PAPER OF THE 21ST ANNUAL ESA MEETING

Postoperative Peritoneal Infection Enhances Migration and

Invasion Capacities of Tumor Cells In Vitro
An Insight Into the Association Between Anastomotic Leak and Recurrence After
Surgery for Colorectal Cancer
Silvia Salvans, MD,∗ † Xavier Mayol, PhD,† Sandra Alonso, MD, PhD,∗ † Ramon Messeguer, PhD,‡
Marta Pascual, MD, PhD,∗ † Sergi Mojal, BSc,§ Luis Grande, MD, PhD,∗ † and Miguel Pera, MD, PhD∗ †

option with curative intention is surgical excision. After complete

Objective: The aim of this study was to investigate the effect of postoperative
resection, recurrence occurs in up to 30% to 40% and the most im-
peritoneal infection on proliferation, migration, and invasion capacities of
portant prognostic factor is tumor stage.2 However, the study of other
cancer cells lines in vitro after surgery for colorectal cancer.
factors that may impinge on the oncological outcome such as those
Background: Anastomotic leakage is associated with higher rates of recur-
related to surgery can provide knowledge about the mechanisms of
rence after surgery for colorectal cancer. However, the mechanisms respon-
recurrence.
sible are unknown. We hypothesized that the infection-induced inflammatory
Several clinical studies have shown that anastomotic leakage
response may enhance tumor progression features of residual cancer cells.
is an independent risk factor of tumor recurrence and is associated
Methods: Prospective matched cohort study. Patients undergoing surgery for
with increased cancer-specific mortality.3–8 Nevertheless, the mech-
colorectal cancer with curative intent (January 2008—March 2012) were in-
anisms responsible for the association between postoperative peri-
cluded. Patients who had an anastomotic leak or intra-abdominal abscess were
toneal infection and tumor recurrence remain to be elucidated. It
included in the infection group (n = 47). For each case patient, another patient
is well known that a systemic inflammatory response is associated
with an uncomplicated postoperative course was selected for the control group
with poor outcome in patients with colorectal cancer after curative
(n = 47).
resection.8–10 Soluble factors released by the inflammatory response
In vitro treatments on cancer cell lines (MDA-MB-231 and SW620) were
might thus stimulate residual viable cancer cells present in the sur-
performed using baseline and postoperative serum and peritoneal fluid samples
gical field, venous blood, or occult micrometastases. Although there
to determine cell proliferation and cell migration/invasion activities.
is extensive literature on the role of inflammatory cytokines in tumor
Results: Postoperative peritoneal fluid from infected patients enhanced both
progression, no particular mechanisms have yet been validated to ex-
cell migration (infection: 140 ± 85 vs control: 94 ± 30; P = 0.016) and cell
plain the increased tumor recurrence after postoperative peritoneal
invasion (infection: 117 ± 31 vs control: 103 ± 16; P = 0.024) capacities of
infection.
cancer cell lines. With serum samples, these effects were only observed in cell
Our group has been engaged in the study of the inflamma-
migration assays (infection: 98 ± 28 vs control: 87 ± 17; P = 0.005). Some
tory response to postoperative infection as a factor associated with
minor activation of cell proliferation was observed by treatment with serum
tumor recurrence. We found that the angiogenic cytokine vascular
from infection group. Two-year cumulative disease-free survival was signif-
endothelial growth factor was significantly induced together with the
icantly lower in patients with postoperative peritoneal infection (infection:
inflammatory response and similar results were observed in a mouse
77.6% vs control: 90.6%; P = 0.032).
model of intraperitoneal infection after colon cancer surgery.11,12
Conclusions: Our results suggest that postoperative peritoneal infection en-
Most strikingly, these elevated levels of vascular endothelial growth
hances the invasive capacity of residual tumor cells after surgery, thus facili-
factor significantly correlated with the rates of tumor recurrence.
tating their growth to recurrent tumors.
Apart from an increase in angiogenesis during the postoper-
Keywords: colorectal cancer, invasion, peritoneal infection, surgery, tumor ative period, other factors related to tumor growth and metastasis
recurrence development may be involved in the association between peritoneal
infection and tumor recurrence. We hypothesized that the infection-
(Ann Surg 2014;260:939–944)
induced inflammatory response may enhance tumor progression fea-
tures of residual cancer cells. The aim was to investigate the effect
C olorectal cancer is currently the third most common cancer
worldwide and the fourth leading cause of cancer-related death.1
In the absence of tumor dissemination, the only current therapeutic
of postoperative peritoneal infection on proliferation, migration, and
invasion capacities of tumor cells in vitro, after surgery for colorectal
cancer.

