GENERAL THORACIC
Prognostic Significance of Positive Circumferential Resection Margin in Esophageal Cancer: A Systematic Review and Meta-Analysis Jie Wu, MD, Qi-Xun Chen, MD, Li-song Teng, MD, PhD, and Mark J. Krasna, MD Department of Surgical Oncology, the First Affiliated Hospital, Zhejiang University School of Medicine, Hangzhou, China; Department of Thoracic Surgery, Zhejiang Cancer Hospital, Hangzhou, China; and Meridian Cancer Care, Jersey Shore University Medical Center, Neptune, New Jersey
Background. To assess the prognostic significance of positive circumferential resection margin on overall survival in patients with esophageal cancer, a systematic review and meta-analysis was performed. Methods. Studies were identified from PubMed, EMBASE, and Web of Science. Survival data were extracted from eligible studies to compare overall survival in patients with a positive circumferential resection margin with patients having a negative circumferential resection margin according to the Royal College of Pathologists (RCP) criteria and the College of American Pathologists (CAP) criteria. Survival data were pooled with hazard ratios (HRs) and their corresponding 95% confidence intervals (CIs). A random-effects model metaanalysis on overall survival was performed. Results. The pooled HRs for survival were 1.510 (95% CI, 1.329–1.717; p < 0.001) and 2.053 (95% CI, 1.597–2.638;
p < 0.001) according to the RCP and CAP criteria, respectively. Positive circumferential resection margin was associated with worse survival in patients with T3 stage disease according to the RCP (HR, 1.381; 95% CI, 1.028–1.584; p [ 0.001) and CAP (HR, 2.457; 95% CI, 1.902–3.175; p < 0.001) criteria, respectively. Positive circumferential resection margin was associated with worse survival in patients receiving neoadjuvant therapy according to the RCP (HR, 1.676; 95% CI, 1.023–2.744; p [ 0.040) and CAP (HR, 1.847; 95% CI, 1.226–2.78; p [ 0.003) criteria, respectively. Conclusions. Positive circumferential resection margin is associated with poor prognosis in patients with esophageal cancer, particularly in patients with T3 stage disease and patients receiving neoadjuvant therapy.
E
two classifications of CRMþ in terms of prognostic significance have revealed contradictory results [11–17]. A meta-analysis [18] on this topic excluded two studies [19, 20], overlooked one [8], and did not include two recent studies [15, 17], in which the authors drew contrary conclusions about the impact of CRMþ on survival in patients with esophageal cancer. Furthermore, it used odds ratio calculated at a fixed point in time as a summary statistic in dealing with survival data, which did not take account of censored information, might be inefficient, and might lead to inappropriate conclusions [21, 22]. To assess the prognostic significance of CRM involvement, according to the RCP and CAP criteria, on overall survival in patients with esophageal cancer, a systematic review and meta-analysis of published original research was performed.
sophageal cancer is one of the most aggressive malignancies and has the sixth highest cancer mortality worldwide [1]. Common histologic characteristics including depth of invasion, lymph node involvement, and proximal and distal resection margin have been proved to be associated with long-term survival for esophageal cancer. Yetm the role of circumferential resection margin (CRM) on survival in patients with esophageal cancer is still under debate. Several studies have demonstrated the prognostic importance of CRM involvement [2–5], whereas others have not [6–8]. In addition, the definition of positive CRM (CRMþ) in esophageal cancer differs between two major schools of pathologists. The Royal College of Pathologists (RCP) [9] considers the CRM to be positive if the tumor is found within 1 mm of the surgical margin, whereas the College of American Pathologists (CAP) [10] defines a positive CRM as tumor found at the cut margin of resection. The optimal definition of CRMþ related to prognosis remains unclear. Studies comparing these
Accepted for publication Oct 11, 2013. Address correspondence to Dr Teng, Department of Surgical Oncology, the First Affiliated Hospital, Zhejiang University School of Medicine, 79 Qingchun Rd, Hangzhou, P.R. China 310003; e-mail: [email protected].
