlining
how improvements in surgical strategy and oncomedical
patient management are necessary. Neoadjuvant therapy
[20,21] and innovative surgical approaches [22,23] could
well represent a benefit to achieve correct resection
margins.
Acknowledgements The authors would like to thank all participating
centers for their contribution to this study.
Author contributions Amiens, CHU Amiens Picardie: Cyril Cosse,
Delphine Lignier, Jean Marc Regimbeau; Angers, CHU Angers:
Julien Barbieux, Emilie Lermite, Antoine Hamy; Beauvais, CH
Beauvais: Franc¸ois Mauvais; Bordeaux, Groupe Hospitalier Saint
Andre
´: Christophe Laurent; Chambery, CH Chambery: Irchid Al
Naasan; Cre
´teil, CHU Henri Mondor: Alexis Laurent, Daniel Azou-
lay, Philippe Compagnon, Nicola de’Angelis; Eaubonne, Ho
ˆpital
Simone Veil: Mohammed Sbai Idrissi; Epinal, Polyclinique de la
Ligne Bleue: Fre
´deric Martin; Gap, CH des Alpes du Sud: Jero
ˆme
Atger; Lyon, Ho
ˆpital de la Croix Rousse: Jacques Baulieux, Benjamin
Darnis, Jean Yves; Mabrut; Lyon, Ho
ˆpital Edouart-Herriot: Vahan
Kepenekian, Julie Perinel, Mustapha Adham; Lyon, CH Lyon Sud:
Olivier Glehen; Lyon, Centre Le
´on Be
´rard: Michel Rivoire; Mar-
seille, Ho
ˆpital de la Conception: Jean Hardwigsen, Anais Palen,
Emilie Gre
´goire, Yves Patrice LeTreut; Marseille, Institut Paoli-
Calmettes: Jean Robert Delpero, Olivier Turrini; Montpellier, Ho
ˆpital
Saint Eloi: Astrid Herrero, Francis Navarro, Fabrizio Panaro; Nancy,
CHU Brabois: Ahmet Ayav, Laurent Bresler; Nancy, Institut de
Cance
´rologie de Lorraine Alexis-Vautrin: Philippe Rauch, Franc¸ois;
Guillemin, Fre
´deric Marchal; Nice, Ho
ˆpital de l’Archet: Jean
Gugenheim, Antonio Iannelli; Kremlin Bice
ˆtre, CHU Kremlin Bice
ˆ-
tre: Stephane Benoist, Antoine Brouquet; Paris, Ho
ˆpital Lariboisie
`re:
Marc Pocard, Re
´a Lo Dico; Paris, Institut Mutualiste Montsouris:
Brice Gayet, David Fuks; Paris, Ho
ˆpital Saint Antoine: Olivier.
Scatton, Olivier Soubrane; Paris, Ho
ˆpital de la Pitie
´Salpetrie
`re: Jean-
Christophe Vaillant; Reims, Ho
ˆpital Robert Debre
´: Tullio Piardi,
Daniel Sommacale, Reza Kianmanesh; La Roche-sur-Yon, Centre
de
´partemental de Vende
´e: Michel Comy; Strasbourg, Ho
ˆpital Hau-
tepierre: Philippe Bachellier, Elie Oussoultzoglou, Pietro Addeo;
Strasbourg, Nouvel Ho
ˆpital Civil: Dimitrios Ntourakis, Patrick Pes-
saux, Didier Mutter, Jacques Marescaux; Toulouse, Ho
ˆpital Rangueil:
Loı
¨c Raoux, Bertrand Suc, Fabrice Muscari; Troyes, Ho
ˆpital des
Hauts-Clos: Georges ELHOMSY; Villejuif, Ho
ˆpital Paul Brousse:
Maximiliano Gelli, Antonio Sa Cunha, Rene
´Adam, Denis Castaing,
Daniel Cherqui; Gabriella Pittau, Oriana Ciacio, Eric Vibert; Ville-
juif, Gustave Roussy: Dominique Elias, Diane Goe
´re, Fabrizio
Vittadello.
Compliance with ethical standards
Conflict of interest All authors declare that they have no conflict of
interest.
References
1. Hamady ZZR, Lodge JPA, Welsh FK et al (2014) One-millimeter
cancer-free margin is curative for colorectal liver metastases. Ann
Surg 259:543–548. doi:10.1097/SLA.0b013e3182902b6e
2. Strasberg SM, Phillips C (2013) Use and dissemination of the
brisbane 2000 nomenclature of liver anatomy and resections. Ann
Surg 257:377–382. doi:10.1097/SLA.0b013e31825a01f6
3. Sadot E, Koerkamp BG, Leal JN et al (2015) Resection margin
and survival in 2368 patients undergoing hepatic resection for
metastatic colorectal cancer. Ann Surg 262:476–485. doi:10.
