Ir J Med Sci DOI 10.1007/s11845-013-1051-6
ORIGINAL ARTICLE
Diagnostic values of carcinoembryonic antigen in predicting peritoneal recurrence after curative resection of gastric cancer: a meta-analysis Y. Xiao • J. Zhang • X. He • J. Ji • G. Wang
Received: 16 October 2013 / Accepted: 3 December 2013 Ó Royal Academy of Medicine in Ireland 2013
Abstract Aim A meta-analysis was performed to assess the diagnostic values of carcinoembryonic antigen (CEA) in predicting the peritoneal recurrence after curative resection of gastric cancer. Methods The Medline, Embase, Web of Science, Ovid and Cochrane databases, Google Scholar and Vivisimo engines were searched to identify studies reporting on the accuracy of CEA protein or CEA mRNA in predicting the postoperative peritoneal recurrence of gastric cancer. Publication bias was demonstrated by Funnel plots and Egger test. The sensitivity, specificity, and diagnostic odds ratio (DOR) were calculated and summary receiver operating characteristic curves were generated. Results Seven and eight studies fulfilled the inclusion criteria for CEA protein and mRNA determination, including 635 and 849 patients, respectively. The pooled sensitivity, specificity and DOR of CEA protein for predicting the peritoneal recurrence were 0.77 (95 % CI 0.69–0.84), 0.89 (95 % CI 0.86–0.92), 29.71 (95 % CI 10.27–85.92), respectively. Similarly, the values for CEA
Y. Xiao J. Ji G. Wang (&) Department of Gastrointestinal Surgery, Union Hospital, Tongji Medical College, HuaZhong University of Science and Technology, Jiefang Ave 1277#, Wuhan 430022, Hubei, China e-mail: [email protected] J. Zhang Department of Emergency Surgery, Union Hospital, Tongji Medical College, HuaZhong University of Science and Technology, Wuhan 430022, China X. He Department of Obstetrics and Gynecology, Union Hospital, Tongji Medical College, HuaZhong University of Science and Technology, Wuhan 430022, China
mRNA were 0.82 (95 % CI 0.75–0.88), 0.82 (95 % CI 0.79–0.85) and 22.97 (95 % CI 10.90–48.41). Meanwhile, the sensitivity and DOR of CEA protein or mRNA were higher than those of cytology, while higher specificity was noted in cytology assay. Conclusion CEA protein and mRNA levels in peritoneal lavage show a high diagnostic accuracy and may help accurately predict the peritoneal recurrence after curative resection of gastric cancer. Keywords Meta-analysis Carcinoembryonic antigen Peritoneal recurrence Gastric cancer Diagnosis Peritoneal lavage cytology
Introduction Gastric cancer is the second most common cause of cancer death worldwide. Despite of recent advances in surgical techniques and adjuvant chemotherapies, the prognosis of advanced gastric cancer still remains poor [1]. Peritoneal metastasis, the most frequent type of recurrence in patients without evidence of residual tumor at the time of surgery, is a significant obstacle in the curative treatment of gastric cancer [2]. Traditionally, risk factors for peritoneal recurrence include the size of serosal invasion [3], the status of lymph node involvement [4], and histologically diffuse type [5]. However, some patients with tumors involving the serosal surface (T3 and T4 according to the TMN classification) do not present evidences of peritoneal recurrence, even without adjuvant chemotherapy [6]. Thus, tumor class alone is not a reliable method for predicting the peritoneal recurrences of gastric cancer [7]. A simple and accurate method to detect the residual cancer cells still needs to be developed.
123
Ir J Med Sci
Intraoperative detection of free cancer cells in the peritoneal cavity appears to be predictive of peritoneal recurrence in gastric cancer patients [8–10]. Cytological examination of lavage fluid obtained at the time of surgery is a conventional method to detect free cancer cells in the peritoneal space. However, the sensitivity of peritoneal lavage cytology (PLC) is relatively low, ranging from 22 to 30 % in gastric carcinoma involving the serosa [11, 12], making it less valuable in determining the postoperative prognosis. Carcinoembryonic antigen (CEA) is a wellknown tumor marker of gastrointestinal tumor, and has been applied to detect a small number of adenocarcinoma cells in the blood [13]. However, the serum CEA level does not elevate until the tumor is enlarged and unresectable, resulting in its limited diagnostic value [14]. Recently, measurement of CEA protein and mRNA level in peritoneal lavage has been used to detect the existence of free cancer cells in the peritoneal cavity [15, 16]. However, due to the limited sample size and varied accuracy in the literature, the overall diagnostic values of CEA in predicting the peritoneal recurrence of gastric cancer still need to be addressed. Therefore, our major interest has been focused on the ultimate benefit of CEA measurement in peritoneal washings, and performed the present meta-analysis to assess its values in predicting the postoperative peritoneal recurrence of gastric cancer.