From the ∗ Section of Colon and Rectal Surgery, Department of Surgery, Hospital
del Mar, Barcelona, Spain; †Colorectal Cancer Research Group, Hospital del
Mar Medical Research Institute (IMIM), Barcelona, Spain; ‡Biomed Division,
Leitat Technological Center, Barcelona, Spain; and §Consulting Service on
Methodology for Biomedical Research, Hospital del Mar Medical Research
Institute (IMIM), Barcelona, Spain.
Disclosure: Supported by the Instituto de Salud Carlos III grant PI11/01093.
The authors declare no conflict of interests.
Reprints: Miguel Pera, MD, PhD, Section of Colon and Rectal Surgery, Department
of Surgery, Hospital del Mar, Passeig Marı́tim 25-29, 08003 Barcelona, Spain.
E-mail: mpera@hospitaldelmar.cat.
Copyright C 2014 by Lippincott Williams & Wilkins
ISSN: 0003-4932/14/26005-0939
DOI: 10.1097/SLA.0000000000000958
METHODS
Study Design and Participants
This was a prospective matched cohort study. Patients under-
going surgery for colorectal cancer with curative intent from January
2008 to March 2012 were included. Patients who had an anastomotic
leak or intra-abdominal abscess were included in the infection group.
For each case patient, another patient with an uncomplicated post-
operative course was selected for the control group. Controls were
matched for six criteria: sex, age (±5 years), tumor location, surgical
approach, date of operation (± 3 years), and tumor stage, according to
the TNM classification. Patients with distant metastases and previous

Annals of Surgery r Volume 260, Number 5, November 2014 www.annalsofsurgery.com | 939

Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Salvans et al
malignant tumor and those requiring an emergency operation were
excluded. The same surgical team performed all operations and la-
paroscopic surgery was performed in selected cases. A closed-suction
drain was left at the resection site at the end of the operation.
Data collected included carcinoembryonic antigen levels, op-
erative time, perioperative transfusion, treatment of anastomotic
leak/abscess, the number of lymph nodes isolated, tumor stage, length
of hospital stay, and neoadjuvant and adjuvant treatment. Oncologic
outcome data were also prospectively collected.
Patients were considered as having an anastomotic leak when
clinical suspicion was confirmed with radiologic or operative explo-
ration. Intra-abdominal abscess was defined as a postoperative intra-
abdominal fluid collection with fever or leukocytosis that required
treatment with intravenous antibiotics and drainage.
The study was approved by the hospital ethics committee and
all patients provided written informed consent.
Blood and Peritoneal Fluid Sampling
Ten milliliters of venous blood were collected before surgery.
Additional samples were obtained 4 days after surgery or at the mo-
ment an intraperitoneal infection was detected. To obtain baseline
peritoneal fluid, peritoneal lavage with 100 mL of saline was done
immediately after laparotomy or after creating the pneumoperitoneum
so that a sample of 10 mL was obtained. Thereafter, 10 mL of peri-
toneal fluid was also obtained from the drain on postoperative day 4
surgery or at the time peritoneal infection was diagnosed. All the sam-
ples, with blood being allowed to clot first, were centrifuged at 3000
rpm for 10 minutes at 4◦ C. The resulting liquid phase was isolated,
aliquoted, and stored at −80◦ C.
In Vitro Assays
Cell Lines and Cell Culture
We used human SW620 colon cancer cells and MDA-MB-231
breast cancer cells as cellular models because they showed optimal
performance in invasion and migration assays, respectively (see later).
Cell cultures were maintained by serial passage in Dulbecco’s Modi-
fied Eagle’s medium supplemented with 10% fetal bovine serum and
antibiotics (penicillin 100 U/mL and streptomycin 100 mg/mL) at
37◦ C in an atmosphere of 5% carbon dioxide.
Cell Proliferation Assay
The proliferation assay was performed according to the col-
orimetric method developed by Landegren13 and modified by Givens
et al.14 It is based on the activity of the lysosomal hexosaminidase
enzyme as a measure of the number of viable cells. Briefly, cells were
plated in 96-well plates in duplicates. The next day, cells were treated
with serum or peritoneal fluid samples diluted to 10% in Iscove’s Mod-
ified Dulbecco’s Medium culture medium supplemented with Insulin,
Transferrin, and Selenium solution (Life Technologies, Paisley, UK)
and 10% of fetal bovine serum. The day after, the hexosaminidase
substrate reagent14 was added (50 μL/well) and the plates were incu-
bated at 37◦ C for 4 hours. The reaction was stopped and the color was
developed by addition of 80 μl/well of hexosaminidase developer.14
Optical density was measured at 410 nm with a microplate spec-
trophotometer. Results were expressed as arbitrary units relative to
internal controls in each assay, the internal control being the preop-
erative sample of each noninfected patient pair.
Cell Migration and Invasion Assays
Before each assay, cells were routinely starved overnight in
Iscove’s Modified Dulbecco’s Medium supplemented with insulin,
transferrin, and selenium solution but without fetal bovine serum.
MDA-MB-231 cells were used for cell migration assays and SW620
940 | www.annalsofsurgery.com
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Annals of Surgery r Volume 260, Number 5, November 2014
for cell invasion assays. Starved cells were disaggregated with
Accutase (Life Technologies, Paisley, UK) and seeded on 96-transwell
permeable supports with 8-μm pore size (Corning Life Sciences,
Corning, NY) using the same starving medium. In the case of cell
invasion assays, transwells were covered with 30 μL of a Basement
Membrane Extract (Trevigen; Gaithersburg, MD) solution poured the
day before seeding. In parallel, serum and peritoneal liquid samples
were diluted to 10% with Iscove’s Modified Dulbecco’s Medium con-
taining insulin, transferrin and selenium solution and 10% fetal bovine
serum, and split throughout 96-well receiver plates above which per-
meable supports containing the cells were placed. This transwell sys-
tem was incubated at 37◦ C in an atmosphere of 5% carbon dioxide for
24 hours. The relative number of cells migrating or invading through
basement membrane extract to the lower side of the transwells was
estimated by using the hexosaminidase assay, mentioned previously.
Statistical Analysis
Continuous variables were expressed as mean ± SD and cate-
gorical data as frequencies and percentages. The following tests were
used: χ 2 test for categorical variables and nonparametric Wilcoxon
test for continuous variables. Bivariate analyses were performed by
employing the Cox proportional hazards model to examine the associ-
ation between clinicopathological variables, in vitro assays results and
peritoneal infection and estimate the independent prognostic effect
of each variable on disease-free survival. Hazard ratios are reported
with 95% confidence interval. Cumulative survival was plotted by
the Kaplan-Meier method, with statistical analysis by means of the
log-rank test. A value of P < 0.05 was used to determine the level
of statistical significance. Statistical analyses were performed with
SPSS 18.0 (SPSS Inc, Chicago, IL).
RESULTS
Patients Characteristics
During the study interval, 420 patients were diagnosed with
nonmetastatic colorectal cancer at this institution and considered el-
igible for the study. A total of 47 patients were included in the in-
fection group (anastomotic leak [n = 34] or intra-abdominal abscess
[n = 13]). Patient characteristics, operative results, and pathological
data of patients and the matched pairs are shown in Table 1. The rate
of patients with American Society of Anesthesiologists status 3 or 4
was higher in the peritoneal infection group although the difference
did not reach statistical significance. Reoperation in the infection
group was necessary in 25 patients, whereas the other 22 patients
were successfully treated with antibiotics and percutaneous drainage.
As expected, length of stay was significantly longer in patients with
peritoneal infection. Three patients with anastomotic dehiscence and
septic shock died.
Surgical resection was R0 in all patients and the majority of
patients were stage II and stage III. There were no differences in
the degree of differentiation or lymphovascular or perineural in-
vasion between both groups. There were no significant differences
between groups in the percentage of patients who received adjuvant
chemotherapy (control group: 21 (44%) vs infection group: 15 (31%);
P = 0.134).
Cell Proliferation Assay and Toxicity
Postoperative serum samples from patients with peritoneal in-
fection significantly increased proliferation in the MDA-MB-231 cell
line compared with controls (control group: 99.42 vs infection group:
101.41; P = 0.013). This effect was most apparent when comparing
the increase in cell proliferation exerted by postoperative samples
respect to preoperative samples within each group (Table 2).

C 2014 Lippincott Williams & Wilkins
Annals of Surgery r Volume 260, Number 5, November 2014 Peritoneal Infection and Tumor Recurrence

