International Journal of Surgery 25 (2016) 134e144

Contents lists available at ScienceDirect

International Journal of Surgery

journal homepage: www.journal-surgery.net

Review

The diagnostic accuracy of carcinoembryonic antigen to detect

colorectal cancer recurrence e A systematic review
Caspar G. Sørensen a, *, William K. Karlsson a, Hans-Christian Pommergaard b,
Jakob Burcharth c, Jacob Rosenberg c
a
Faculty of Health Sciences e University of Copenhagen, Blegdamsvej 3 e 2200 København N, Denmark
b
Hvidovre Hospital e University of Copenhagen, Department of Surgery, Kettegård Alle 30 e 2650 Hvidovre, Denmark
c
Herlev Hospital e University of Copenhagen, Centre for Perioperative Optimization, Department of Surgery, Herlev Ringvej 75 e 2730 Herlev, Denmark

h i g h l i g h t s

Reporting on the reference standards used was not optimal.

Results point toward a sensitivity of CEA ranging between 50% and 80%.

Results point toward a specificity and negative predictive value above 80%.

A clinically relevant effect on patient mortality remains to be proven.

a r t i c l e i n f o a b s t r a c t

Article history: Introduction: Carcinoembryonic Antigen (CEA) has been used as a tumor marker in the follow-up of
Received 10 November 2015 colorectal cancer for more than 40 years. Controversy exists regarding its diagnostic applicability due to a
Accepted 29 November 2015 relatively low sensitivity and a questionable effect on mortality. The aim of this review was to assess the
Available online 15 December 2015
diagnostic accuracy of CEA in detecting recurrence after intended curative surgery for primary colorectal
cancer.
Keywords:
Methods: Systematic literature searches were performed in PubMed, EMBASE and Cochrane databases,
Carcinoembryonic antigen
and articles were chosen based on predefined inclusion criteria. Reference lists from included articles
Biomarkers
Colorectal cancer
were manually searched for additional publications of relevance.
Curative surgery Results: Forty-two original studies with generally representative populations and long follow-up were
Cancer recurrence included. Data were reported on outcomes from 9,834 CEA tests during follow-up. Reporting on the
Follow-up reference standards used was not optimal. Sensitivity of CEA ranged from 17.4 % to 100 %, specificity
ranged from 66.1 % to 98.4 %, positive predictive value ranged from 45.8 % to 95.2% and negative pre-
dictive value ranged from 74.5 % to 100 %.
Conclusion: Results point toward a sensitivity of CEA ranging between 50 % and 80 %, and a specificity
and negative predictive value above 80 %. Results on positive predictive value showed low reliability.
Overall, CEA did not effectively detect treatable recurrences at an early stage, and a clinically relevant
effect on patient mortality remains to be proven.
© 2015 IJS Publishing Group Limited. Published by Elsevier Ltd. All rights reserved.

1. Introduction curative treatment, remains a complex challenge for the healthcare

system, both in terms of diagnostics, treatment and monetary costs.
Colorectal cancer is the third most common cancer worldwide, As a means of assistance, biomarkers have become increasingly
accounting for almost 10% of all cases [1]. Recent studies with large popular to use in cancer follow-up.
cohorts show recurrence rates of 20e30% [2,3]. Follow-up after Discovered in 1965, Carcinoembryonic Antigen (CEA) is an
oncofetal antigen produced by endodermally derived epithelial
tumor cells in the digestive tract [4]. As a blood-borne biomarker,
CEA represents a potentially cheap [5], safe and noninvasive test for
* Corresponding author. Godthåbsvej 89, 4. tv, 2000 Frederiksberg, Denmark.
the follow-up of colorectal cancer patients. However, results
E-mail address: [email protected] (C.G. Sørensen).

http://dx.doi.org/10.1016/j.ijsu.2015.11.065
1743-9191/© 2015 IJS Publishing Group Limited. Published by Elsevier Ltd. All rights reserved.
 C.G. Sørensen et al. / International Journal of Surgery 25 (2016) 134e144 135

concerning its effect on reducing mortality have been contradictive compared with the standard of clinical diagnostics (biopsy, surgery,
[6,7], and CEA can be elevated due to a number of reasons, that is endoscopy, radiology, and clinical examination).
both malignant [8] and benign diseases [9], as well as in smokers If patient data overlapped between publications, we selected the
[10]. most appropriate publication to avoid duplicate data based upon
The purpose of this systematic review was to assess the diag- the following factors in descending order: Data on both true posi-
nostic accuracy of CEA in detecting recurrence, after intended tive, false positive, false negative and true negative values, size and
curative surgery for primary colorectal cancer. relevance of the cohort (relevant exclusions, both colon and rectal
cancers represented), and duration of follow-up.
2. Material and methods
2.4. Inclusions and exclusions
2.1. Protocol and reporting
The following types of publications were excluded: Reviews,
A protocol was developed and registered in the PROSPERO caseecontrol studies, experimental studies, meeting abstracts,
database of systematic reviews [11]. comments, correspondences etc.
This systematic review was reported in accordance with the Studies were also excluded for the following reasons:
PRISMA guidelines [12].