Ó 2014 by The Society of Thoracic Surgeons Published by Elsevier Inc
(Ann Thorac Surg 2014;97:446–53) Ó 2014 by The Society of Thoracic Surgeons
Material and Methods Search Strategy and Selection Criteria The PICO (Patients, Intervention, Comparison, Outcomes) format of clinical question was as follows: In patients with esophageal cancer after radical resection, whether positive CRM is associated with a worse prognosis compared with negative CRM. A systematic 0003-4975/$36.00 http://dx.doi.org/10.1016/j.athoracsur.2013.10.043
literature search was conducted in the electronic databases, including PubMed, EMBASE, and Web of Science. Search terms were “esophageal cancer and/or carcinoma or tumor or neoplasm” and “circumferential/radial/ lateral resection margin.” The searches were limited to articles published in English until April 2013. A manual cross-reference search of the bibliographies of relevant articles was performed to identify studies not found through the computerized search. To be eligible for inclusion, studies had to meet the following criteria: (1) Studies investigated the association between CRM status and prognosis in esophageal cancer patients. (2) All patients underwent potentially curative operations regardless of which type of surgical procedure was performed. (3) Hazard ratios (HRs) for overall survival according to CRM status had to be reported or could be calculated from the data presented. (4) When the same patient population was reported in more than one publication, only the one with the most relevant data or complete data was included. The exclusion criteria were review articles, case reports, conference abstracts, and studies that did not report outcomes of interest.
Quality Assessment and Data Extraction Two reviewers (J.W. and Q.C.) independently searched the literature, extracted data, and scored the eligible studies according to the Newcastle-Ottawa Scale (NOS) for cohort studies, which consists of three parameters of quality: selection, comparability, and outcome assessment [23]. A
WU ET AL CIRCUMFERENTIAL RESECTION MARGIN
447
score of at least 7 stars from a maximum of 9 was considered to represent a study of higher quality. Any disagreements were settled by a consensus reviewer (L.T.). The following data were extracted in a predefined form: first author, year of publication, origin country, study type, number of patients, age, sex, histology type, T stage, neoadjuvant therapy, patients’ stratification, incidence of CRMþ, follow-up time, and HR estimation. The authors of the primary studies were not contacted for additional or unreported information. Studies examining overall survival were pooled with studies examining cancer-specific survival, inasmuch as previous research has shown that there are few competing causes of death in these patients because of the highly fatal nature of esophageal cancer [1]. Five studies [14, 16, 24–26] included in this analysis stratified patients into three groups: a tumor found at the surgical margin, within 1 mm of the surgical margin, and at more than 1 mm of the surgical margin. To enable pooled analysis, the group with tumors within 1 mm was taken as a reference group. It was compared with the group with tumors found at more than 1 mm of the surgical margin according to RCP criteria, and with the group with tumors found at the surgical margin according to CAP criteria. To combine the data, a HR with a 95% confidence interval (CI) was used. When described in original articles, these values were obtained directly. If these statistical variables were not directly provided, the total number of events, the number of patients at risk in each group, and the Fig 1. Flow chart.
GENERAL THORACIC
Ann Thorac Surg 2014;97:446–53
GENERAL THORACIC a
UK
R
50
62
72
UK
P
135
64
UK
R
329
UK
R
UK
Histology Type
CRMþ (%) Neoadjuvant Patient T Stage Therapy (n) Stratification RCP CAP
Follow-up (mo) Median
Mean
HR est
Range
RCP/CAP
NOS Score
24–52
Log rank þ p
4
Report
6
SC
5
Reportb
7
NR
No
RCP
40.0
36
65
ADC, SCC, others ADC, SCC
T1–3
No
RCP
47.4
16
65
66
ADC, SCC
T1–3
No
RCP
20.3
249
64
78
T1–3
CT (34)
RCP
31.7
R
242
65
74
ADC, SCC, others ADC, SCC
T1–4
RCP
23.1
12–92
Report
5
Australia
R
240
62
78
ADC, SCC
T1–4
CT (142) CRT (9) CRT (124)
RCP
35.4
A: 1–120
Report
6
USA
R
135
60
85
T3
CT (59)
RCP, CAP
61.5
11.8
12–192
SC/SC
6
Holland
R
110
64
73
ADC, SCC, others ADC, SCC
T1–3
CT (31)
3 groups
38.2
15.5
Report/SC
6
UK
P
105
61
83
ADC
yPT0–4
CT (105)
RCP
36.2
A:26
Reportc
7
UK
R
320
NR
NR
T1–3
CT (121)
RCP
28.4
NR
Report
5
UK
R
314
63
79
T1–4
No
RCP
46.5
1–90
Report
4
Holland
R
98
65
78
T1–4
No
3 groups
47.9
25.5
4.7–124
SC/SC
5
Taiwan
R
151
58
96
ADC, SCC, others SCC
yPT3
CRT (151)
3 groups
51.0
17.2
SC/SC
8
Holland
R
132
63
85
ADC
T3
No
RCP, CAP
67.4
19.7
Report/Report
8
USA
R
160
60
92
ADC
yPT3
CRT (160)
RCP, CAP
26.3
5.0
Report/SC
7
UK
R
115
64
66
ADC, SCC
T1–4
CT (37)
3 groups
49.6
14.8
A: 36
SC/SC
5
UK
R
269
NR
77
ADC, SCC
T1–4
RCP
38.0a
4–157
Report
7
UK
R
232
62
76
3 groups
45.3
16.4
Ireland
R
157
63
69
ADC, SCC, yPT1–4 others ADC, SCC, T3 others
CT (124) CRT (42) CT (232) CRT (82)
RCP, CAP
60.0
18.0
CRM status reported in 258 patients.