Inclusion and exclusion criteria Studies relevant to the diagnostic values of CEA in predicting the peritoneal recurrence after curative resection of gastric cancer were included only if they fulfilled the following criteria: (1) CEA protein or mRNA in peritoneal lavage was used to predict the peritoneal recurrence after the curative resection of histologically proved gastric cancer. (2) Clinical diagnosis of peritoneal recurrence after surgery was used as reference standard. (3) On the patient basis, sufficient data were available to calculate the true positive (TP), true negative (TN), false positive (FP) and false negative (FN). TP were defined as the patients that had CEA expression in intraoperative peritoneal lavage and postoperative peritoneal recurrence. TN findings were patients who had no evidence of intraoperative CEA expression and postoperative peritoneal recurrence. FP findings were defined as patients with intraoperative CEA expression but not postoperative peritoneal recurrence. FN findings were failures to identify intraoperative CEA expression, with postoperative peritoneal recurrence. Conference abstracts were excluded because of the limited data presented in them. Although no language restrictions were imposed initially, the search was limited to studies published in the English language for the full-text review and final analysis. Data extraction
Materials and methods Literature search We searched the online databases, including Medline (using PubMed as the search engine), Embase, Web of Science, Ovid and Cochrane, to identify suitable studies documenting the measurement of CEA in predicting the peritoneal recurrence after curative resection of gastric cancer between 1995 and the end of May 2011. The search was also made using GoogleTM Scholar, and VivisimoTM search engine. The terms ‘‘carcinoembryonic antigen’’, ‘‘peritoneal lavage’’, ‘‘peritoneal washing’’, ‘‘gastric cancer’’, ‘‘gastric carcinoma’’, ‘‘sensitivity’’ and ‘‘specificity’’ were used as Mesh search headings. The above terms and their combinations were also searched as text words. All scanned abstracts, studies, and citations were reviewed. Moreover, references of the retrieved manuscripts were also manually cross-searched for further relevant publications. The review was limited to the published studies and no attempt was made to contact the authors to obtain unpublished data or unclear information in the studies. All searches were limited to ‘‘humans’’. The latest eligible study for the above search period was July 31, 2007.
123
The full article was retrieved when they met the inclusion criteria. Two reviewers (LDZ and QST) independently evaluated and extracted data from potentially relevant studies. Data regarding the following factors were considered: first author, publication year, country origin of the study, study population characteristics, study design, methods to measure CEA, cut-off values, numbers of TP, TN, FP and FN observations. For analysis of the outcomes of interest, any disagreements between the reviewers were discussed in committee and a consensus was reached. Quality assessment of included studies The methodological quality of the selected studies was graded by two reviewers (LDZ and QST), with the standards for reporting diagnostic accuracy (STARD) [17] and quality assessment of diagnostic accuracy studies (QUADAS) tools [18]. Each diagnostic study was scored on the basis of the 25 STARD and 14 QUADAS criteria [17, 18]. The STARD or QUADAS scoring system awards one point if a specific criterion is met; zero points if a criterion is unclear, and one point is deducted if a criterion is not fulfilled. Disagreement among the two reviewers (LDZ and QST) was discussed in committee and a consensus was reached. The Pearson correlation coefficient was used to
Ir J Med Sci
find a correlation between the two reviewers. The value for a Pearson’s can fall between 0.00 (no correlation) and 1.00 (perfect correlation). Both reviewers demonstrated significant correlation (Pearson’s r = 0.94). Statistical analysis Recommended standard methods were used for metaanalyses of diagnostic test evaluations [19]. Analyses were performed using two statistical software programs, Stata (version 11.0; Stata Corporation) and Meta-DiSc (version 1.4; XI Cochrane Colloquium). For each study, the sensitivity, specificity, positive likelihood ratio (PLR), negative likelihood ratio (NLR) and diagnostic odds ratio (DOR) were computed to measure the test accuracy. A randomeffects model was used to calculate the average sensitivity, specificity and other measures across studies. The presence of publication bias, if any, was evaluated using graphical exploration with funnel plots and the Egger test [20].
Results Quality of reporting and study characteristics We identified 52 potential studies published between 1995 and 2011 that might match the selection criteria, of which 13 were excluded due to review article (n = 1), no detailed findings (n = 2), and conference abstracts (n = 10). Thirty-nine articles were then retrieved for full-text review, of which 25 were excluded due to insufficient data (n = 8), no prediction of peritoneal recurrence (n = 10), combined analysis with cytokeratin 20 (n = 2), duplicate studies by the same author (n = 4), and inconsistent data (n = 1). Fourteen studies had sufficient data for the calculation of a summary sensitivity and specificity value [14, 15, 21–32], and were included in the meta-analysis. Of the included studies, 92.9 % (13/14) were published in 2000–2007, and 7.1 % (1/14) were published in 1995 (Table 1). The majority of studies (78.6 %) were performed in the Japan (11/14), whereas 14.3 % (2/14) were performed in China (mainland and Taiwan), and 7.1 % (1/ 14) in Turkey (Table 1). Seven and eight studies fulfilled the inclusion criteria for CEA protein and mRNA determination, including 635 and 849 patients, respectively (Tables 2, 3). In all studies included in the meta-analysis, the diagnoses of patients with peritoneal recurrence were made based on clinical findings, supported by ultrasonography, computed tomography or magnetic resonance imaging. Radiometric immunoassay, enzyme immunoassay or sphere turbidimetric assay were used to detect CEA protein in seven studies [14, 15, 21, 27, 28, 31, 32], whereas CEA mRNA in peritoneal lavage was measured by