The methods or language used were unclear.
2.2. Data sources and search methods
The cohort consisted entirely of either recurred, CEA positive or
CEA negative patients.
A literature search was conducted in PubMed, EMBASE and
A cut-off value for CEA was not stated, or if the same cut-off
Cochrane databases on October 8, 2014. value did not apply to everyone.
The full search was as follows:
CEA was only measured within 3 months postoperatively, since
this was considered as prognosticating to account for residual
((((((((((((((((((((((((((((((((((((((((((((((((((((((((cancer of the co- disease.
lon) OR cancers of the colon) OR colon cancer) OR colon cancers)
OR colon carcinoma) OR colon carcinomas) OR colon neoplasm) Furthermore, studies were excluded if data were mixed in an
OR colon neoplasms) OR colon tumor) OR colon tumors) OR inappropriate way, e.g. curatively and non-curatively operated pa-
colon tumour) OR colon tumours) OR colonic cancer) OR colonic tients, primary cancers with recurrences, first time recurrences
cancers) OR colonic carcinoma) OR colonic carcinomas) OR with second time recurrences or if results were calculated per test
colonic neoplasm) OR colonic neoplasms) OR colonic tumor) OR or per recurrence, instead of per patient.
colonic tumors) OR colonic tumour) OR colonic tumours) OR
cancer of the rectum) OR cancers of the rectum) OR rectal 2.5. Study selection
cancer) OR rectal cancers) OR rectal carcinoma) OR rectal car-
cinomas) OR rectal neoplasm) OR rectal neoplasms) OR rectal Results from the database search were transferred into a
tumor) OR rectal tumors) OR rectal tumour) OR rectal tumours) document making up one long list of publications. Publications
OR rectum cancer) OR rectum cancers) OR rectum carcinoma) written in English were assessed for duplicates. After this, authors
OR rectum carcinomas) OR rectum neoplasm) OR rectum neo- CGS and WKK independently screened the titles and abstracts for
plasms) OR rectum tumor) OR rectum tumors) OR rectum potentially relevant publications. Disagreements were discussed
tumour) OR rectum tumours) OR colorectal cancer) OR colo- and in cases of doubt, the discussion was settled by authors JB, JR
rectal cancers) OR colorectal carcinoma) OR colorectal carci- and HCP.
nomas) OR colorectal neoplasm) OR colorectal neoplasms) OR
colorectal tumor) OR colorectal tumors) OR colorectal tumour) 2.6. Data extraction
OR colorectal tumours))
AND Data were extracted independently by the first author into a
(((((((((CEA) OR carcinoembryonic antigen) OR carcinoem- predefined Excel sheet. This included article demographics, recur-
bryonic antigens) OR CD66e Antigen) OR CD66e Antigens) OR rence information, the diagnostic methods used and the diagnostic
CD66e-Antigen) OR CD66e-Antigens) OR CD66eAntigen) OR properties of CEA measurements. The Excel sheet was piloted using
CD66eAntigens)) 10 studies, and revised.
AND If studies examined more than one cut-off value, data were re-
((((((((recurrence) OR relapse) OR recrudescence) OR follow-up) ported on the cut-off the authors seemed to prefer or mentioned
OR follow up) OR surveillance) OR monitoring) OR sustained the most. If this was not clear, data were reported on the cut-off
remission). used in standard practice.

2.7. Definitions of outcome measures

2.3. Eligibility criteria

A true positive test was defined as a positive CEA test (CEA above
Cohort studies and randomized controlled trials written in En- cut-off) before or at the time of recurrence detection. However,
glish were included. five studies required a positive CEA test before the detection of
No restrictions were made on the year of publication. recurrence [28,39,51e53], and these were handled like the other
The populations had to comprise adult humans with primary studies. Recurrences had to be detected within the follow-up
colorectal cancer, operated on with curative intent. period.
At least one postoperative CEA measurement was required. Data
A true negative was defined as a patient with negative CEA
for calculation of either sensitivity, specificity, positive predictive testing and no evidence of recurrence at the end of follow-up.
value or negative predictive value for CEA had to be available.
A false negative was defined as a diagnosis of recurrence in a
The diagnostic abilities of CEA in detecting recurrence, had to be patient with normal CEA at the time. In five studies
136 C.G. Sørensen et al. / International Journal of Surgery 25 (2016) 134e144

Fig. 1. Flow diagram of systematic literature search.

[28,39,51e53], a test was also false negative, if a diagnosis of thereby the potential to introduce bias or limit applicability.
recurrence coincided with the first positive CEA test. Accordingly “þ” refers to optimal methodology, “-” is less than

A false positive was defined as a positive CEA test without evi- optimal, and “?” means data were insufficient to decide.
dence of recurrence in the follow-up period.
3. Results
In this respect, it has to be mentioned that some studies
required two consecutive CEA values above cut-off to qualify as a
3.1. Study selection
positive test [20,30,31,33,37,41,46,49,54,55].
The primary endpoint was the sensitivity, specificity, positive
The literature search produced 6378 results. 5259 publications
predictive value and negative predictive value of CEA in detecting
were written in English and assessed for duplicates. When dupli-
colorectal cancer recurrence.
cates were removed, the titles and abstracts from 3569 publications
were assessed for inclusion.
2.8. Assessment of quality Fig. 1 illustrates the process of identifying eligible studies as a
flow diagram, in accordance with the PRISMA guidelines for sys-
Methodological quality assessment was performed using the tematic reviews [12].
QUADAS tool, modified by the Cochrane Collaboration [13]. This in From the database search, 128 publications were deemed
an 11 point questionnaire with up to 9 additional items. Items are potentially relevant. Five of these could not be obtained. A further
rated “þ”, “-” and “?”, depending on methodological quality and four publications were identified by manually searching the
 C.G. Sørensen et al. / International Journal of Surgery 25 (2016) 134e144 137

Table 1
Article demographics, disease information, follow-up duration and study design of included studies.

Demographics Disease information Follow-up Study

duration design
Year Lead Patient population Exclusions of comorbidities, risk Number of Site of recurrences (%) Method of
author (n) factors and conditions known to recurrences detecting
interfere with CEA measurements (%) recurrence (%)