b
ADC, SCC, others ADC, SCC
Multivariate analysis (Cox proportional hazard model) performed in 225 cases.
c
19 A: 60
D:14 A:70
37 NR
A: 12–84
37 43
50 28.4
2.3–212.3
NR
88 18 A:30 22
1.2–108 A: Report/SC 2.4–108 3–88 SC/SC
6 6
Multivariate analysis (Cox proportional hazard model) performed in 104 cases.
A ¼ length of follow-up referring to patients who were still alive; ADC ¼ adenocarcinoma; CAP ¼ College of American Pathologists; CRM ¼ circumferential resection margin; CRT ¼ chemoradiotherapy; CT ¼ chemotherapy; D ¼ length of follow-up referring to patients who were dead; HR est ¼ hazard ratio estimation; NOS ¼ Newcastle Ottawa Scale; NR ¼ not reported; P ¼ prospective; R ¼ retrospective; RCP ¼ Royal College of Pathologists; SC ¼ survival curve; SCC ¼ squamous cell carcinoma; 3 groups ¼ tumor cell found at, within, and at more than 1 mm of the surgical margin.
Ann Thorac Surg 2014;97:446–53
Sagar et al (1993) Dexter et al (2001) Khan et al (2003) Griffiths et al (2006) Sujendran et al (2008) Thompson et al (2008) Deeter et al (2009) Scheepers et al (2009) Saha et al (2009) Sillah et al (2009) Mirnezami et al (2010) Pultrum et al (2010) Chao et al (2011) Verhage et al (2011) Harvin et al (2012) Rao et al (2012) Reid et al (2012) Salih et al (2013) O’Farrell et al (2013)
Mean or Median Male Origin No. of Country Design Patients (n) Age (y) (%)
WU ET AL CIRCUMFERENTIAL RESECTION MARGIN
Study (year)
448
Table 1. Main Characteristics of Eligible Studies
WU ET AL CIRCUMFERENTIAL RESECTION MARGIN
449
Fig 2. (A) Forest plot of studies evaluating hazard ratio (HR) of positive circumferential resection margin (CRMþ) according to Royal College of Pathologists criteria. (B) Forest plot of studies evaluating HR of CRMþ according to College of American Pathologists criteria. (CI ¼ confidence interval.)
log-rank statistic or its P value were used to derive an approximate estimation of the HRs [21]. If only KaplanMeier curves were available, data were extracted from the survival plots and the HR was then estimated by use of the described methods [21, 22].
