Professional Documents
Culture Documents
Clinical Science
Division of General and Gastrointestinal Surgery, Massachusetts General Hospital and Harvard
Medical School, Boston, MA, USA
KEYWORDS: Abstract
Emergency surgery; BACKGROUND: Emergency presentation with colon cancer is intuitively related to advanced dis-
Colon cancer; ease. We measured its effect on outcomes of surgically treated colon cancer.
Survival; METHODS: A retrospective cohort of 1,071 surgical colon cancer patients (2004 to 2011), with 102
Disease-free survival; emergency cases requiring surgery within the index admission, was analyzed.
Pathology; RESULTS: Emergency patients required longer surgeries (median 141 vs 124 minutes; P 5 .04),
Perioperative outcomes longer median admissions (8% vs 5%; P , .001), more readmissions (12.7% vs 7.1%; P 5 .040),
and perioperative mortality (7.8% vs .8%; P , .001). Surgical pathology displayed higher rates of
node-positive disease (56.6% vs 38.6%; P , .001), extramural vascular invasion (39.6% vs 29.1%;
P 5 .021), and metastatic disease (19.6% vs 8%; P , .001). Consequently, adjusting for staging, emer-
gency presentations had considerably higher mortality (odds ratio 5 2.07; P 5 .003) and shorter
disease-free survival (hazard ratio 5 1.39; P 5 .042).
CONCLUSIONS: Emergency presentation is a stage-independent poor prognostic factor associated
with aggressive tumor biology, resulting in longer surgeries and admissions, frequent readmissions,
worsening outcomes, and increasing healthcare costs.
Ó 2015 Elsevier Inc. All rights reserved.
Colon cancer is a malignancy where outcomes are as high as 97.4% for early stage disease.1 Conversely, late
strongly related to the stage of disease on presentation. detection with widespread metastatic disease can lower
Timely detection leads to 5-year survival rates that can be this survival rate to 8.1% at 5 years.2 Screening initiatives
This work was conducted with support from Harvard Catalyst, The Harvard Clinical and Translational Science Center (National Center for Research
Resources and the National Center for Advancing Translational Sciences, National Institutes of Health Award 8UL1TR000170-05, and financial contributions
from Harvard University and its affiliated academic healthcare centers). The content is solely the responsibility of the authors and does not necessarily repre-
sent the official views of Harvard Catalyst, Harvard University, and its affiliated academic healthcare centers, or the National Institutes of Health. This work
was also supported in part by the Dutch Cancer Society, the Dutch Digestive Society, the Amsterdam University Funds, the Royal Netherlands Academy of
Arts and Sciences, and the Fulbright Foundation.
There were no relevant financial relationships or any sources of support in the form of grants, equipment, or drugs.
Oral presentation at the 9th Academic Surgical Congress, February 4 to 6, 2014, San Diego, California.
* Corresponding author. Tel.: 11-617-724-6980; fax: 11-617-724-0067.
E-mail address: dberger@mgh.harvard.edu
Manuscript received May 23, 2014; revised manuscript June 27, 2014
0002-9610/$ - see front matter Ó 2015 Elsevier Inc. All rights reserved.
http://dx.doi.org/10.1016/j.amjsurg.2014.07.014
R. Amri et al. Emergency surgery for colon cancer 247
are currently being assessed and implemented worldwide to presentation. The pathological characteristics compared
lower the disease burden and to stimulate early detection.3 included the tumor, node, metastasis (TNM) classification
On the other side of the spectrum, research has also focused as well as pathological characteristics with prognostic
on factors that predict delayed presentation, worse staging significance. Poor prognostic factors compared included
at baseline, and thus poorer outcomes. Emerging evidence positive resection margins (expressed as R1), tumor grade
has already shown that certain symptoms on presentation (expressed as high-grade disease), extramural vascular inva-
predict advanced disease and poorer outcomes.4 Higher sion,8 perineural invasion,9 infiltrating tumor border configu-
risk of advanced disease on presentation has also been iden- ration,10 and absence of peritumoral lymphocytic response.11
tified in certain demographic groups, including underprivi- Inpatient characteristics which were analyzed included
leged socioeconomic groups,5 and especially ethnic surgical duration, rates of laparoscopic surgery and con-
minorities.6,7 version to open surgery, rates of multivisceral resection,
Intuitively, emergency presentation is expected to be an and characteristics of the surgical admission including
unequivocally poor prognostic factor in patients with colon length of stay, rates of major surgical complications, and
cancer, as the symptoms that lead patients to present at an 30-day rates of readmission, reoperation, and death.