2014 Shibutani [14] Operated with Patients undergoing preoperative 32 (13) Liver (55), Lung (24), 48e132 months RC
curative intent for therapy Anastomosis (24),
stage II colorectal Peritoneal dissimination
cancer (238) (10)
2014 Zhong [15] Operated with Deaths from other causes, lost to 144 (39) 60 months RC
curative intent for follow-up, preoperative
colorectal cancer chemotherapy or radiotherapy
(373)
2013 Kim [16] Operated with Abnormal CEA < 3 months after 79 (24) Pathology, 36e134 months RC
curative intent for surgery, follow-up less than 3 radiology
stage IIeIII years, history of other cancers,
colorectal cancer chemoradiation before primary
(336) surgery
2012 Fora [17] Operated with Patients who did not complete 44 (25) >1 site (30), Liver (23), Minimum 36 RC
curative intent for the follow-up programme Peritoneum/perineum/ months
stage IIeIII ovary (18), Lung (14), (median 60)
colorectal cancer Other (11), Anastomosis
(177) (5)
2012 Su [18] Operated with Excluded: Patients without 90 (22) Local (46), Liver (19), Pathology, 3e118 months RC
curative intent for available preoperative CEA value, Lung (11), Bones (8), radiology (median 69)
colorectal cancer recurrences < 3 months after Adrenal (3), Renal (2),
(413) surgery Distal lymphatic (2),
Brain (1), Spinal (1), Not
accounted (7)
2011 Banaszkiewicz Operated with Patients with concommitant 112 (33) Liver (39), Local (29), Endoscopy (9), 60 months RC
[19] curative intent for ilness making survival of a second Disseminated (11), Two clinically (63),
colorectal cancer operation unlikely lesions (10), Lung (6), CEA (29)
(340) Other (5)
2010 Chen [20] Operated with Failure of CEA to normalize after 999 (25) Liver (49), Local (20), Pathology, Median 68 RC
curative intent for surgery Lung (9), Other (23) positive image months
stage IeIII colorectal study
adenocarcinoma supported by
(4076) symptoms
2010 Holt [21] Operated with Failure of CEA to normalize after 22 (12) Pathology, Median 36.5 RC
curative intent for surgery, missing preoperative radiology months
stage IeIII colorectal CEA, < 2 postoperative CEA
cancer (186) measurements
2010 Yakabe [22] Operated with Inappropriate follow-up 62 (27) Pathology or 60 months PC
curative intent for radiology
stage IeIII colorectal
cancer (227)
2009 Park [23] Operated with Synchronous metastatic disease, 291 (23) Liver (35), Other (65) Pathology, 1e156 months RC
curative intent for carcinoma in situ, T1 cancer, radiology, (median 48)
colorectal inflammatory bowel disease, clinically
adenocarcinoma familial adenomatous polyposis, supported by
(1263) chronic obstructive pulmonary biochemistry
disease, chronic liver disease,
peptic ulcer, diabetes
2008 Hara [24] Operated with Failure of CEA to normalize after 51 (29) Liver (33), Lung (28), Pathology (39), 60 months RC
curative intent for surgery, multiple cancers, Local (22), Lymph node radiology (61)
Dukes stage C insufficient examinations, (11), Peritoneum (5),
colorectal cancer squamous-cell carcinoma Hematologic (2)
(174)
2007 Grossmann Operated with Patients without at least 1 pre- 282 (30) Minimum 72 RC
[25] curative intent for and postoperative CEA value months
stage IeIII rectal
cancer (954)
2007 Irvine [26] Operated with Failure of CEA to normalize after 46 (33) Liver (44), Other (66) 27 months RC
curative intent for surgery, recurrences detected < 3 (mean)
Dukes stage AeC months after surgery
colorectal cancer
(139)
2007 Ko€rner [27] Operated with Patients > 75 years 37 (24) Locoregional (49), Liver Pathology, 60 months PC
curative intent for (27), Lung (14), Multiple radiology
colorectal (16), Other (11)
adenocarcinoma
(153)
2006 Fernandes [28] Operated with Previous malignancy, residual 23 (20) Radiology, 0.2e75 months PC
curative intent for macroscopic or microscopic endoscopy, (mean 22.3)
colorectal disease clinically
adenocarcinoma
(117)
(continued on next page)
138 C.G. Sørensen et al. / International Journal of Surgery 25 (2016) 134e144

Table 1 (continued )

Demographics Disease information Follow-up Study

duration design
Year Lead Patient population Exclusions of comorbidities, risk Number of Site of recurrences (%) Method of
author (n) factors and conditions known to recurrences detecting
interfere with CEA measurements (%) recurrence (%)

2004 Carpelan- Operated with Patients who received follow-up 40 (39) Local (43), Liver (25), Minimum 72 RC
Holmstro€m curative intent for at other hospitals, patients Other (33) months
[29] colorectal cancer without serum samples before
(102) surgery and at recurrence
detection
2003 Engarås [30] Operated with 39 (30) Radiology, 60e108 months PC
curative intent for autopsy,
colorectal cancer endoscopy
(132)
1999 Carriquiry [31] Operated with Postoperative deaths, patients 40 (28) Locoregional (50), Liver Pathology, Minimum 24 RC
curative intent for with <3 postoperative CEA (33), Lung (8), Bone (6), surgery, months
Dukes stage AeC measurements Spleen (2), Brain (2) radiology
colorectal cancer
(142)
1998 Li Destri [32] Operated with 45 (19) Liver (40), local (36), 12e120 months RC
curative intent for lung (9), brain (2), (mean 51.4)
colorectal cancer supraclavicular lymph
(239) node (2), mixed (11)
1997 Lucha Jr. [33] Operated with Synchronous cancers 62 (22) 1e156 months RC
curative intent for (mean 80)
Astler-Coller stage A
eC colorectal cancer
(280)
1997 Tobaruela [34] Operated with 21 (35) Liver (43), Locoregional 1e62 months PC
curative intent for (29), Combined (14), (median 27)
Dukes stage C Lung (14)
colorectal
adenocarcinoma
(60)
1995 Miles [35] Operated with Incomplete follow-up, patients 53 (42) Pathology, Mean 28 RC
curative intent for with <1 preoperative and <2 surgery, months
colorectal cancer postoperative CEA measurements endoscopy,
(125) radiology,
clinically
1994 Chiang [36] Operated with Failure of CEA to normalize after 101 (27) 60e144 months RC
curative intent for surgery, patients without
Dukes stage BeC complete data or follow-up
colorectal
adenocarcinoma
(380)
1994 Filella [37] Operated with 72 (21) Locoregional (50), Liver Always Mean 24.8 PC
curative intent for (21), Lung (10), Bone (4), pathology months
colorectal cancer Liver and locoregional (median 14.5)
(346) (6), Various (8), Other (1)
1994 McCall [38] Operated with <2 years follow-up, failure to 98 (32) 24e60 months RCT
curative intent for obtain CEA levels (median 54)
Dukes stage AeC
colorectal cancer
(311)
1993 Moertel [39] Operated with Recurrences and deaths < 3 417 (41) Liver (30), Lung (13), Pathology, 60e102 months RCT
curative intent for months from surgery Locoregional (11), surgery, (median 72)
stage IIeIII colon Peritoneal (7), endoscopy,
cancer (1017) Retroperitoneal (4), radiology
Other single sites (5),
Multiple sites (29)
1993 Zeng [40] Operated with 32 (28) PC
curative intent for
lymph node positive
colon cancer, with
normal preoperative
CEA (114)
1992 Meling [41] Operated with Failure of CEA to normalize after 50 (21) Liver (43), Locoregional Pathology, 33e60 months PC
curetive intent for surgery (37), Lung (11), endoscopy, (median 38)
Dukes stage AeC Subcutaneous lymph radiology
colorectal node (4), Liver and lung
adenocarcinoma (4), Bone (1)
(241)
1992 Seregni [42] Operated with 136 (40) Distant (63), Local (37) 2e185 months RC
curative intent for (mean 42.3)
colorectal cancer
(336)
 C.G. Sørensen et al. / International Journal of Surgery 25 (2016) 134e144 139

Table 1 (continued )

Demographics Disease information Follow-up Study

duration design
Year Lead Patient population Exclusions of comorbidities, risk Number of Site of recurrences (%) Method of
author (n) factors and conditions known to recurrences detecting
interfere with CEA measurements (%) recurrence (%)