Statistical Analysis A test of heterogeneity of combined HRs was conducted using I2 statistics, which describe the percentage of variability in effect estimates caused by heterogeneity. Values of 25%, 50%, and 75% were regarded as reflecting low, moderate, and high
Table 2. Subgroup Analyses RCP
Total Histology type ADC SCC Cancer T stage T3 Other Neoadjuvant therapy Yes No Mix HR estimation KM Reported Quality Low High Country UK Others Reference Yes No
CAP
No. of Studies
HR (95% CI)
I2 (%)
p
No. of Studies
HR (95% CI)
I2 (%)
p
19
1.510 (1.329–1.717)
36.810
Prognostic Significance of Positive Circumferential Resection Margin in Esophageal Cancer: A Systematic Review and Meta-Analysis Jie Wu, MD, Qi-Xun Chen, MD, Li-song Teng, MD, PhD, and Mark J. Krasna, MD Department of Surgical Oncology, the First Affiliated Hospital, Zhejiang University School of Medicine, Hangzhou, China; Department of Thoracic Surgery, Zhejiang Cancer Hospital, Hangzhou, China; and Meridian Cancer Care, Jersey Shore University Medical Center, Neptune, New Jersey
Background. To assess the prognostic significance of positive circumferential resection margin on overall survival in patients with esophageal cancer, a systematic review and meta-analysis was performed. Methods. Studies were identified from PubMed, EMBASE, and Web of Science. Survival data were extracted from eligible studies to compare overall survival in patients with a positive circumferential resection margin with patients having a negative circumferential resection margin according to the Royal College of Pathologists (RCP) criteria and the College of American Pathologists (CAP) criteria. Survival data were pooled with hazard ratios (HRs) and their corresponding 95% confidence intervals (CIs). A random-effects model metaanalysis on overall survival was performed. Results. The pooled HRs for survival were 1.510 (95% CI, 1.329–1.717; p < 0.001) and 2.053 (95% CI, 1.597–2.638;
p < 0.001) according to the RCP and CAP criteria, respectively. Positive circumferential resection margin was associated with worse survival in patients with T3 stage disease according to the RCP (HR, 1.381; 95% CI, 1.028–1.584; p [ 0.001) and CAP (HR, 2.457; 95% CI, 1.902–3.175; p < 0.001) criteria, respectively. Positive circumferential resection margin was associated with worse survival in patients receiving neoadjuvant therapy according to the RCP (HR, 1.676; 95% CI, 1.023–2.744; p [ 0.040) and CAP (HR, 1.847; 95% CI, 1.226–2.78; p [ 0.003) criteria, respectively. Conclusions. Positive circumferential resection margin is associated with poor prognosis in patients with esophageal cancer, particularly in patients with T3 stage disease and patients receiving neoadjuvant therapy.
E
two classifications of CRMþ in terms of prognostic significance have revealed contradictory results [11–17]. A meta-analysis [18] on this topic excluded two studies [19, 20], overlooked one [8], and did not include two recent studies [15, 17], in which the authors drew contrary conclusions about the impact of CRMþ on survival in patients with esophageal cancer. Furthermore, it used odds ratio calculated at a fixed point in time as a summary statistic in dealing with survival data, which did not take account of censored information, might be inefficient, and might lead to inappropriate conclusions [21, 22]. To assess the prognostic significance of CRM involvement, according to the RCP and CAP criteria, on overall survival in patients with esophageal cancer, a systematic review and meta-analysis of published original research was performed.
sophageal cancer is one of the most aggressive malignancies and has the sixth highest cancer mortality worldwide [1]. Common histologic characteristics including depth of invasion, lymph node involvement, and proximal and distal resection margin have been proved to be associated with long-term survival for esophageal cancer. Yetm the role of circumferential resection margin (CRM) on survival in patients with esophageal cancer is still under debate. Several studies have demonstrated the prognostic importance of CRM involvement [2–5], whereas others have not [6–8]. In addition, the definition of positive CRM (CRMþ) in esophageal cancer differs between two major schools of pathologists. The Royal College of Pathologists (RCP) [9] considers the CRM to be positive if the tumor is found within 1 mm of the surgical margin, whereas the College of American Pathologists (CAP) [10] defines a positive CRM as tumor found at the cut margin of resection. The optimal definition of CRMþ related to prognosis remains unclear. Studies comparing these
Accepted for publication Oct 11, 2013. Address correspondence to Dr Teng, Department of Surgical Oncology, the First Affiliated Hospital, Zhejiang University School of Medicine, 79 Qingchun Rd, Hangzhou, P.R. China 310003; e-mail: [email protected].