reverse-transcriptase polymerase chain reaction (RT-PCR), real-time RT-PCR or transcription–reverse transcription concerted reaction (TRC) in eight studies [22–26, 29–31]. The clinical characteristics of these studies, along with STARD and QUADAS scores, were outlined in Table 1. In all, the quality of study design and reporting diagnostic accuracy of most studies were good, as all studies had higher QUADAS scores ([10), and only two studies had lower STARD scores (\13). Diagnostic accuracy of CEA protein Seven studies documented the sensitivity and specificity of CEA protein in prediction of the peritoneal recurrence of gastric cancer (Table 2) [14, 15, 21, 27, 28, 31, 32]. As shown in Fig. 1, the sensitivity ranged from 0.50 to 1.00 (pooled 0.77, 95 % CI 0.69–0.84), whereas specificity ranged from 0.73 to 0.95 (pooled 0.89, 95 % CI 0.86–0.92); it was noted that PLR was 6.89 (95 % CI 4.00–11.85), NLR was 0.29 (95 % CI 0.16–0.54) and DOR was 29.71 (95 % CI 10.27–85.92). The summary receiver operating characteristic (SROC) curve showed that the maximum joint sensitivity and specificity was 0.8699 (SEM 0.038); whereas the area under the curve (AUC) was 0.934 (SEM 0.031), indicating that the level of overall accuracy of CEA protein was high (Fig. 2). In addition, five included studies compared the sensitivity and specificity of CEA protein and PLC in predicting the peritoneal recurrence of gastric cancer [15, 21, 27, 28, 31]. Analysis of the pooled data revealed that the sensitivity of CEA protein and PLC was 0.72 (95 % CI 0.60–0.82) and 0.45 (95 % CI 0.33–0.57), respectively, whereas their specificity was 0.88 (95 % CI 0.84–0.92) and 0.92 (95 % CI 0.89–0.95). Diagnostic accuracy of CEA mRNA Eight studies reported the sensitivity and specificity of CEA mRNA in the prediction of peritoneal recurrence of gastric cancer (Table 3) [22–26, 29–31]. As shown in Fig. 1, the sensitivity ranged from 0.65 to 1.00 (pooled 0.82, 95 % CI 0.75–0.88), whereas specificity ranged from 0.65 to 0.90 (pooled 0.82, 95 % CI 0.79–0.85); meanwhile, the PLR was 4.59 (95 % CI 3.30–6.37), NLR was 0.22 (95 % CI 0.12–0.40) and DOR was 22.97 (95 % CI 10.90–48.41). The SROC curve showed that the maximum joint sensitivity and specificity was 0.8221 (SEM 0.024); whereas the area under the curve (AUC) was 0.891 (SEM 0.023), indicating the high level of overall accuracy of CEA mRNA (Fig. 2). In the included studies, five articles compared the sensitivity and specificity of CEA mRNA and PLC in the prediction of peritoneal recurrence of gastric cancer [22, 26, 29–31]. The pooled analysis
123
Ir J Med Sci Table 1 Characteristics of included studies Author
Year
Country
Duration
Design
Quality score STARD
QUADAS 12
Nishiyama
1995
Japan
NA
Prospective
11
Tsutsumi
2000
Japan
1991–1996
Prospective
12
11
Abe
2001
Japan
1994
Prospective
16
13
Fujii
2002
Japan
1995–1998
Prospective
15
13
Ueno
2003
Japan
1998–2001
Prospective
17
13
Tokuda
2003
Japan
1997–1999
Prospective
15
13
Ishii Wang
2004 2005
Japan Taiwan
1999–2002 2001–2002
Prospective Prospective
14 17
14 12
Ito
2005
Japan
1995–1999
Prospective
14
14
Cetin
2005
Turkey
1998–2001
Prospective
16
12
Li
2005
China
1995–1997
Prospective
17
12
Kodera
2006
Japan
1995–1999
Prospective
17
13
Katsuragi
2007
Japan
NA
Prospective
14
12
Yamamoto
2007
Japan
1996–2001
Prospective
14
12
STARD standards for reporting diagnostic accuracy, QUADAS quality assessment for studies of diagnostic accuracy, NA non-available
Table 2 Summary of included studies of CEA protein in predicting peritoneal recurrence of gastric cancer Author
No. of patients
Cut-off
CEA protein TP
Nishiyama Tsutsumi
Cytology
FP
FN
TN
TP
FP
FN
TN
118
100 ng/g
10
5
4
99
2
1
12
103
60
100 ng/g
18
3
0
39
NA
NA
NA
NA
Abe
54
210 ng/g
6
3
1
44
3
0
4
47
Cetin Li
70 64
10 ng/g 100 ng/g
11 18
14 8
8 1
37 37
9 14
17 1
10 5
34 44
40
200 ng/g
6
2
6
26
4
2
8
26
229
0.4 ng/ml
29
21
9
170
NA
NA
NA
NA
FP
FN
TN
Wang Yamamoto
TP true positive, FP false positive, FN false negative, TN true negative, NA non-available Table 3 Summary of included studies of CEA mRNA in predicting peritoneal recurrence of gastric cancer Author
No. of patients
Cut-off
CEA mRNA
Cytology
TP
FP
FN
TN
TP
Fujii
49
10 cells/S
15
12
0
22
5
1
10
33
Ueno
78
27.11a
10
21
3
44
3
1
10
64
10 cells/S
15
15
1
105
5
0
11
120
9
5
0
46
NA
NA
NA
NA
8 11
3 9
4 2
25 64
4 NA
2 NA
8 NA
26 NA
32
46
4
202
17
8
19
240
24
14
13
65
NA
NA
NA
NA
Tokuda
136
Ishii
60
1.2
Wang Ito
40 86
10 cells/S 0.1 cells/S
Kodera
284
0.1 cells/S
Katsuragi
116
2.490 910
-2b
TP true positive, FP false positive, FN false negative, TN true negative, S sample, NA non-available a
Crossing-point value
b
CEA/GAPDH ratio
123
Ir J Med Sci
Fig. 1 Forest plot of the sensitivity and specificity of CEA protein or mRNA in the prediction of peritoneal recurrence of gastric cancer. The point estimates of sensitivity and specificity from each study were shown as solid circles. Error bars indicate 95 % confidence intervals
Fig. 2 Summary receiver operating characteristic curves for CEA protein or mRNA. Each study included in the meta-analysis was represented by the solid circle, with the size indicated by the size of the solid circle
revealed that the sensitivity of CEA mRNA and PLC was 0.87 (95 % CI 0.78–0.93) and 0.37 (95 % CI 0.27–0.48), respectively, whereas their specificity was 0.80 (95 % CI 0.77–0.84) and 0.98 (95 % CI 0.96–0.99).