1991 ~ as [43]
Camun Operated with Failure of CEA to normalize after 61 (40) Local (38), Disseminated Pathology, 24e120 months PC
curative intent for surgery, postoperative deaths, (28), Liver (25), Lung surgery, (mean 58)
colorectal adjuvant therapy, follow-up < 24 (10) radiology,
adenocarcinoma months clinically
(151)
1989 Secco [44] Operated with 38 (40) Distant (58), Local (42) Mininum 60 PC
curative intent for months
Dukes stage AeC
colorectal cancer
(96)
1985 Graffner [45] Operated with Deaths from other causes, 47 (28) Liver (38), Perineal (15), 12e60 months PC
curative intent for patients who moved from the Skin (11), Anastomotic (median 26)
colorectal cancer area or did not complete the (9), Lung (9), Bone (2),
(167) follow-up programme Multiple (17)
1984 Hine [46] Operated with Deaths from other causes, 171 (27) Liver (50), Local (32), Pathology, Mean 39.7 PC
curative intent for conditions known to interfere Other distant (15), radiology, months
colorectal cancer with CEA measurement, Metachronous (3) clinically
(626) psychiatric illness, patients lost to
follow-up or no CEA
measurement the last year
1983 Beart Jr. [47] Operated with 48 (29) Pathology 12e48 months PC
curative intent for
Dukes stage B2eC
colorectal cancer
(166)
1983 Carlsson [48] Operated with Deaths < 3 months from surgery, 50 (36) 3e70 months PC
curative intent for patients lost to follow-up (median 47)
colorectal cancer
(139)
1982 Lunde [49] Operated with 44 (25) Abdomen/pelvis (34), 6e72 months RC
curative intent for Liver (25), Locoregional (mean 27.4)
colorectal (25), Lung (16)
adenocarcinoma
(179)
1982 Steele Jr. [50] Operated with 218 (30) Liver (22), Locoregional Pathology, Colon: RCT
curative intent for (22), Lung (16), Single radiology, Minimum 21
Dukes stage B2eC other (21), Multiple perineal pain months
colorectal cancer other (8), Liver þ Other occurring (median 46).
(734) (5), Other (4), Site acutely after a Rectal:
Unconfirmed (2) pain-free Minimum 8
interval months
(median 44)
1982 Tate [51] Operated with Early recurrences, patients lost to 68 (15) Minimum 24 PC
curative intent for follow-up, other malignancies months
Dukes stage AeC
colorectal cancer
(468)
1981 Gray [52] Operated with 15 (20) Peritoneum (33), Liver Pathology, Up to 60 PC
curative intent for (27), Local (20), Liver/ radiology, months
colorectal cancer peritoneum (13), Bone clinically
(75) (7)
1981 Wedell [53] Operated with 86 (37) Local (53), Distant (47) Pathology, 60 months RC
curative intent for endoscopy,
colorectal cancer radiology
(235)
1980 Wood [54] Operated with 37 (25) Local (46), Liver (22), 20e56 months PC
curative intent for Bone (14), Lung (5), Local
colorectal and liver (5), Local and
adenocarcinoma bone (5), Local and
(148) metachronous primary
(3)
1974 Sorokin [55] Operated with Metachronous cancers, 6 (8) Supraclavicular lymph Pathology (50), 1e54 months PC
curative intent for previously reported recurrences, node (33), Lung (17), surgery (17), (mean 16)
colorectal recurrences before the time the Rectal wall (17), radiology (17),
adenocarcinoma study was initiated Locoregional (17), Liver clinically (17)
(74) and vaginal apex (17)

Abbreviations for Table 1:

PC: Prospective Cohort Study.
RC: Retrospective Cohort Study.
RCT: Randomized Controlled Trial.
140 C.G. Sørensen et al. / International Journal of Surgery 25 (2016) 134e144

Table 2
The diagnostic properties of CEA.

Year Lead author Carcinoembryonic antigen (CEA)

Method of measurement Double measurements Cut-off Required to be positive Sens Spec PPV NPV
(ng/ before detection of
mL) recurrence