considered an inhibition of cell growth. When comparing infection

TABLE 1. Clinical Characteristics of Patients, Operative
group with control group, no relevant differences were observed ei-
Results, and Pathological Data
ther between preoperative samples or between postoperative samples
Control Group Infection Group (data not shown).
Variables (n = 47)∗ (n = 47)∗ P
Age, yr 72 ± 9 74 ± 9 0.404 Cell Migration and Invasion Assays
Sex, male 33 (70) 33 (70) 1.000 With regard to serum samples, cell migration assays displayed
Tumor location significant results: postoperative samples from patients with peri-
right colon 18 (38) 18 (38) 1.000 toneal infection increased cell migration activity compared with con-
left colon 15 (32) 15 (32)
trols (control group: 87 ± 17 vs infection group: 98 ± 28; P = 0.005).
rectum 14 (30) 14 (30)
CEA, ng/mL 10 ± 11 11 ± 33 0.893 In fact, no induction of cell migration or invasion was detected after
ASA status surgery in controls. A significant increase in the infection group with
2 19 (40.4) 10 (21.4) 0.195 respect to the control group was also observed when comparing the
3 26 (55.3) 32 (68) difference in cell migration exerted by preoperative and postopera-
4 2 (4.3) 5 (10.6) tive samples (Table 3). Cell invasion assays using sera did not reveal
Operative time, min 194 ± 78 204 ± 89 0.734 significant differences between groups (Table 3).
Laparoscopy 10 (21) 10 (21) 1.000 Postoperative peritoneal fluids taken from the infection group
Blood transfusion 5 (11) 6 (13) 1.000 rendered a significant increase in the capacity of both in vitro cell
Reoperation 0 (0) 25 (53) 0.001
migration and cell invasion compared to the effect of peritoneal fluids
Hospital stay, d 7±8 25 ± 11 0.001
No. of lymph nodes isolated 23 ± 15 22 ± 14 0.559 from the control group (cell migration, control group: 94 ± 30 vs
TNM stage infection group: 140 ± 85, P = 0.016; cell invasion, control group:
Stage I 7 (15) 7 (15) 1.000 103 ± 16 vs infection group: 117± 31, P = 0.024). This effect was
Stage II 20 (42.5) 20 (42.5) also apparent when comparing the difference in cell migration and
Stage III 20 (42.5) 20 (42.5) invasion activities between preoperative and postoperative peritoneal
Degree of differentiation liquids in both study groups (Table 3).
Low grade 35 (74) 33 (70) 0.646
High grade 12 (26) 14 (30)
Lymphovascular invasion 20 (42) 15 (32) 0.289 Recurrence
Perineural invasion 15 (32) 11 (23) 0.358 The median follow-up for survivors was 38 months (25-55
∗ months). During this period, 18 patients developed tumor recurrence
Data are the number of patients with percentages in parentheses or mean ± SD.
ASA indicates American Society of Anesthesiologists; CEA, carcinoembryonic that was more frequent in patients who had a peritoneal infection
antigen. (control group: 6 [12.5%] vs infection group: 12 [25.5%]; P = 0.001)

TABLE 2. Paired Samples Statistics (Wilcoxon Test) TABLE 3. Paired Samples Statistics (Wilcoxon Test)
Comparing Differences Between Preoperative and Comparing Differences Between Preoperative and
Postoperative Serum Samples From Cell Proliferation Postoperative Samples From Migration/Invasion Assays
Assays Using Both Serum and Peritoneal Fluid Samples
Relative Relative
Sample Group Cell No∗ P Cell Migration/
Sample Group Invasion∗ P
MDA-MB-231
Difference (day 0 to day 4) Control − 0.30 ± 5.01 0.010 Sera
Difference (day 0 to day Infection 2.71 ± 5.53 Cell migration
of infection) Difference (day 0 to day 4) Control − 11.27 ± 16.98 <0.005
SW620 Difference (day 0 to day Infection 13.44 ± 29.77
Difference (day 0 to day 4) Control − 0.04 ± 5.56 0.216 of infection)
Difference (day 0 to day Infection 1.16 ± 4.89 Cell Invasion
of infection) Difference (day 0 to day 4) Control − 9.18 ± 34.08 0.861
∗
Difference (day 0 to day Infection − 7.82 ± 23.96
Relative cell numbers were expressed as the percentages of the cell proliferation of infection)
assay values relative to the preoperative control value in each pair of samples.
Peritoneal fluids†
Cell migration
Difference (day 0 to day 4) Control − 5.65 ± 30.50 0.008
Difference (day 0 to day Infection 55.59 ± 83.21
Some postoperative peritoneal liquid samples exerted a cyto- of infection)
toxic effect on cell cultures that was not related to the presence of in- Cell invasion
fection (data not shown), so the peritoneal fluid analysis was restricted Difference (day 0 to day 4) Control 2.81 ± 16.16 <0.005
to the subset of noncytotoxic samples. Postoperative peritoneal fluid Difference (day 0 to day Infection 23.01 ± 23.36
samples rendered lower levels of cell proliferation when compared of infection)
with their respective preoperative samples, both in the control group ∗
Relative cell numbers were expressed as the percentages of the cell migration/
(MDA-MB-231 cell line; preoperative: 100 vs postoperative: 94.10; invasion assay values relative to the preoperative control value in each pair of
P < 0.005) and in the infection group (MDA-MB-231 cell line; pre- samples.
operative: 100.7 vs postoperative: 93.4; P < 0.016). However, cell †After excluding samples with a cytotoxic effect on cell cultures, 19 complete
matched pairs were available for analysis.
proliferation was diminished by approximately 5%, which cannot be