Ó 2014 by The Society of Thoracic Surgeons Published by Elsevier Inc
(Ann Thorac Surg 2014;97:446–53) Ó 2014 by The Society of Thoracic Surgeons
Material and Methods Search Strategy and Selection Criteria The PICO (Patients, Intervention, Comparison, Outcomes) format of clinical question was as follows: In patients with esophageal cancer after radical resection, whether positive CRM is associated with a worse prognosis compared with negative CRM. A systematic 0003-4975/$36.00 http://dx.doi.org/10.1016/j.athoracsur.2013.10.043
literature search was conducted in the electronic databases, including PubMed, EMBASE, and Web of Science. Search terms were “esophageal cancer and/or carcinoma or tumor or neoplasm” and “circumferential/radial/ lateral resection margin.” The searches were limited to articles published in English until April 2013. A manual cross-reference search of the bibliographies of relevant articles was performed to identify studies not found through the computerized search. To be eligible for inclusion, studies had to meet the following criteria: (1) Studies investigated the association between CRM status and prognosis in esophageal cancer patients. (2) All patients underwent potentially curative operations regardless of which type of surgical procedure was performed. (3) Hazard ratios (HRs) for overall survival according to CRM status had to be reported or could be calculated from the data presented. (4) When the same patient population was reported in more than one publication, only the one with the most relevant data or complete data was included. The exclusion criteria were review articles, case reports, conference abstracts, and studies that did not report outcomes of interest.
Quality Assessment and Data Extraction Two reviewers (J.W. and Q.C.) independently searched the literature, extracted data, and scored the eligible studies according to the Newcastle-Ottawa Scale (NOS) for cohort studies, which consists of three parameters of quality: selection, comparability, and outcome assessment [23]. A
WU ET AL CIRCUMFERENTIAL RESECTION MARGIN
447
score of at least 7 stars from a maximum of 9 was considered to represent a study of higher quality. Any disagreements were settled by a consensus reviewer (L.T.). The following data were extracted in a predefined form: first author, year of publication, origin country, study type, number of patients, age, sex, histology type, T stage, neoadjuvant therapy, patients’ stratification, incidence of CRMþ, follow-up time, and HR estimation. The authors of the primary studies were not contacted for additional or unreported information. Studies examining overall survival were pooled with studies examining cancer-specific survival, inasmuch as previous research has shown that there are few competing causes of death in these patients because of the highly fatal nature of esophageal cancer [1]. Five studies [14, 16, 24–26] included in this analysis stratified patients into three groups: a tumor found at the surgical margin, within 1 mm of the surgical margin, and at more than 1 mm of the surgical margin. To enable pooled analysis, the group with tumors within 1 mm was taken as a reference group. It was compared with the group with tumors found at more than 1 mm of the surgical margin according to RCP criteria, and with the group with tumors found at the surgical margin according to CAP criteria. To combine the data, a HR with a 95% confidence interval (CI) was used. When described in original articles, these values were obtained directly. If these statistical variables were not directly provided, the total number of events, the number of patients at risk in each group, and the Fig 1. Flow chart.
GENERAL THORACIC
Ann Thorac Surg 2014;97:446–53
GENERAL THORACIC a
UK
R
50
62
72
UK
P
135
64
UK
R
329
UK
R
UK
Histology Type
CRMþ (%) Neoadjuvant Patient T Stage Therapy (n) Stratification RCP CAP
Follow-up (mo) Median
Mean
HR est
Range
RCP/CAP
NOS Score
24–52
Log rank þ p
4
Report
6
SC
5
Reportb
7
NR
No
RCP
40.0
36
65
ADC, SCC, others ADC, SCC
T1–3
No
RCP
47.4
16
65
66
ADC, SCC
T1–3
No
RCP
20.3
249
64
78
T1–3
CT (34)
RCP
31.7
R
242
65
74
ADC, SCC, others ADC, SCC
T1–4
RCP
23.1
12–92
Report
5
Australia
R
240
62
78
ADC, SCC
T1–4
CT (142) CRT (9) CRT (124)
RCP
35.4
A: 1–120
Report
6
USA
R
135
60
85
T3
CT (59)
RCP, CAP
61.5
11.8
12–192
SC/SC
6
Holland
R
110
64
73
ADC, SCC, others ADC, SCC
T1–3
CT (31)
3 groups
38.2
15.5
Report/SC
6
UK
P
105
61
83
ADC
yPT0–4
CT (105)
RCP
36.2
A:26
Reportc
7
UK
R
320
NR
NR
T1–3
CT (121)
RCP
28.4
NR
Report
5
UK
R
314
63
79
T1–4
No
RCP
46.5
1–90
Report
4
Holland
R
98
65
78
T1–4
No
3 groups
47.9
25.5
4.7–124
SC/SC
5
Taiwan
R
151
58
96
ADC, SCC, others SCC
yPT3
CRT (151)
3 groups
51.0
17.2
SC/SC
8
Holland
R
132
63
85
ADC
T3
No
RCP, CAP
67.4
19.7
Report/Report
8
USA
R
160
60
92
ADC
yPT3
CRT (160)
RCP, CAP
26.3
5.0
Report/SC
7
UK
R
115
64
66
ADC, SCC
T1–4
CT (37)
3 groups
49.6
14.8
A: 36
SC/SC
5
UK
R
269
NR
77
ADC, SCC
T1–4
RCP
38.0a
4–157
Report
7
UK
R
232
62
76
3 groups
45.3
16.4
Ireland
R
157
63
69
ADC, SCC, yPT1–4 others ADC, SCC, T3 others
CT (124) CRT (42) CT (232) CRT (82)
RCP, CAP
60.0
18.0
CRM status reported in 258 patients.