2014 Shibutani [14] Radioimmunoassay (Abbott, Chiba, Japan) No 5 58.6%

2014 Zhong [15] Electrochemiluminescence immunoassay system No 5 87.5%
(Elecsys 2010, Roche, Basel, Switzerland)
2013 Kim [16] Immunoassay (ADVIA Centaur XP, Siemens, No 5 43.0% 91.1% 59.6% 83.9%
Erlangen, Germany)
2012 Fora [17] No 2.5 52.4%
2012 Su [18] CEA Elecsys analyzers (Roche Diagnostics GmbH, No 5 54.4%
USA)
2011 Banaszkiewicz No 5 85.7% 91.2% 82.8% 92.9%
[19]
2010 Chen [20] Increased values 5 67.6%
2010 Holt [21] No 5a 54.5%
2010 Yakabe [22] Latex immunoassay (Mitsubishi chemichal Ltd, No 5 66.1%
Japan)
2009 Park [23] No 7 41.9%
2008 Hara [24] No 5 51.0% 88.6% 65.0% 81.3%
2007 Grossmann No 5 54.6%
[25]
2007 Irvine [26] Bayer immunoassay (Siemens, Erlangen, No 10 65.2%
Germany)
2007 Ko€ rner [27] Immunoassay (Abbott, IL, USA) No 10 51.4%
2006 Fernandes [28] Delfia method (PerkinElmer, MA, USA) No 5 Yes 30.4%
2004 Carpelan- Immuno-flourometric assay (AutoDELFIA; Wallac, No 5 67.5%
Holmstro €m Turku, Finland)
[29]
2003 Engarås [30] DELFIA test kit (Wallac, Turku, Finland) Always 5.6 64.1%
1999 Carriquiry [31] Increased values 5 76.9% 97.8% 95.2% 88.0%
1998 Li Destri [32] Radioimmunoassay method No 5 71.1% 88.1% 58.2% 92.9%
1997 Lucha Jr. [33] Increased values 5 66.7% 91.8% 71.0% 90.1%
1997 Tobaruela [34] Enzyme-linked immunoassay (Boehringer, No 5a 42.9% 89.7% 69.2% 74.5%
Mannheim, Germany)
1995 Miles [35] International Reference Preparation 73/601 No 10b 86.8% 88.9% 85.2% 90.1%
(National Institute for Biological Standards and
Control)
1994 Chiang [36] Radioimmunoassay (Abbott) No 5 78.2% 93.2% 80.6% 92.2%
1994 Filella [37] Radioimmunoassay (Abbott, IL, USA) Increased values (further rise 5 51.4% 92.3% 63.8% 87.8%
between them also required)
1994 McCall [38] Enzyme-linked immunoassay (Abbott, IL, USA) No 5 58.2% 93.0% 79.2% 82.8%
1993 Moertel [39] No 5 Yes 59.2% 83.7% 71.6% 74.7%
1993 Zeng [40] No 5 43.8%
1992 Meling [41] Radioimmunoassay Increased values, unless it 5 56.0% 98.4% 90.3% 89.5%
coincides with recurrence
1992 Seregni [42] No 5 63.2% 92.5% 85.1% 78.7%
1991 Camun ~ as [43] No 5 68.9%
1989 Secco [44] EIA DUOMAB 60 (Roche) No 7.5 76.3% 94.8% 90.6% 85.9%
1985 Graffner [45] Radioimmunoassay No 10 66.0%
1984 Hine [46] Double antibody radioimmunoassay (Egan/ Increased values 20 78.9% 80.4% 60.3% 91.0%
Laurence)
1983 Beart Jr. [47] No 5 68.8% 68.6% 47.1% 84.4%
1983 Carlsson [48] Direct radioimmunoassay method No 7.5 76.0% 82.0% 70.4% 85.9%
1982 Lunde [49] CEA-Roche RiA (Roche) Always 3.5 75.0% 71.1% 45.8% 89.7%
1982 Steele Jr. [50] Ziconyl-phosphate method (Hansen Z-gel) No 5 68.8% 66.1% 46.2% 83.4%
1982 Tate [51] Double antibody radioimmunoassay (Egan/ No 40 Yes 58.8% 88.5% 46.5% 92.7%
Laurence)
1981 Gray [52] CEA-RIAKIT (Dainabot Radio Isotope Lab, Japan) No 7.1 Yes 66.7%
1981 Wedell [53] CEA-EIA (Abbott) No 5 Yes 17.4%
1980 Wood [54] Double antibody radioimmunoassay (Egan/ Increased values 25 97.3% 97.3% 92.3% 99.1%
Laurence)
1974 Sorokin [55] Radioimmunoassay (Thomson) and/or Ziconyl- Increased values 2.5 100.0% 91.2% 50.0% 100.0%
phosphate method (Hansen Z-gel)

Symbols for Table 2.

a
Only preoperative cut-off value is mentioned. Postoperative is assumed to be the same.
b
100 U/L corresponding to 10 ng/ml.

reference lists of the included studies, resulting in a total of 127 3.2. Study characteristics
publications assessed in full.
After appropriate exclusions were made, 42 studies were Article demographics, disease information, follow-up duration
included in this review. and study design are summarized in Table 1.
 C.G. Sørensen et al. / International Journal of Surgery 25 (2016) 134e144 141

Fig. 2. Forest plot from studies presenting outcome data on both specificity and sensitivity.

A total of 42 studies [14e55] comprising 16,249 patients were the additional 9 items, numbers 1e3 and 7e9 were included.
included. This gave a total of 9834 patients with CEA as an The populations examined were generally representative. Ob-
outcome, since some studies only reported CEA on recurred pa- jectives and cut-offs were usually decided in advance, and the
tients. The included articles consisted of 18 prospective definitions of a positive result were generally clear.
cohort studies, 21 retrospective cohort studies, and 3 randomized However, the studies differed regarding the reference standards
controlled trials. Recurrence rates ranged from 8 to 42% and 37 used, meaning the gold standard of clinical diagnostics. During
studies had recurrence rates between 20 and follow-up, the treating physician may be influenced by the
40% [15e20,22e31,33,34,36e38,40e50,52e54]. Mean follow-up knowledge of a CEA test. Blinding needed to prevent this influence
ranged from 16 to 80 months. Median follow-up ranged from was generally not performed, and only 23 studies reported the
14.5 to 72 months. The reference standard for detecting recurrence reference standard used. Furthermore, the quality and availability
was reported in 23 studies and 1 study used CEA as part of it. of diagnostic imaging has improved greatly over time. The tech-
nology of the index test has also improved. This is not taken into
3.3. Primary outcome measure account. The degree of instrument variation was generally not
considered among the studies.
The diagnostic properties of CEA are summarized in Table 2.
Data on sensitivity were reported in 41 studies, ranging from 4. Discussion
17.4% to 100% and from 50% to 80% in 30 studies. Two studies
showed a sensitivity of less than 40%, and both required CEA to be This systematic review included a total of forty-two studies
positive before detection of recurrence. Data on specificity were which reported a total of 9834 outcomes from CEA testing. The
reported in 23 studies, ranging from 66.1% to 98.4% and above 80% included studies generally showed sensitivities between 50% and
in 20 studies. Data on positive predictive value were reported in 24 80% and specificities and negative predictive values above 80%.
studies, ranging from 45.8% to 95.2% with great disparity between Results on positive predictive values ranged from 45.8% to 95.2%
results. This disparity, caused by a high number of false positives, and showed little reliability.
persisted among the studies performing double measurements. There are several limitations to this review. A considerable un-
Data on negative predictive value were reported in 23 studies certainty lies in the comparison between CEA and the gold standard
ranging from 74.5% to 100% and above 80% in 20 studies. Cut-off of clinical diagnostics as only 23 studies reported how recurrences
values for CEA ranged from 2.5 to 40 ng/ml, and a cut-off of 5 ng/ were verified. Moreover, diagnostic radiology has evolved greatly
ml was used in 27 studies. Two studies only reported a preoperative since the introduction of the CT scanner with improvements in
cut-off value, and the assumption was made that the same cut-off availability, scan time and resolution, as well as new imaging mo-
also applied for postoperative testing. The type of assay or dalities. These factors are of importance when evaluating the
analyzer used to measure CEA was reported in 28 studies, and these quality of reference standards and thereby CEA. If it takes longer to
differed considerably. Two studies performed double measure- diagnose a recurrence in the older studies, CEA will have more time
ments of each blood sample, and eight studies performed a to rise, and this will theoretically lead to an overestimation of
confirmatory measurement only when CEA was increased. sensitivity. Considering Table 2, this is possibly the reason for the
A forest plot of results from studies presenting outcome data on decrease seen in sensitivity over the years. The technology of the
both specificity and sensitivity, and the associated confidence in- CEA test has also improved. However, the clinical implications of
tervals, can be seen in Fig. 2. these developments are uncertain.
Additionally, only two studies blinded investigators from CEA
3.4. Methodological quality assessment values and 18 studies reported that CEA testing influenced the
further management of patients. If increased CEA values leads to
Methodological quality assessment is summarized in Table 3. Of more intensive diagnostics, sensitivity may be overestimated
Table 3
Quality Assessment of included studies using the QUADAS tool, modified by the Cochrane Collaboration.
 C.G. Sørensen et al. / International Journal of Surgery 25 (2016) 134e144 143