C 2014 Lippincott Williams & Wilkins www.annalsofsurgery.com | 941

Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Salvans et al Annals of Surgery r Volume 260, Number 5, November 2014

due to a higher percentage of systemic recurrence. There was no

tumor recurrence in patients in stage I.
The relationship between clinical variables, results of in vitro
assays, peritoneal infection, and disease-free survival are shown in
Table 4. On bivariate analysis, migration activity induced by postop-
erative peritoneal fluid samples, the difference between postoperative
and preoperative proliferation activity induced by peritoneal fluid,
the number of lymph nodes, and peritoneal infection were all sig-
nificantly associated with disease-free survival (Table 4). Two-year
cumulative disease-free survival was significantly lower in patients
with postoperative peritoneal infection (infection: 77.6% vs control:
90.6%; P = 0.032) (Fig).

DISCUSSION
In this study, we have shown that both serum samples and
peritoneal fluid harvested from patients undergoing surgery for col- FIGURE. Kaplan-Meier analysis of disease-free survival in pa-
orectal cancer, complicated with peritoneal infection, significantly tients with or without peritoneal infection after surgery for
enhanced the in vitro invasiveness capacity of cancer cell lines. The colorectal cancer.
acquisition of an invasive phenotype is associated with tumor dis-
semination, tumor cell survival, and the cancer stem-cell phenotype,
and all of them are hallmark features related to tumor progression
association,18,19 a recent meta-analysis concluded that anastomotic
and tumor recurrence.15 Therefore, our results are consistent with
leakage has a negative impact on the oncological outcome.20
the hypothesis that the inflammatory response to infection releases
There is evidence that the inflammatory response in pa-
soluble factors that may impinge on the phenotype of residual tumor
tients with postoperative peritoneal infection affects the oncologi-
cells, in particular their invasive capacity.
cal outcome.8–10 Katoh et al8 found that anastomotic leakage was
The finding that tumor recurrence was significantly higher
associated with a higher incidence of hematogenous metastases in
in patients who had a postoperative peritoneal infection has already
patients with stage II colorectal cancer. The authors also observed
been well established by previous reports.3–8 This association has also
that a sustained increase of the serum C-reactive protein at 2 weeks
been reported after resection of liver metastases16 and other gastroin-
was an independent predictor of poor prognosis. Our previous studies
testinal malignancies.17 Although some authors have questioned this
suggested that the inflammatory reaction, in response to infection,
leads to an increased expression of local and circulating proinflam-
matory and proangiogenic factors that might facilitate the survival
and growth of residual tumor cells in their path to recurrence.11 In
TABLE 4. Relationships Between Clinical Variables, In particular, postoperative serum and peritoneal fluid interleukin-6 and
Vitro Assays Results, Tumor Pathology, Peritoneal vascular endothelial growth factor concentrations were significantly
Infection, and Disease-free Survival associated with disease-free survival in patients with peritoneal in-
fection. Although the postoperative inflammatory response and the
Variable HR (95% CI) P
activation of angiogenesis should be considered part of the wound
Sex, male 0.570 (0.187−1.732) 0.321 healing process, magnification of these responses could have a nega-
Age 0,996 (0.948−1.016) 0.858 tive effect on patients with cancer.
CEA 0.997 (0.980−1.017) 0.784 Our present results point to an additional mechanism by which
Serum proliferation assay∗ 0.978 (0.886−1.081) 0.667 postoperative infection might facilitate tumor recurrence: the inflam-
Peritoneal fluid proliferation assay∗ 0.996 (0.985−1.007) 0.491 matory environment created in response to infection seemed to act
Serum migration assay∗ 1.003 (0.982−1.025) 0.792
Peritoneal fluid migration assay∗ 1.016 (1.000−1.032) 0.044
directly on tumor cells, enhancing their capacity of tissue invasive-
Serum invasion assay∗ 0.984 (0.963−1.007) 0.167 ness, as inferred from the in vitro migration/invasion assays. Thus, the
Peritoneal fluid invasion assay∗ 1.024 (0.955−1.055) 0.104 acquisition of an invasive/metastatic phenotype is a cellular feature
Serum proliferation assay† 0.948 (0.839−1.070) 0.386 that functionally links postsurgical inflammation with tumor recur-
Peritoneal fluid proliferation assay† 0.986 (0.973−0.999) 0.041 rence. Moreover, the serum samples from patients with peritoneal
Serum migration assay † 1.010 (0.992−1.029) 0.285 infection also displayed the ability to enhance the proliferation of
Peritoneal fluid migration assay† 1.017 (1.001−1.033) 0.034 cancer cell lines in vitro compared to controls, an effect that may
Serum invasion assay† 1.001 (0.973−1.029) 0.955 be interpreted as a growth advantage of residual tumor cells in the
Peritoneal fluid invasion assay† 1.016 (0.985−1.047) 0.317 presence of infection. It is important to note that some of the in vitro
Lymph node involvement 1.298 (0.509−3.312) 0.585
No. of lymph nodes 0.932 (0.870−0.998) 0.044
results were significantly associated, along with peritoneal infection,
Degree of differentiation 2.110 (0.816−5.459) 0,124 with disease-free survival. Therefore, enhancing tumor invasiveness
Lymphovascular invasion 2.146 (0.849−5.421) 0.106 in residual cancer cells present in the surgical field, venous blood or
Perineural invasion 1.850 (0.715−4.789) 0.205 occult micrometastases might facilitate their progression to local or
Adjuvant treatment 2.029 (0.786−5.240) 0.144 distant recurrence.21–23
Peritoneal infection 2.954 (1.100−7.931) 0.032 In this sense, it has been suggested that proinflammatory medi-
∗
Samples were obtained on postoperative day 4 or at the time the infection ators, which can be secreted from macrophages and leukocytes in the
occurred. tumor microenvironment, play a role in the progression of colorec-
†Difference between postoperative and preoperative samples. tal cancer and other tumors by promoting proliferation and migra-
CEA indicates carcinoembryonic antigen; CI, confidence interval; HR, hazard tion of cancer cells. Leukotriene D4 can induce β-catenin activation
ratio.
and, thereby, migration of colon cancer cells.24 Recent studies have