Publication bias Evaluation of publication bias for CEA protein and mRNA showed that the Egger test was not significant (P = 0.193,
123
Ir J Med Sci
Fig. 3 Funnel graph for the assessment of potential publication bias in CEA protein or mRNA determination. Each study included in the meta-analysis was represented by the solid circle. Funnel graph plots
the log of the diagnostic odds ratio (DOR) against the SE of the log of the DOR (an indicator of sample size). The line in the center indicates the summary DOR
P = 0.07, respectively). The funnel plots for publication bias did not show obvious asymmetry (Fig. 3). In addition, meta-regression analysis of the STARD and QUADAS scores indicated no effects of study quality on relative DOR of CEA protein (P = 0.645 and P = 0.251) or CEA mRNA (P = 0.492 and P = 0.227) in predicting the peritoneal recurrence of gastric cancer.
of gastric cancer. In 1991, Asao et al. [37] first reported that CEA protein levels in peritoneal washings can be a useful guide to peritoneal recurrence of gastric cancer. A series of studies demonstrated that CEA protein levels in peritoneal washings could be more responsive to the presence of peritoneal dissemination than PLC, with low falsepositive rate [14, 15, 21, 28]. In 64 patients with gastric cancer, all the 15 patients with PLC positive findings were CEA protein positive, while 11 of 49 patients with PLC negative findings were also CEA positive [27]. We further analyzed five studies comparing the sensitivity and specificity between CEA protein and PLC [22, 26, 29–31], and the results indicated that CEA protein determination was more sensitive (0.72 versus 0.45) but less specific (0.88 versus 0.92) than PLC assay, suggesting that CEA protein level in peritoneal washings is of potential value as an indicator of peritoneal recurrence [38]. As a more sensitive method than conventional PLC, the polymerase chain reaction (PCR) method has recently been developed to detect CEA in peritoneal washing fluid [39]. In general, almost all gastric cancer cells are considered to have and secrete CEA mRNA [40]. Nakanishi et al. [40] reported a higher sensitivity of CEA RT-PCR assay (31 %) over routine PLC (10 %) in the prediction of peritoneal recurrence; combining PLC and CEA RT-PCR assays improved the sensitivity rate up to 57 % [41]. In another study, 93.7 % patients (15/16) with peritoneal recurrence after surgery were positive for CEA mRNA, but only 31.2 % (5/16) were positive by PLC examination [29]. Thus, CEA RT-PCR assay may serve as a useful method for predicting the peritoneal recurrence of gastric cancer patients. Quantitative measurement of CEA mRNA with subsequent automatic data analysis can be completed well within 3 h [42]. Nakanishi et al. [43] also reported that the results obtained from real-time RT-PCR were essentially
Discussion Peritoneal lavage cytology (PLC) is the gold standard for assessing the presence or absence of peritoneal dissemination of gastric cancer. However, its sensitivity is relatively low in gastric cancer involving the serosa [33, 34]. Some patients with cytologically negative peritoneal washing develop peritoneal recurrence of gastric cancer after surgery [35]. In recent years, novel diagnostic tools, including immunohistochemical and molecular biological methods, have been developed to detect free cancer cells [33, 36]. Although several investigators have reported that immunohistochemistry with antibody panels can be an aid to conventional cytology [33, 36], the sensitivity still needs to be increased. In addition, these diagnostic techniques are time-consuming and laborious compared to conventional PLC [16]. Therefore, there is a need for a more accurate test to early detect the peritoneal metastasis of gastric cancer. In the present meta-analysis, our results indicate that determining CEA protein levels in peritoneal lavage is sensitive (0.77, 95 % CI 0.69–0.84), suggesting that CEA determination is feasible to predict the peritoneal recurrence of gastric cancer after surgery. Specificity was also high (0.89, 95 % CI 0.86–0.92), and this was very helpful in confirming peritoneal recurrence after curative resection
123
Ir J Med Sci
the same as those obtained by conventional RT-PCR. However, some CEA secreting cells found in peritoneal washing fluids, such as macrophages, leukocytes and necrotic cancer cells, may alter the sensitivity of PCR results, resulting in false positivity [42, 44]. In this metaanalysis, we noted the sensitivity (0.82, 95 % CI 0.75–0.88) and specificity (0.82, 95 % CI 0.79–0.85) of CEA mRNA determination were suitable for the prediction of peritoneal recurrence of gastric cancer. Similarly, higher sensitivity (0.87 versus 0.37) and lower specificity (0.80 versus 0.98) were noted for CEA mRNA determination than PLC in five included studies [22, 26, 29–31]. Based on the comprehensive search strategy, including screening, study selection and quality assessment, the