because the CEA negative patients are not investigated as thor- predictive value above 80%. Results on positive predictive value
oughly and some of these patients will have an undetected recur- ranged from 45.8% to 95.2% and showed low reliability.
rence. On the other hand, some of the false positive patients might The primary strength of CEA seems to be ability to confirm
develop signs of recurrence after the follow-up period, although negative results from other diagnostic examinations. Other
the majority of recurrences develop within the first three years strengths are its low cost, and simple interpretation.
after curative resection [2,3]. Because the follow-up periods were The impact of CEA as a supplement to CT-scan on mortality is
generally long and only eight studies reported a mean or median questionable, and therefore the clinical use of CEA may be
duration of less than 36 months, it is reasonable to assume that reconsidered.
most recurrences developed during this time.
Furthermore, half of the studies were retrospective in design, Ethical approval
and these are known to be less accurate than prospective studies
[56]. No Ethical Approval was given, since this is a Systematic Review.
Finally, only three studies investigated the degree of instrument
variation, which seems relevant when evaluating a test that is Sources of funding
prone to false positives and measures blood contents in minuscule
amounts. None.
The clinical utility of CEA as an early indicator of recurrence, is
conditioned by an impact on survival. However, a review examined Author contribution
surveillance strategies in relation to survival, after intended cura-
tive resection of colorectal cancer [6]. Of the five studies investi- Caspar G. Sørensen: 1 (conception and design, acquisition of
gating CEA, two studies (n ¼ 406) showed a significant survival data, analysis and interpretation), 2 (first draft), 3 (final approval).
advantage from CEA testing, and three studies (n ¼ 743) did not. A William K. Karlsson: 1 (acquisition of data, analysis and inter-
large randomized study has recently been published with 1211 pretation), 2 (critical revisions), 3 (final approval).
patients randomized to four main types of follow-up after intended Hans-Christian Pommergaard: 1 (conception and design, anal-
curative resection: CEA, CT, CEA þ CT, or minimal [7]. The study ysis and interpretation), 2 (critical revisions), 3 (final approval).
showed that CEA- and CT-monitored patients with recurrence were Jakob Burcharth: 1 (conception and design, analysis and inter-
more often amenable to curative surgery, compared to the minimal pretation), 2 (critical revisions), 3 (final approval).
follow-up group, but the difference in overall survival was not Jacob Rosenberg: 1 (conception and design, analysis and inter-
statistically significant. Combining CEA and CT was not superior to pretation), 2 (critical revisions), 3 (final approval).
CEA directed or CT directed follow-up alone, making CEA seemingly
redundant. There are several possible reasons for this questionable Conflicts of interest
impact on survival. First of all, the relatively low sensitivity. In order
to detect a recurrence in an early stage, a positive test result is None.
still needed. A high specificity is not of use here. Second, the low
incidence of curable recurrences in general makes an effect Guarantor
difficult to prove statistically. A third reason could be that
CEA seems to have a lower sensitivity for local recurrences Caspar Godthaab Sørensen.
[18,19,31,32,38e42,45,46,49,50] compared with metastases
[18,31,32,40e42,46,47,49]. These are fundamentally different dis-
Acknowledgements
eases and the failure to detect potentially curable local recurrences
will reduce the impact of the test. Conversely, CEA is mostly
None declared.
regarded as being relatively sensitive to hepatic metastases
[19,23,30e32,34,38,39,41,42,45e47,49,50], of which resection can
References
result in an improved prognosis [57].
A disadvantage of CEA, is the high rate of false positives. [1] J. Ferlay, I. Soerjomataram, M. Ervik, et al., GLOBOCAN 2012 v1.0, Cancer
Therefore, optimal use of CEA measurements requires consider- Incidence and Mortality Worldwide: IARC CancerBase No. 11 [Internet], In-
ations of factors and diseases known to influence results [8e10]. ternational Agency for Research on Cancer, Lyon, France, 2013. Available
from: http://globocan.iarc.fr.
Intuitively, confirmatory measurements when CEA is increased, [2] H. Kobayashi, H. Mochizuki, K. Sugihara, et al., Characteristics of recurrence
should also address this problem. However, the evidence for this is and surveillance tools after curative resection for colorectal cancer: a multi-
sparse. Another necessary consideration is whether the increase in center study, Surgery 141 (1) (2007) 67e75. Epub 2006 Sep. 14.
[3] M.J. O'Connell, M.E. Campbell, R.M. Goldberg, et al., Survival following
CEA is part of a fluctuating pattern, or if it rises progressively. The recurrence in stage II and III colon cancer: findings from the ACCENT data set,
primary strength of CEA seems to be its ability to confirm negative J. Clin. Oncol. 26 (14) (2008) 2336e2341.
results from other diagnostic examinations, even though metastatic [4] P. Gold, S.O. Freedman, Specific carcinoembryonic antigens of the human
digestive system, J. Exp. Med. 122 (3) (1965) 467e481.
disease can still occur without CEA elevation. This is thought to [5] L. Staib, K.H. Link, H.G. Beger, Follow-up in colorectal cancer: cost-
happen more often in poorly differentiated tumors, although the effectiveness analysis of established and novel concepts, Langenbecks Arch.
correlation is weak [58]. Other strengths of CEA are its low cost and Surg. 385 (6) (2000) 412e420.
[6] B.N. Fahy, Follow-up after curative resection of colorectal cancer, Ann. Surg.
simple interpretation. Oncol. 21 (3) (2014) 738e746.
[7] J.N. Primrose, R. Perera, A. Gray, et al., Effect of 3 to 5 years of scheduled CEA
5. Conclusion and CT follow-up to detect recurrence of colorectal cancer: the FACS ran-
domized clinical trial, JAMA 311 (3) (2014) 263e270.
[8] A.H. Chevinsky, CEA in tumors of other than colorectal origin, Semin. Surg.
Overall, the populations were largely representative and follow- Oncol. 7 (3) (1991) 162e166.
up duration long. However, an uncertainty may lie in the different [9] A. Ruibal Morell, CEA serum levels in non-neoplastic disease, Int. J. Biol.
reference standards in the studies, and thereby in the comparison Markers 7 (3) (1992) 160e166.
[10] I. Fukuda, M. Yamakado, H. Kiyose, Influence of smoking on serum carci-
between clinical diagnostics and CEA. Results point toward a noembryonic antigen levels in subjects who underwent multiphasic health
sensitivity between 50% and 80%, and a specificity and negative testing and services, J. Med. Syst. 22 (2) (1998) 89e93.
144 C.G. Sørensen et al. / International Journal of Surgery 25 (2016) 134e144