942 | www.annalsofsurgery.com 
C 2014 Lippincott Williams & Wilkins

Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Annals of Surgery r Volume 260, Number 5, November 2014
also documented that interleukin-6 enhances cell migration in human
osteosarcoma25 and oral squamous cell carcinoma.26 Interleukin-1β
is also particularly compelling as it was shown to induce a phenotypic
shift in vitro in colon cancer cells to a cancer stem-cell–like pheno-
type, including features of epithelial-to-mesenquimal transition and
increased invasiveness similar to our present results.27
On the basis of the present results and those of other previous
investigations, we propose that the mechanism by which postoperative
peritoneal infection and subsequent inflammatory response enhance
tumor recurrence is multifactorial. It is conceivable that a combi-
nation of mechanisms—amplification of angiogenesis, stimulation of
migration and invasion capacities, and others yet to be identified—act
in combination to increase the risk of tumor recurrence.
Our model not only supports tumor cell invasiveness as a new
mechanism linking infection with tumor recurrence but also offers
an experimental setting to identify particular inflammatory factors
involved in tumor recurrence. Therefore, characterizing the patterns
of cytokine expression associated with infection and tumor recur-
rence, especially in peripheral blood, might lead to the identification
of better biomarkers to predict the risk of tumor recurrence after
surgery.
REFERENCES
1. Ferlay J, Shin HR, Bray F, et al. Estimates of worldwide burden of cancer in
2008: GLOBOCAN 2008. Int J Cancer. 2010;127:2893–2917.
2. Lay YT, Chang SC, Yang SH, et al. Comparison of clinicopathological charac-
teristics and prognosis between early and late recurrence after curative surgery
for colorectal cancer. Am J Surg. 2014;207:922–930.
3. Bell SW, Walker KG, Rickard MJ, et al. Anastomotic leakage after curative
anterior resection results in a higher prevalence of local recurrence. Br J Surg.
2003;90:1261–1266.
4. Walker KG, Bell SW, Rickard MJ, et al. Anastomotic leakage is predictive of
diminished survival after potentially curative resection for colorectal cancer.
Ann Surg. 2004;240:255–259.
5. McArdle CS, McMillan DC, Hole DJ. Impact of anastomotic leakage on long-
term survival of patients undergoing curative resection for colorectal cancer.
Br J Surg. 2005;92:1150–1154.
6. Law WI, Choi H, Lee Y, et al. Anastomotic leakage is associated with poor
long-term outcome in patients after curative colorectal resection for malig-
nancy. J Gastrointest Surg. 2007;11:8–15.
7. Ptok H, Marusch F, Meyer F, et al. Impact of anastomotic leakage on oncolog-
ical outcome after rectal cancer resection. Br J Surg. 2007;94:1548–1554.
8. Katoh H, Yamashita K, Wang G, et al. Anastomotic leakage contributes to the
risk for systemic recurrence in stage II colorectal cancer. J Gastrointest Surg.
2011;15:120–129.
9. McMillan DC, Wotherspoon HA, Fearon KC, et al. A prospective study of
tumor recurrence and acute-phase response after apparently curative colorectal
cancer surgery. Am J Surg. 1995;170:319–332
10. McMillan DC, Canna K, McArdle CS. Systemic inflammatory response pre-
dicts survival following curative resection of colorectal cancer. Br J Surg.
2003;90:215–219.
11. Alonso S, Pascual M, Salvans S, et al. Postoperative peritoneal infection in-
creases colorectal cancer recurrence: a prospective matched cohort study of
inflammatory and angiogenic responses as mechanisms responsible for this
association. Colorectal Dis. 2010;12(suppl 3):24.
12. Bohle B, Pera M, Pascual M, et al. Postoperative intra-abdominal infection
increases angiogenesis and tumor recurrence after surgical excision of colon
cancer in mice. Surgery. 2010;147:120–126.
13. Landegren U. Measurement of cell numbers by means of the endogenous
enzyme hexosaminidase. Applications to detection of lymphokines and cell
surface antigens. J Immunol Methods. 1984;67:379–388.
14. Givens KT, Kitada S, Chen AK, et al. Proliferation of human ocular fibroblasts.
An assessment of in vitro colorimetric assays. Invest Ophthalmol Vis Sci.
1990;31:1856–1862.
15. Vaiopoulos AG, Kostakis ID, Koutsilieris M, et al. Colorectal cancer stem cells.
Stem Cells. 2012;30:363–371.
16. Farid SG, Aldouri A, Morris-Stiff G, et al. Correlation between postoperative
infective complications and long-term outcomes after hepatic resection for
colorectal liver metastasis. Ann Surg. 2010;251:91–100.