quality of study design and reporting diagnostic accuracy of most studies were good. To present a global summary of test performance, we further applied the SROC curve to show the trade-off between sensitivity and specificity. The results showed that the maximum joint sensitivity and specificity of CEA protein and mRNA determination was 0.8699 and 0.8221, respectively, with AUC being 0.934 and 0.891. In addition, as an indicator of test accuracy that combines data from sensitivity and specificity into a single number [45], DOR was applied to indicate better discriminatory test performance. We found that the DOR values of CEA protein and mRNA determination were 29.71 and 22.97, respectively, further conforming that their roles in the prediction of peritoneal recurrence of gastric cancer. The present meta-analysis had some limitations. First, although obvious publication bias was not noted by funnel plot and Egger test, the conference abstracts and nonEnglish language studies were excluded, resulting in potential publication bias. Second, due to the lack of sufficient data in the original publications, it was not possible to compare the accuracy of CEA protein and mRNA determination in the prediction of peritoneal recurrence of gastric cancer. Third, there were different methods to measure CEA protein and CEA mRNA. Due to the limited study number, it was not possible to analyze the effects of laboratory infrastructure, expertise with assay technology and patient spectrum. Thus, a series of studies with larger case number and standard methods are necessary to further explore the accuracy of CEA protein and mRNA. In conclusion, CEA protein and mRNA levels in peritoneal lavage show a high diagnostic accuracy and may help accurately predict the peritoneal recurrence after curative resection of gastric cancer. The sensitivity and diagnostic odds ratio of CEA protein or mRNA are higher than those of PLC, while higher specificity of PLC is noted. Combining determination of CEA and PLC may further improve the diagnostic accuracy for predicting the peritoneal recurrence of gastric cancer.
Acknowledgments This study was funded by Natural Science Fund Project of Hubei China (No. 2011CDB391). Conflict of interest authors.
There are no conflicts of interest for any of the
References 1. GASTRIC (Global Advanced/Adjuvant Stomach Tumor Research International Collaboration) Group, Paoletti X, Oba K, Burzykowski T et al (2010) Benefit of adjuvant chemotherapy for resectable gastric cancer: a meta-analysis. JAMA 303(17):1729– 1737 2. Gunderson LL, Sosin H (1982) Adenocarcinoma of the stomach: areas of failure in a re-operation series (second or symptomatic look) clinicopathologic correlation and implications for adjuvant therapy. Int J Radiat Oncol Biol Phys 8(1):1–11 3. Kaibara N, Iitsuka Y, Kimura A et al (1987) Relationship between area of serosal invasion and prognosis in patient with gastric carcinoma. Cancer 60(1):136–139 4. Maehara Y, Moriguchi S, Kakeji Y et al (1991) Pertinent risk factors and gastric carcinoma with synchronous peritoneal dissemination or liver metastasis. Surgery 110(5):820–823 5. Noda S, Soejima K, Inokuchi K (1980) Clinicopathological analysis of the intestinal type and diffuse type of gastric carcinoma. Jpn J Surg 10(4):277–283 6. Saito H, Tsujitani S, Maeda Y et al (2001) Combined resection of invaded organs in patients with T4 gastric carcinoma. Gastric Cancer 4(4):206–211 7. To EM, Chan WY, Chow C et al (2003) Gastric cancer cell detection in peritoneal washing: cytology versus RT-PCR for CEA transcripts. Diagn Mol Pathol 12(2):88–95 8. Boku T, Nakane Y, Minoura T et al (1990) Prognostic significance of serosal invasion and free intraperitoneal cancer cells in gastric cancer. Br J Surg 77(4):436–439 9. Bonenkamp JJ, Songun I, Hermans J et al (1996) Prognostic value of positive cytology findings from abdominal washings in patients with gastric cancer. Br J Surg 83(5):672–674 10. Schott A, Vogel I, Krueger U et al (1998) Isolated tumor cells are frequently detectable in the peritoneal cavity of gastric and colorectal cancer patients and serve as a new prognostic marker. Ann Surg 227(3):372–379 11. Bando E, Yonemura Y, Takeshita Y et al (1999) Intraoperative lavage for cytological examination in 1,297 patients with gastric carcinoma. Am J Surg 178(3):256–262 12. Suzuki T, Ochiai T, Hayashi H et al (1999) Peritoneal lavage cytology findings as prognostic factor for gastric cancer. Semin Surg Oncol 17(2):103–107 13. Mori K, Aoyagi K, Ueda T et al (2004) Highly specific marker genes for detecting minimal gastric cancer cells in cytology negative peritoneal washings. Biochem Biophys Res Commun 313(4):931–937 14. Tsutsumi S, Asao T, Shimura T et al (2000) A novel rapid colorimetric assay of carcinoembryonic antigen levels in the abdominal cavity to detect peritoneal micrometastasis during gastric cancer surgery. Cancer Lett 149(1–2):1–5 15. Abe N, Watanabe T, Toda H et al (2001) Prognostic significance of carcinoembryonic antigen levels in peritoneal washes in patients with gastric cancer. Am J Surg 181(4):356–361 16. Kodera Y, Nakanishi H, Ito S et al (2002) Quantitative detection of disseminated free cancer cells in peritoneal washes with realtime reverse transcriptase-polymerase chain reaction: a sensitive predictor of outcome for patients with gastric carcinoma. Ann Surg 235(4):499–506