[11] http://www.crd.york.ac.uk/PROSPERO/display_record.asp? monitoring in curative surgery for recurrent colorectal carcinoma, Dis. Colon
ID¼CRD42014014005#.VSuhuPmsWT9. Assessed September 21, 2015. Rectum 40 (2) (1997) 145e149.
[12] D. Moher, A. Liberati, J. Tetzlaff, D.G. Altman, The PRISMA group, Preferred [34] E. Tobaruela, J.M. Enriquez, M. Diez, J. Camunas, J. Muguerza, J. Granell,
reporting items for systematic reviews and metaanalyses: the PRISMA Evaluation of serum carcinoembryonic antigen monitoring in the follow-up of
statement, PLoS Med. 6 (7) (2009) e1000097. Assessed May 15, 2015. colorectal cancer patients with metastatic lymph nodes and a normal pre-
[13] J.B. Reitsma, A.W.S. Rutjes, P. Whiting, V.V. Vlassov, M.M.G. Leeflang, J.J. Deeks, operative serum level, Int. J. Biol. Markers 12 (1) (1997) 18e21.
Chapter 9: assessing methodological quality, in: J.J. Deeks, P.M. Bossuyt, [35] W.F.A. Miles, J.D. Greig, J. Seth, C. Sturgeon, S.J. Nixon, Raised carcinoem-
C. Gatsonis (Eds.), Cochrane Handbook for Systematic Reviews of Diagnostic bryonic antigen level as an indicator of recurrent disease in the follow up of
Test Accuracy Version 1.0.0, The Cochrane Collaboration, 2009. Available patients with colorectal cancer, Br. J. Gen. Pract. 45 (395) (1995) 287e288.
from: http://srdta.cochrane.org/. [36] J.M. Chiang, J.Y. Wang, The value of CEA in the management of colorectal
[14] M. Shibutani, K. Maeda, H. Nagahara, et al., Significance of CEA and CA19-9 cancer, J. Surg. Assoc. ROC 27 (4) (1994) 2513e2521.
combination as a prognostic indicator and for recurrence monitoring in pa- [37] X. Filella, R. Molina, J.M. Pique, et al., Use of CA 19-9 in the early detection of
tients with stage II colorectal cancer, Anticancer Res. 34 (7) (2014) recurrences in colorectal cancer: comparison with CEA, Tumour Biol. 15 (1)
3753e3758. (1994) 1e6.
[15] W. Zhong, Z. Yu, J. Zhan, et al., Association of serum levels of CEA, CA199, [38] J.L. McCall, R.B. Black, C.A. Rich, et al., The value of serum carcinoembryonic
CA125, CYFRA21-1 and CA72-4 and disease characteristics in colorectal can- antigen in predicting recurrent disease following curative resection of colo-
cer, Pathol. Oncol. Res. 21 (1) (2015) 83e95. rectal cancer, Dis. Colon Rectum 37 (9) (1994) 875e881.
[16] H.S. Kim, M.R. Lee, Diagnostic accuracy of elevated serum carcinoembryonic [39] C.G. Moertel, T.R. Fleming, J.S. Macdonald, D.G. Haller, J.A. Laurie, C. Tangen, An
antigen for recurrence in postoperative stage II colorectal Cancer patients: evaluation of the carcinoembryonic antigen (CEA) test for monitoring patients
comparison with stage III, Ann. Coloproctol. 29 (4) (2013) 155e159. with resected colon cancer, JAMA 270 (8) (1993) 943e947.
[17] A. Fora, A. Patta, K. Attwood, G. Wilding, M. Fakih, Intensive radiographic and [40] Z. Zeng, A.M. Cohen, C. Urmacher, Usefulness of carcinoembryonic antigen
biomarker surveillance in stage II and III colorectal cancer, Oncology 82 (1) monitoring despite normal preoperative values in node-positive colon cancer
(2012) 41e47. patients, Dis. Colon Rectum 36 (11) (1993) 1063e1068.
[18] B.B. Su, H. Shi, J. Wan, Role of serum carcinoembryonic antigen in the [41] G.I. Meling, T.O. Rognum, O.P. Clausen, et al., Serum carcinoembryonic antigen
detection of colorectal cancer before and after surgical resection, World J. in relation to survival, DNA ploidy pattern, and recurrent disease in 406
Gastroenterol. 18 (17) (2012) 2121e2126. colorectal carcinoma patients, Scand. J. Gastroenterol. 27 (12) (1992)
[19] Z. Banaszkiewicz, P. Jarmocik, J. Frasz, K. Tojek, M. Mrozowski, A. Jawien , 1061e1068.
Usefulness of CEA concentration measurement and classic colonoscopy in [42] E. Seregni, E. Bombardieri, A. Bogni, F. Crippa, E. De Jager, G.L. Buraggi, The role
follow-up after radical treatment of colorectal cancer, Pol. Przegl Chir. 83 (6) of serum carcinoembryonic antigen (CEA) in the management of patients with
(2011) 310e318. colorectal carcinoma: the experience of the Istituto Tumori of Milan, Int. J.
[20] C.H. Chen, M.C. Hsieh, C.C. Lai, et al., Lead time of carcinoembryonic antigen Biol. Markers 7 (3) (1992) 167e170.
elevation in the postoperative follow-up of colorectal cancer did not affect the [43] J. Camun ~ as, J.M. Enriquez, J.M. Devesa, V. Morales, I. Millan, Value of follow-up
survival rate after recurrence, Int. J. Colorectal Dis. 25 (5) (2010) 567e571. in the management of recurrent colorectal cancer, Eur. J. Surg. Oncol. 17 (5)
[21] A. Holt, R.A. Nelson, L. Lai, Surveillance with serial serum carcinoembryonic (1991) 530e535.
levels detect colorectal cancer recurrences in patients who are initial non- [44] G.B. Secco, R. Fardelli, E. Campora, et al., CEA as a prognostic factor and early
secretors, Am. Surg. 76 (10) (2010) 1100e1103. indicator of recurrence in colorectal cancer, J. Exp. Clin. Cancer Res. 8 (3)
[22] T. Yakabe, Y. Nakafusa, K. Sumi, et al., Clinical significance of CEA and CA19-9 (1989) 173e177.