C 2014 Lippincott Williams & Wilkins
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Peritoneal Infection and Tumor Recurrence
17. Nagasako Y, Satoh S, Isogaki J, et al. Impact of anastomotic complications
on outcome after laparoscopic gastrectomy for early gastric cancer. Br J Surg.
2012;99:894–854.
18. Eriksen MT, Wibe A, Norstein J, et al. Norwegian Rectal Cancer Group.
Anastomotic leakage following routine mesorectal excision for rectal cancer
in a national cohort of patients. Colorectal Dis. 2005;7:51–57.
19. Richards CH, Platt JJ, Anderson JH, et al. The impact of perioperative risk,
tumor pathology and surgical complications on disease recurrence following
potentially curative resection of colorectal cancer. Ann Surg. 2011;254:83–89.
20. Mirnezami A, Mirnezami R, Chandrakumaran K, et al. Increased local recur-
rence and reduced survival from colorectal cancer following anastomotic leak:
systematic review and meta-analysis. Ann Surg. 2011;253:890–899.
21. Fermor B, Umpleby HC, Lever JV, et al. Proliferative and metastatic potential
of exfoliated colorectal cancer cells. J Natl Cancer Inst. 1986;76:347–349.
22. Allen-Mersh TG, McCullough TK, Patel H, et al. Role of circulating tumour
cells in predicting recurrence after excision of primary colorectal carcinoma.
Br J Surg. 2007;94:96–105.
23. Peeters C, de Waal R, Wobbes T, et al. Metastatic dormancy imposed by the
primary tumor: does it exists in humans? Ann Surg Oncol. 2008;15:3308–3315.
24. Salim T, Sand-Dejmek J, Sjölander A. The inflammatory mediator leukotriene
D4 induces subcellular β-catenin translocation and migration of colon cancer
cells. Exp Cell Res. 2014;321:255–266.
25. Lin YM, Chang ZL, Liao YY, et al. IL-6 promotes ICAM-1 expression and
cell motility in human osteosarcoma. Cancer Lett. 2013;328:135–143.
26. Chuang JY, Huang YL, Yen WL, et al. Syk/JNK/AP-1 signaling pathway me-
diates interleukin-6-promoted cell migration in oral squamous cell carcinoma.
Int J Mol Sci. 2014;15:545–559.
27. Li Y, Wang L, Pappan L, et al. IL-1β promotes stemness and invasiveness of
colon cancer cells through Zeb1 activation. Mol Cancer. 2012;11:87.
DISCUSSANTS
P.M. Schneider (Zurich, Switzerland):
Thank you, Dr. Pera, for the clear and comprehensive presen-
tation of your important study.
You engaged in a very interesting but also difficult study, to
identify mechanisms, which could explain potential higher recurrence
rates after colorectal cancer surgery in cases of postoperative peri-
toneal infection due to anastomotic leakage and/or intra-abdominal
abscess.
This is a particularly hot topic of debate, which has caused
concern among gastrointestinal (GI) cancer surgeons, and we know
little to nothing about its underlying mechanisms.
You aimed to study the effect of the capacities of postoperative
peritoneal infection on proliferation, migration, and invasion of 2
cancer cell lines in vitro, after surgery for colorectal cancer, in a
matched cohort with and without infection.
In summary, you tried to demonstrate and prove that peritoneal
fluid (local factors) taken from infected patients enhanced both the
cell migration and cell invasion capacities of cancer cell lines. With
respect to the serum samples (systemic factors), you observed an
effect only in the cell migration assays.
Your scientific model might enable us to link infection with tu-
mor recurrence, introducing a way to further explore factors involved
in tumor recurrence.
As is always the case in difficult endeavors, the answer to a
difficult problem is not easy and some points need further clarification.
To open the discussion of your paper, I have 3 questions.
Some postoperative peritoneal liquid samples exerted a cyto-
toxic effect in the cell cultures. You claimed that this was not related to
the presence of infection and your analysis was, therefore, restricted
to the subset of noncytotoxic samples. Could you comment on this
phenomenon, as it might represent a bias in your model? How would
the results differ if the so-called “toxic samples” had been included?
The matching process of your patients is undoubtedly impor-
tant, as this can influence your study results. Could you explain how
this was done in more detail?
www.annalsofsurgery.com | 943
Salvans et al
What are the potential clinical consequences of your observa-
tion, taking into consideration that one logical consequence would be
adjuvant chemotherapy, which is not frequently possible?