123
Ir J Med Sci 17. Bossuyt PM, Reitsma JB, Bruns DE et al (2003) Standards for reporting of diagnostic accuracy towards complete and accurate reporting of studies of diagnostic accuracy: the STARD initiative. BMJ 326(7379):41–44 18. Whiting P, Rutjes AW, Reitsma JB et al (2003) The development of QUADAS: a tool for the quality assessment of studies of diagnostic accuracy included in systematic reviews. BMC Med Res Methodol 3:25 19. Deville´ WL, Buntinx F, Bouter LM et al (2002) Conducting systematic reviews of diagnostic studies: didactic guidelines. BMC Med Res Methodol 2:9 20. Egger M, Davey Smith G, Schneider M et al (1997) Bias in metaanalysis detected by a simple, graphical test. BMJ 315(7109):629–634 21. Cetin B, Atalay C, Aslan S et al (2005) Peritoneal carcinoembryonic antigen level for predicting locoregional and distant spread of gastric cancer. Surg Today 35(11):919–924 22. Fujii S, Kitayama J, Kaisaki S et al (2002) Carcinoembryonic antigen mRNA in abdominal cavity as a useful predictor of peritoneal recurrence of gastric cancer with serosal exposure. J Exp Clin Cancer Res 21(4):547–553 23. Ishii T, Fujiwara Y, Ohnaka S et al (2004) Rapid genetic diagnosis with the transcription-reverse transcription concerted reaction system for cancer micrometastasis. Ann Surg Oncol 11(8):778–785 24. Ito S, Nakanishi H, Kodera Y et al (2005) Prospective validation of quantitative CEA mRNA detection in peritoneal washes in gastric carcinoma patients. Br J Cancer 93(9):986–992 25. Katsuragi K, Yashiro M, Sawada T et al (2007) Prognostic impact of PCR-based identification of isolated tumor cells in the peritoneal lavage fluid of gastric cancer patients who underwent a curative R0 resection. Br J Cancer 97(4):550–556 26. Kodera Y, Nakanishi H, Ito S et al (2006) Prognostic significance of intraperitoneal cancer cells in gastric carcinoma: analysis of real time reverse transcriptase-polymerase chain reaction after 5 years of followup. J Am Coll Surg 202(2):231–236 27. Li JK, Zheng M, Miao CW et al (2005) Peritoneal lavage cytology and carcinoembryonic antigen determination in predicting peritoneal metastasis and prognosis of gastric cancer. World J Gastroenterol 11(46):7374–7377 28. Nishiyama M, Takashima I, Tanaka T et al (1995) Carcinoembryonic antigen levels in the peritoneal cavity: useful guide to peritoneal recurrence and prognosis for gastric cancer. World J Surg 19(1):133–137 29. Tokuda K, Natsugoe S, Nakajo A et al (2003) Clinical significance of CEA-mRNA expression in peritoneal lavage fluid from patients with gastric cancer. Int J Mol Med 11(1):79–84 30. Ueno H, Yoshida K, Hirai T et al (2003) Quantitative detection of carcinoembryonic antigen messenger RNA in the peritoneal cavity of gastric cancer patients by real-time quantitative reverse transcription polymerase chain reaction. Anticancer Res 23(2C): 1701–1708 31. Wang JY, Lin SR, Lu CY et al (2005) Gastric cancer cell detection in peritoneal lavage: RT-PCR for carcinoembryonic antigen transcripts versus the combined cytology with peritoneal carcinoembryonic antigen levels. Cancer Lett 223(1):129–135
123
32. Yamamoto M, Baba H, Toh Y et al (2007) Peritoneal lavage CEA/CA125 is a prognostic factor for gastric cancer patients. J Cancer Res Clin Oncol 133(7):471–476 33. Benevolo M, Mottolese M, Cosimelli M et al (1998) Diagnostic and prognostic value of peritoneal immunocytology in gastric cancer. J Clin Oncol 16(10):3406–3411 34. Wu CC, Chen JT, Chang MC et al (1997) Optimal surgical strategy for potentially curable serosa-involved gastric carcinoma with intraperitoneal free cancer cells. J Am Coll Surg 184(6): 611–617 35. Abe S, Yoshimura H, Tabara H et al (1995) Curative resection of gastric cancer: limitation of peritoneal lavage cytology in predicting the outcome. J Surg Oncol 59(4):226–229 36. Juhl H, Stritzel M, Wroblewski A et al (1994) Immunocytological detection of micrometastatic cells: comparative evaluation of findings in the peritoneal cavity and the bone marrow of gastric, colorectal and pancreatic cancer patients. Int J Cancer 57(3):330–335 37. Asao T, Fukuda T, Yazawa S et al (1991) Carcinoembryonic antigen levels in peritoneal washings can predict peritoneal recurrence after curative resection of gastric cancer. Cancer 68(1):44–47 38. Korenaga D, Okuyama T, Orita H et al (1994) Role of intraoperative assessment of lymph node metastasis and serosal invasion in patients with gastric cancer. J Surg Oncol 55(4):250–254 39. Kodera Y, Nakanishi H, Yamamura Y et al (1998) Prognostic value and clinical implications