emergency department (ED), including intestinal bleeding, Finally, the long-term outcomes compared were the need
perforation, or obstruction, are usually attributable to for postoperative chemotherapy as well as survival and
advanced disease. disease-free survival, both expressed as duration and
This article analyzes colon cancer patients who are dichotomized as death, colon cancer-related death, overall
admitted after emergency presentation and eventually un- metastatic disease, and metastatic recurrence. These end
dergo surgical resection for colon cancer. It assesses the outcomes were also analyzed in multivariate models
magnitude of the effects associated with emergency presen- controlling for any significant covariates encountered
tation on staging, surgical stay, and cancer-related outcomes. during baseline analysis, as well as staging, expressed in
the American Joint Committee on Cancer (AJCC) 7th
Methods edition classification (ie, subdividing disease in stages 0, I,
IIA, IIB, IIIA, IIIB, IIIC, and IV)12
Patient population
Statistical analysis
All colon cancer patients surgically treated at Massa-
chusetts General Hospital in the 2004 to 2011 timespan A 2-tailed P value below .05 was considered statistically
were included for analysis. Data on all cases were collected significant. We compared dichotomous outcomes among
from prospectively maintained internal data repositories or emergency and elective patients using the chi-square test
accrued from patient records under an institutional review and a relative risk (RR) calculation, while a 1-way analysis
board-approved protocol. of variance or a univariate linear regression, with the un-
standardized B regression coefficient as a point estimate,
was used for continuous outcomes. Multivariate analysis
Definitions and objectives was performed to control the findings for any potentially
significant confounders found during univariate analysis.
Emergency presentation is defined as presentation or The Cox proportional hazards model was used for time-
referral to our center through the ED with symptoms and related outcomes, while binary logistic regression was
clinical findings requiring admission and urgent surgical used for dichotomous outcomes.
treatment within the index admission. All cases had a
working diagnosis of colon cancer, or at least colonic
malignancy as part of the differential diagnosis before
Results
operation, which was in all cases followed by subsequent
confirmation of malignancy through surgical pathology. We included 1,071 patients, of whom 102 were emer-
Cases where the presence of colonic adenocarcinoma was gency admissions, 83 of which came directly from our ED.
never confirmed pathologically were not part of the
included sample. Baseline characteristics
Differences between emergency admissions and the
remainder of patients were assessed on 4 main levels: Baseline characteristics are displayed in Table 1. There
presentation characteristics, surgical pathology, surgical stay were nonsignificant differences in general characteristics,
outcomes, and long-term outcomes. Presentation characteris- with patients with emergency presentation being slightly
tics included general demographics (age, sex, ethnicity), older, with lower body mass index, more ethnic minority
lifestyle (alcohol and tobacco consumption, body habitus), patients, and a relative overrepresentation of women. Emer-
comorbidity (expressed through a colon cancer-adjusted gency patients had lower but nonsignificant rates of polyp-
Charlson comorbidity score as well as type 2 diabetes, osis, a previous history of colorectal carcinoma, and
separately), and finally symptoms present at baseline inflammatory bowel disease and otherwise comparable
248 The American Journal of Surgery, Vol 209, No 2, February 2015
be explained by higher rates of multivisceral resections resections (69.6% vs 79.7%; P 5 .045), as well as a
(21.6% vs 13%; P 5 .014), and lower rates of laparo- nonsignificant trend toward lower rates of specialized
scopic surgery (RR 5 .44; P 5 .014) and segmental colorectal surgeons as the operating attending for
indicated a clear increase in risk of death and shorter sur- been shown to be quite different in tumor biology22 and
vival (hazards ratio 5 2.18; P , .001). Similarly, adjusting subsequent prognosis23 and arguments to split both are a
disease-free survival for baseline AJCC staging still indi- recurring issue in outcome studies.24
cated a significantly higher hazard for recurrence and hence Our work is therefore the first to review a large cohort
shorter disease-free survival (hazards ratio 5 1.39; P 5 of colon cancer-specific patients in a single center. This
.042). Logistic regression reiterates these findings after population also has the heterogeneous composition of a
adjustment for the same covariates for odds ratios (ORs) large urban public hospital in the screening era, with on
of death and metastatic disease, while metastatic recurrence one side of the spectrum a significant portion benefitting
(in patients without metastatic disease on presentation), from early detection through screening25 and on the other
which was significant on univariate analysis (OR 5 1.9; side patients who present through the ED and require
P 5 .033), was no longer significant after adjustment for emergency admission and surgery. We aimed to provide
baseline staging (OR 5 1.55; P 5 .17). These survival char- a comprehensive overview of the impact of these emer-
acteristics are also shown in survival curves with and gency presentations on all aspects of their subsequent
without adjustment of covariates, in Kaplan–Meier survival care, including the course of the index admission, patho-
curves, and Cox proportional hazards survival estimates, logical characteristics, and both short- and long-term
respectively, as displayed in Fig. 1. outcomes.