b
ADC, SCC, others ADC, SCC
Multivariate analysis (Cox proportional hazard model) performed in 225 cases.
c
19 A: 60
D:14 A:70
37 NR
A: 12–84
37 43
50 28.4
2.3–212.3
NR
88 18 A:30 22
1.2–108 A: Report/SC 2.4–108 3–88 SC/SC
6 6
Multivariate analysis (Cox proportional hazard model) performed in 104 cases.
A ¼ length of follow-up referring to patients who were still alive; ADC ¼ adenocarcinoma; CAP ¼ College of American Pathologists; CRM ¼ circumferential resection margin; CRT ¼ chemoradiotherapy; CT ¼ chemotherapy; D ¼ length of follow-up referring to patients who were dead; HR est ¼ hazard ratio estimation; NOS ¼ Newcastle Ottawa Scale; NR ¼ not reported; P ¼ prospective; R ¼ retrospective; RCP ¼ Royal College of Pathologists; SC ¼ survival curve; SCC ¼ squamous cell carcinoma; 3 groups ¼ tumor cell found at, within, and at more than 1 mm of the surgical margin.
Ann Thorac Surg 2014;97:446–53
Sagar et al (1993) Dexter et al (2001) Khan et al (2003) Griffiths et al (2006) Sujendran et al (2008) Thompson et al (2008) Deeter et al (2009) Scheepers et al (2009) Saha et al (2009) Sillah et al (2009) Mirnezami et al (2010) Pultrum et al (2010) Chao et al (2011) Verhage et al (2011) Harvin et al (2012) Rao et al (2012) Reid et al (2012) Salih et al (2013) O’Farrell et al (2013)
Mean or Median Male Origin No. of Country Design Patients (n) Age (y) (%)
WU ET AL CIRCUMFERENTIAL RESECTION MARGIN
Study (year)
448
Table 1. Main Characteristics of Eligible Studies
WU ET AL CIRCUMFERENTIAL RESECTION MARGIN
449
Fig 2. (A) Forest plot of studies evaluating hazard ratio (HR) of positive circumferential resection margin (CRMþ) according to Royal College of Pathologists criteria. (B) Forest plot of studies evaluating HR of CRMþ according to College of American Pathologists criteria. (CI ¼ confidence interval.)
log-rank statistic or its P value were used to derive an approximate estimation of the HRs [21]. If only KaplanMeier curves were available, data were extracted from the survival plots and the HR was then estimated by use of the described methods [21, 22].
Statistical Analysis A test of heterogeneity of combined HRs was conducted using I2 statistics, which describe the percentage of variability in effect estimates caused by heterogeneity. Values of 25%, 50%, and 75% were regarded as reflecting low, moderate, and high
Table 2. Subgroup Analyses RCP
Total Histology type ADC SCC Cancer T stage T3 Other Neoadjuvant therapy Yes No Mix HR estimation KM Reported Quality Low High Country UK Others Reference Yes No
CAP
No. of Studies
HR (95% CI)
I2 (%)
p
No. of Studies
HR (95% CI)
I2 (%)
p
19
1.510 (1.329–1.717)
36.810