of disseminated cancer cells in the peritoneal cavity detected by reverse transcriptase-polymerase chain reaction and cytology. Int J Cancer 79(4):429–433 40. Nakanishi H, Kodera Y, Torii A et al (1997) Detection of carcinoembryonic antigen-expressing free tumor cells in peritoneal washes from patients with gastric carcinoma by polymerase chain reaction. Jpn J Cancer Res 88(7):687–692 41. Yonemura Y, Endou Y, Fujimura T et al (2001) Diagnostic value of preoperative RT-PCR-based screening method to detect carcinoembryonic antigen-expressing free cancer cells in the peritoneal cavity from patients with gastric cancer. ANZ J Surg 71(9):521–528 42. Nakanishi H, Kodera Y, Yamamura Y et al (2000) Rapid quantitative detection of carcinoembryonic antigen-expressing free tumor cells in the peritoneal cavity of gastric-cancer patients with real-time RT-PCR on the lightcycler. Int J Cancer 89(5):411–417 43. Nakanishi H, Kodera Y, Yamamura Y et al (1999) Molecular diagnostic detection of free cancer cells in the peritoneal cavity of patients with gastrointestinal and gynecologic malignancies. Cancer Chemother Pharmacol 43:S32–S36 44. Kojima N, Kunieda K, Matsui K et al (2003) Evaluation of carcinoembryonic antigen mRNA in living, necrotic, and apoptotic gastric cancer cells by reverse transcriptase-polymerase chain reaction. Surg Today 33(11):839–846 45. Glas AS, Lijmer JG, Prins MH et al (2003) The diagnostic odds ratio: a single indicator of test performance. J Clin Epidemiol 56(11):1129–1135
ORIGINAL ARTICLE
Diagnostic values of carcinoembryonic antigen in predicting peritoneal recurrence after curative resection of gastric cancer: a meta-analysis Y. Xiao • J. Zhang • X. He • J. Ji • G. Wang
Received: 16 October 2013 / Accepted: 3 December 2013 Ó Royal Academy of Medicine in Ireland 2013
Abstract Aim A meta-analysis was performed to assess the diagnostic values of carcinoembryonic antigen (CEA) in predicting the peritoneal recurrence after curative resection of gastric cancer. Methods The Medline, Embase, Web of Science, Ovid and Cochrane databases, Google Scholar and Vivisimo engines were searched to identify studies reporting on the accuracy of CEA protein or CEA mRNA in predicting the postoperative peritoneal recurrence of gastric cancer. Publication bias was demonstrated by Funnel plots and Egger test. The sensitivity, specificity, and diagnostic odds ratio (DOR) were calculated and summary receiver operating characteristic curves were generated. Results Seven and eight studies fulfilled the inclusion criteria for CEA protein and mRNA determination, including 635 and 849 patients, respectively. The pooled sensitivity, specificity and DOR of CEA protein for predicting the peritoneal recurrence were 0.77 (95 % CI 0.69–0.84), 0.89 (95 % CI 0.86–0.92), 29.71 (95 % CI 10.27–85.92), respectively. Similarly, the values for CEA
Y. Xiao J. Ji G. Wang (&) Department of Gastrointestinal Surgery, Union Hospital, Tongji Medical College, HuaZhong University of Science and Technology, Jiefang Ave 1277#, Wuhan 430022, Hubei, China e-mail: [email protected] J. Zhang Department of Emergency Surgery, Union Hospital, Tongji Medical College, HuaZhong University of Science and Technology, Wuhan 430022, China X. He Department of Obstetrics and Gynecology, Union Hospital, Tongji Medical College, HuaZhong University of Science and Technology, Wuhan 430022, China
mRNA were 0.82 (95 % CI 0.75–0.88), 0.82 (95 % CI 0.79–0.85) and 22.97 (95 % CI 10.90–48.41). Meanwhile, the sensitivity and DOR of CEA protein or mRNA were higher than those of cytology, while higher specificity was noted in cytology assay. Conclusion CEA protein and mRNA levels in peritoneal lavage show a high diagnostic accuracy and may help accurately predict the peritoneal recurrence after curative resection of gastric cancer. Keywords Meta-analysis Carcinoembryonic antigen Peritoneal recurrence Gastric cancer Diagnosis Peritoneal lavage cytology
Introduction Gastric cancer is the second most common cause of cancer death worldwide. Despite of recent advances in surgical techniques and adjuvant chemotherapies, the prognosis of advanced gastric cancer still remains poor [1]. Peritoneal metastasis, the most frequent type of recurrence in patients without evidence of residual tumor at the time of surgery, is a significant obstacle in the curative treatment of gastric cancer [2]. Traditionally, risk factors for peritoneal recurrence include the size of serosal invasion [3], the status of lymph node involvement [4], and histologically diffuse type [5]. However, some patients with tumors involving the serosal surface (T3 and T4 according to the TMN classification) do not present evidences of peritoneal recurrence, even without adjuvant chemotherapy [6]. Thus, tumor class alone is not a reliable method for predicting the peritoneal recurrences of gastric cancer [7]. A simple and accurate method to detect the residual cancer cells still needs to be developed.