in postoperative follow-up of colorectal cancer, Ann. Surg. Oncol. 17 (9) [45] H. Graffner, B. Hultberg, B. Johansson, T. Mo €ller, B.G. Petersson, Detection of
(2010) 2349e2356. recurrent cancer of the colon and rectum, J. Surg. Oncol. 28 (2) (1985)
[23] I.J. Park, G.S. Choi, K.H. Lim, B.M. Kang, S.H. Jun, Serum carcinoembryonic 156e159.
antigen monitoring after curative resection for colorectal cancer: clinical [46] K.R. Hine, P.W. Dykes, Serum CEA testing in the post-operative surveillance of
significance of the preoperative level, Ann. Surg. Oncol. 16 (11) (2009) colorectal carcinoma, Br. J. Cancer 49 (6) (1984) 689e693.
3087e3093. [47] R.W. Beart Jr., M.J. O'Connell, Postoperative follow-up of patients with carci-
[24] M. Hara, Y. Kanemitsu, T. Hirai, K. Komori, T. Kato, Negative serum carci- noma of the colon, Mayo Clin. Proc. 58 (6) (1983) 361e363.
noembryonic antigen has insufficient accuracy for excluding recurrence from [48] U. Carlsson, J. Stewe nius, G. Ekelund, L. Leandoer, B. Nosslin, Is CEA analysis of
patients with dukes C colorectal cancer: analysis with likelihood ratio and value in screening for recurrences after surgery for colorectal carcinoma? Dis.
posttest probability in a follow-up study, Dis. Colon Rectum 51 (11) (2008) Colon Rectum 26 (6) (1983) 369e373.
1675e1680. [49] O.C. Lunde, O. Havig, Clinical significance of carcinoembryonic antigen (CEA)
[25] I. Grossmann, G.H. de Bock, W.M. Meershoek-Klein Kranenbarg, C.J. van de in patients with adenocarcinoma in colon and rectum, Acta Chir. Scand. 148
Velde, T. Wiggers, Carcinoembryonic antigen (CEA) measurement during (2) (1982) 189e193.
follow-up for rectal carcinoma is useful even if normal levels exist before [50] G. Steele Jr., S. Ellenberg, K. Ramming, et al., CEA monitoring among patients
surgery. A retrospective study of CEA values in the TME trial, Eur. J. Surg. in multi-institutional adjuvant G.I. therapy protocols, Ann. Surg. 196 (2)
Oncol. 33 (2) (2007) 183e187. (1982) 162e169.
[26] T. Irvine, M. Scott, C.I. Clark, A small rise in CEA is sensitive for recurrence after [51] H. Tate, Plasma CEA in the post-surgical monitoring of colorectal carcinoma,
surgery for colorectal cancer, Colorectal Dis. 9 (6) (2007) 527e531. Br. J. Cancer 46 (3) (1982) 323e330.
[27] H. Ko €rner, K. So € reide, P.J. Stokkeland, J.A. So€ reide, Diagnostic accuracy of [52] B.N. Gray, C. Walker, R. Barnard, Value of serial carcinoembryonic antigen
serum-carcinoembryonic antigen in recurrent colorectal cancer: a receiver determinations for early detection of recurrent cancer, Med. J. Aust. 1 (4)
operating characteristic curve analysis, Ann. Surg. Oncol. 14 (2) (2007) (1981) 177e178.
417e423. [53] J. Wedell, P. Meier zu Eissen, T.H. Luu, et al., A retrospective study of serial CEA
[28] L.C. Fernandes, S.B. Kim, S.S. Saad, D. Matos, Value of carcinoembryonic an- determinations in the early detection of recurrent colorectal cancer, Dis. Colon
tigen and cytokeratins for the detection of recurrent disease following cura- Rectum 24 (8) (1981) 618e621.
tive resection of colorectal cancer, World J. Gastroenterol. 12 (24) (2006) [54] C.B. Wood, J.G. Ratcliffe, R.W. Burt, A.J. Malcolm, L.H. Blumgart, The clinical
3891e3894. significance of the pattern of elevated serum carcinoembryonic antigen (CEA)
[29] M. Carpelan-Holmstro €m, J. Louhimo, U.H. Stenman, H. Alfthan, H. J€ arvinen, levels in recurrent colorectal cancer, Br. J. Surg. 67 (1) (1980) 46e48.
C. Haglund, CEA, CA 242, CA 19-9, CA 72-4 and hCGbeta in the diagnosis of [55] J.J. Sorokin, P.H. Sugarbaker, N. Zamcheck, M. Pisick, H.Z. Kupchik, F.D. Moore,
recurrent colorectal cancer, Tumour Biol. 25 (5e6) (2004) 228e234. Serial carcinoembryonic antigen assays. Use in detection of cancer recurrence,
[30] B. Engarås, Individual cutoff levels of carcinoembryonic antigen and CA 242 JAMA 228 (1) (1974) 49e53.
indicate recurrence of colorectal cancer with high sensitivity, Dis. Colon [56] A.M. Euser, C. Zoccali, K.J. Jager, F.W. Dekker, Cohort studies: prospective
Rectum 46 (3) (2003) 313e321. versus retrospective, Nephron Clin. Pract. 113 (3) (2009) c214e217.
[31] L.A. Carriquiry, A. Pin ~ eyro, Should carcinoembryonic antigen be used in the [57] R. Lupinacci, C. Penna, B. Nordlinger, Hepatectomy for resectable colorectal
management of patients with colorectal cancer? Dis. Colon Rectum 42 (7) cancer metastaseseindicators of prognosis, definition of resectability, tech-
(1999) 921e929. niques and outcomes, Surg. Oncol. Clin. N. Am. 16 (3) (2007), 493e506, vii-viii.
[32] G. Li Destri, S. Greco, C. Rinzivillo, A. Racalbuto, R. Curreri, A. Di Cataldo, [58] C.G. Moertel, J.R. O'Fallon, V.L. Go, M.J. O'Connell, G.S. Thynne, The preoper-
Monitoring carcinoembryonic antigen in colorectal cancer: is it still useful? ative carcinoembryonic antigen test in the diagnosis, staging, and prognosis of
Surg. Today 28 (12) (1998) 1233e1236. colorectal cancer, Cancer 58 (3) (1986) 603e610.
[33] P.A. Lucha Jr., L. Rosen, J.A. Olenwine, et al., Value of carcinoembryonic antigen