1097/SLA.0000000000001427
4. Pandanaboyana S, White A, Pathak S et al (2015) Impact of
margin status and neoadjuvant chemotherapy on survival, recur-
rence after liver resection for colorectal liver metastasis. Ann
Surg Oncol 22:173–179. doi:10.1245/s10434-014-3953-6
5. Nuzzo G, Giuliante F, Ardito F et al (2008) Influence of surgical
margin on type of recurrence after liver resection for colorectal
metastases: a single-center experience. Surgery 143:384–393.
doi:10.1016/j.surg.2007.09.038
6. Elias D, Cavalcanti A, Sabourin JC et al (1998) Results of 136
curative hepatectomies with a safety margin of less than 10 mm
for colorectal metastases. J Surg Oncol 69:88–93
7. Liu W, Sun Y, Zhang L, Xing B (2015) Negative surgical margin
improved long-term survival of colorectal cancer liver metastases
after hepatic resection: a systematic review and meta-analysis. Int
J Colorectal Dis. doi:10.1007/s00384-015-2323-6
8. Fong Y, Fortner J, Sun RL et al (1999) Clinical score for pre-
dicting recurrence after hepatic resection for metastatic colorectal
cancer: analysis of 1001 consecutive cases. Ann Surg
230:309–321. doi:10.1097/00000658-199909000-00004
9. Nordlinger B, Sorbye H, Glimelius B et al (2013) Perioperative
FOLFOX4 chemotherapy and surgery versus surgery alone for
resectable liver metastases from colorectal cancer (EORTC
40983): long-term results of a randomised, controlled, phase 3
trial. Lancet Oncol 14:1208–1215. doi:10.1016/S1470-
2045(13)70447-9
10. Welsh FKS, Tekkis PP, O’Rourke T et al (2008) Quantification of
risk of a positive (R1) resection margin following hepatic
resection for metastatic colorectal cancer: an aid to clinical
decision-making. Surg Oncol 17:3–13. doi:10.1016/j.suronc.
2007.12.003
11. Poultsides GA, Schulick RD, Pawlik TM (2010) Hepatic resec-
tion for colorectal metastases: the impact of surgical margin
status on outcome. HPB (Oxf) 12:43–49. doi:10.1111/j.1477-
2574.2009.00121.x
12. Dematteo RP, Palese C, Jarnagin WR et al (2000) Anatomic
segmental hepatic resection is superior to wedge resection as an
oncologic operation for colorectal liver metastases. J Gastrointest
Surg 4:178–184
13. Wray CJ, Lowy AM, Mathews JB et al (2005) The significance
and clinical factors associated with a subcentimeter resection of
colorectal liver metastases. Ann Surg Oncol 12:374–380. doi:10.
1245/ASO.2005.06.038
14. Zorzi D, Mullen JT, Abdalla EK et al (2006) Comparison
between hepatic wedge resection and anatomic resection for
colorectal liver metastases. J Gastrointest Surg 10:86–94. doi:10.
1016/j.gassur.2005.07.022
15. Puglisi F, Capuano P, Iambrenghi OC et al (2009) Laparoscopic
repair of Morgagni hernia in an adult: use of a porcine small
intestine submucosa biocompatible prosthesis. Chir Ital
61:351–356
16. de’Angelis N, Memeo R, Calderaro J et al (2014) Open and
laparoscopic resection of hepatocellular adenoma: trends over
World J Surg
123
23 years at a specialist hepatobiliary unit. HPB (Oxf)
16:783–788. doi:10.1111/hpb.12257
17. Memeo R, de’Angelis N, Compagnon P et al (2014) Laparo-
scopic vs. open liver resection for hepatocellular carcinoma of
cirrhotic liver: a case-control study. World J Surg 38:2919–2926.
doi:10.1007/s00268-014-2659-z
18. Salloum C, Subar D, Memeo R et al (2013) Laparoscopic robotic
liver surgery: the Henri Mondor initial experience of 20 cases.
J Robot Surg. doi:10.1007/s11701-013-0437-9
19. Ali MA, Di Sandro S, Lauterio A et al (2015) Repeat hepatec-
tomy for recurrent colorectal liver metastases: is it worth the
challenge? J Gastrointest Surg 19:2192–2198. doi:10.1007/
s11605-015-2939-4
20. Sjoquist KM, Burmeister BH, Smithers BM et al (2011) Survival
after neoadjuvant chemotherapy or chemoradiotherapy for
resectable oesophageal carcinoma: an updated meta-analysis.
Lancet Oncol 12:681–692. doi:10.1016/S1470-2045(11)70142-5
21. Markar SR, Bodnar A, Rosales J et al (2013) The impact of
neoadjuvant chemoradiotherapy on perioperative outcomes,
tumor pathology, and survival in clinical stage II and III eso-
phageal cancer. Ann Surg Oncol 20:3935–3941. doi:10.1245/
s10434-013-3137-9
22. Memeo R, de’Angelis N, de Blasi V, Cherkaoui Z, Brunetti O,
Longo V, Piardi T, Sommacale D, Marescaux J, Mutter D, Pes-
saux P (2016) Innovative surgical approaches for hepatocellular
carcinoma. World J Hepatol 8:591–596. doi:10.5662/wjm.v1.i1.4
23. Ntourakis D, Memeo R, Soler L et al (2015) Augmented reality
guidance for the resection of missing colorectal liver metastases:
an initial experience. World J Surg. doi:10.1007/s00268-015-
3229-8
World J Surg
123