We found that patients admitted in an emergency setting
largely had the same general characteristics as the
Comments remainder of the population, except for having a slightly
worse lifestyle. These patients did of course present with
The aim of this study was to verify whether patients far higher rates of symptoms that are prodromal of
diagnosed with colon cancer in an emergency setting with complicated and advanced disease, likely contributing to
admission and subsequent surgical treatment differed in longer surgeries, higher complication rates, and a longer
presentation characteristics and had more advanced staging surgical stay. Surgical pathology subsequently showed far
and worse outcomes, both during and after surgical more advanced and aggressive disease on pathology, which
admission. in the long term led to shorter survival and disease-free
The current literature supporting this hypothesis is survival, which was not identified in previous studies using
scarce and crippled by applicability issues for the current smaller populations.19,26 This difference in long-term out-
colon cancer population. Some papers have produced comes was significant and of considerable magnitude.
valid points that echo the intuitive assumptions many Our findings show that emergency presentation predicts
clinicians have on the topic, but many of the papers laying more advanced disease and far worse outcomes, even after
the foundation for this work are outdated as they are from adjusting for staging at diagnosis. The detailed pathological
the prescreening era: a 1995 study by Scott et al13 gave characteristics of tumors of emergency admission patients
some of the first evidence pointing at more advanced dis- hint at a possible explanation for this stage-independent
ease and worse perioperative outcomes for emergency disparity. Aside from classical TNM staging, many histo-
colorectal cancer patients, underlining the importance of pathological indicators of aggressive tumor behavior were
screening initiatives, and a 1998 study by Porta et al14 far more common in emergency presentation patients.
demonstrated shorter survival rates for patients admitted These included high-grade disease, lymphovascular and
through the ED in a small set of 80 colon cancer patients. perineural invasion, transmural growth, and margin posi-
Finally, with data from the same decade, Biondo et al15 tivity; all these factors are likely to shorten the time interval
confirmed differences in recurrence and survival in a pro- between onset of disease and appearance of (serious)
spective cohort (1996 to 1998), although not in all stages symptoms, and are independent predictors of poor out-
of disease. comes in their own right.
In a more recent work, a British pilot study indicated a Another potential explanation lies in the baseline
dramatic decrease in emergency admissions for colorectal lifestyle characteristics: emergency presentations had
cancer by virtue of fecal occult blood test screening,16 higher smoking and alcohol abuse rates. Aside from
although later reports of a local study in Canada17 and an intrinsically adding even more risk and disease burden,
audit in England18 showed that rates of emergency admis- these characteristics are also associated with low socioeco-
sions for colon cancer remained alarmingly high, despite nomic status,27,28 low health literacy,29 and poor access to
both areas providing universal health care with a colorectal health care,30 which may explain in part why emergency
screening program. A second issue that makes the findings presentations come with more serious disease: for these un-
of previous work hard to interpret specifically for colon derprivileged patients, often reluctant to seek medical
cancer is the lack of differentiation that all but 214,15 attention until their symptoms become too severe to ignore,
of the discussed articles make between colon and rectal the ED is often the primary presentation site. In that sense,
diseasedan issue shared with many publications on the it is hardly a surprise to see these patients with both more
matter.19–21 Malignancies of the colon and rectum have advanced and more aggressive disease.
252 The American Journal of Surgery, Vol 209, No 2, February 2015
Figure 1 Univariate and multivariate survival and disease-free survival models. The red lines depict patients admitted in an emergency
setting, while the blue lines show the same outcome for the remainder of the population. (For interpretation of the references to color in this
Figure, the reader is referred to the web version of this article.)