Again, your study shows very interesting and promising data.
You must be commended for your excellent work. I would like to
thank the Association for the privilege of being the first discussant.
Response From M. Pera:
Thank you for your comments. First, regarding the cytotoxic
effect of the peritoneal fluid samples, this information was included
only in the manuscript due to time limitations. Some peritoneal fluid
samples in this study and in other studies we have done exerted cy-
totoxic effects on the tumor cells in vitro. So, we had no viable cells
after incubation. We have not been able to deduct what the mech-
anisms responsible for these cytotoxic effects are. However, they
are not related to the presence of infection, contrarily to what we
could imagine. But, we found the same number of cytotoxic sam-
ples between the infection group and the control group, and for this
reason, we do not think that this is a bias, as the main differences,
even with a minor number of samples, were observed when we com-
pared migration and invasion in the peritoneal samples. We are now
investigating which mechanisms are responsible for this cytotoxic
effect. We think that cells, which survive, have the opportunity to
proliferate.
Second, regarding the matching process, we, of course, didn’t
know which patients were going to have an infection. As such, during
this period, we collected samples from all patients who underwent
surgery with a curative intent. Patients needed to have an absolutely
uncomplicated postoperative course and there were several matching
criteria. We faced problems and even had to enlarge the recruitment
period to obtain controls for our patients with infection. Briefly, there
was no more than one control for each patient with an infection.
Finally, regarding the clinical implications, we’d like to identify tar-
gets that we could use to treat these patients. If we can identify the
mechanisms, we can maybe provide a more appropriate treatment.
I think that the first step would be to follow a randomized control
trial, comparing the benefit of adjuvant treatment in this group of
944 | www.annalsofsurgery.com
Copyright © 2014 Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Annals of Surgery r Volume 260, Number 5, November 2014
patients. This is something that hasn’t been done. We know that
some of these patients may not even receive chemotherapy due to
the complications. However, the majority of these patients, provided
with an early diagnosis, could receive chemotherapy. We just need a
randomized control trial to uphold this.
DISCUSSANTS
J. van Lanschot (Rotterdam, The Netherlands):
It has been suggested, for different GI tumors, that it is not
only anastomotic leakage but complications in general that facilitate
recurrence. Based on this suggestion, are you certain that recurrence
is specifically related to the local problem of anastomotic leakage?
Could it be that tumor outgrowth was stimulated by a more general
effect of complications; for instance, of pneumonia or a myocardial
infarction? It has also been suggested that, for other GI tumors, com-
plications have an impact on the timing of the recurrence; however,
if you wait long enough, both survival lines will cross again. What
was the time-span you observed on the x axis? You saw a difference
in local recurrence between the 2 groups. Could it be that if you were
to have a longer follow-up period, these lines would overlap again,
as has been suggested for esophageal cancer? It is the timing of the
recurrence that is influenced by the complications, but in the end, the
recurrence is comparable, at least in esophageal cancer.
Response From M. Pera:
Yes, I agree. We do not believe in a mechanistic hypothesis,
as has been suggested by other authors, who propose that exfoliated
tumors inside the lumen of the bowel may deposit extraluminally
through the leak. We think that the problem is related to the inflam-
matory response, which is much more intense at a local level of the
peritoneum. However, anything that increases the surgical response to
trauma, even pneumonia, might have an influence on the oncological
outcome. With regard to your second question, esophageal cancer has
a worse prognosis than colorectal cancer. But, in our group the me-
dian follow-up was 48 months, and there was nothing that suggested
that the lines were going to cross.

C 2014 Lippincott Williams & Wilkins