123
Ir J Med Sci
Intraoperative detection of free cancer cells in the peritoneal cavity appears to be predictive of peritoneal recurrence in gastric cancer patients [8–10]. Cytological examination of lavage fluid obtained at the time of surgery is a conventional method to detect free cancer cells in the peritoneal space. However, the sensitivity of peritoneal lavage cytology (PLC) is relatively low, ranging from 22 to 30 % in gastric carcinoma involving the serosa [11, 12], making it less valuable in determining the postoperative prognosis. Carcinoembryonic antigen (CEA) is a wellknown tumor marker of gastrointestinal tumor, and has been applied to detect a small number of adenocarcinoma cells in the blood [13]. However, the serum CEA level does not elevate until the tumor is enlarged and unresectable, resulting in its limited diagnostic value [14]. Recently, measurement of CEA protein and mRNA level in peritoneal lavage has been used to detect the existence of free cancer cells in the peritoneal cavity [15, 16]. However, due to the limited sample size and varied accuracy in the literature, the overall diagnostic values of CEA in predicting the peritoneal recurrence of gastric cancer still need to be addressed. Therefore, our major interest has been focused on the ultimate benefit of CEA measurement in peritoneal washings, and performed the present meta-analysis to assess its values in predicting the postoperative peritoneal recurrence of gastric cancer.
Inclusion and exclusion criteria Studies relevant to the diagnostic values of CEA in predicting the peritoneal recurrence after curative resection of gastric cancer were included only if they fulfilled the following criteria: (1) CEA protein or mRNA in peritoneal lavage was used to predict the peritoneal recurrence after the curative resection of histologically proved gastric cancer. (2) Clinical diagnosis of peritoneal recurrence after surgery was used as reference standard. (3) On the patient basis, sufficient data were available to calculate the true positive (TP), true negative (TN), false positive (FP) and false negative (FN). TP were defined as the patients that had CEA expression in intraoperative peritoneal lavage and postoperative peritoneal recurrence. TN findings were patients who had no evidence of intraoperative CEA expression and postoperative peritoneal recurrence. FP findings were defined as patients with intraoperative CEA expression but not postoperative peritoneal recurrence. FN findings were failures to identify intraoperative CEA expression, with postoperative peritoneal recurrence. Conference abstracts were excluded because of the limited data presented in them. Although no language restrictions were imposed initially, the search was limited to studies published in the English language for the full-text review and final analysis. Data extraction
Materials and methods Literature search We searched the online databases, including Medline (using PubMed as the search engine), Embase, Web of Science, Ovid and Cochrane, to identify suitable studies documenting the measurement of CEA in predicting the peritoneal recurrence after curative resection of gastric cancer between 1995 and the end of May 2011. The search was also made using GoogleTM Scholar, and VivisimoTM search engine. The terms ‘‘carcinoembryonic antigen’’, ‘‘peritoneal lavage’’, ‘‘peritoneal washing’’, ‘‘gastric cancer’’, ‘‘gastric carcinoma’’, ‘‘sensitivity’’ and ‘‘specificity’’ were used as Mesh search headings. The above terms and their combinations were also searched as text words. All scanned abstracts, studies, and citations were reviewed. Moreover, references of the retrieved manuscripts were also manually cross-searched for further relevant publications. The review was limited to the published studies and no attempt was made to contact the authors to obtain unpublished data or unclear information in the studies. All searches were limited to ‘‘humans’’. The latest eligible study for the above search period was July 31, 2007.
123
The full article was retrieved when they met the inclusion criteria. Two reviewers (LDZ and QST) independently evaluated and extracted data from potentially relevant studies. Data regarding the following factors were considered: first author, publication year, country origin of the study, study population characteristics, study design, methods to measure CEA, cut-off values, numbers of TP, TN, FP and FN observations. For analysis of the outcomes of interest, any disagreements between the reviewers were discussed in committee and a consensus was reached. Quality assessment of included studies The methodological quality of the selected studies was graded by two reviewers (LDZ and QST), with the standards for reporting diagnostic accuracy (STARD) [17] and quality assessment of diagnostic accuracy studies (QUADAS) tools [18]. Each diagnostic study was scored on the basis of the 25 STARD and 14 QUADAS criteria [17, 18]. The STARD or QUADAS scoring system awards one point if a specific criterion is met; zero points if a criterion is unclear, and one point is deducted if a criterion is not fulfilled. Disagreement among the two reviewers (LDZ and QST) was discussed in committee and a consensus was reached. The Pearson correlation coefficient was used to
Ir J Med Sci
find a correlation between the two reviewers. The value for a Pearson’s can fall between 0.00 (no correlation) and 1.00 (perfect correlation). Both reviewers demonstrated significant correlation (Pearson’s r = 0.94). Statistical analysis Recommended standard methods were used for metaanalyses of diagnostic test evaluations [19]. Analyses were performed using two statistical software programs, Stata (version 11.0; Stata Corporation) and Meta-DiSc (version 1.4; XI Cochrane Colloquium). For each study, the sensitivity, specificity, positive likelihood ratio (PLR), negative likelihood ratio (NLR) and diagnostic odds ratio (DOR) were computed to measure the test accuracy. A randomeffects model was used to calculate the average sensitivity, specificity and other measures across studies. The presence of publication bias, if any, was evaluated using graphical exploration with funnel plots and the Egger test [20].
Results Quality of reporting and study characteristics We identified 52 potential studies published between 1995 and 2011 that might match the selection criteria, of which 13 were excluded due to review article (n = 1), no detailed findings (n = 2), and conference abstracts (n = 10). Thirty-nine articles were then retrieved for full-text review, of which 25 were excluded due to insufficient data (n = 8), no prediction of peritoneal recurrence (n = 10), combined analysis with cytokeratin 20 (n = 2), duplicate studies by the same author (n = 4), and inconsistent data (n = 1). Fourteen studies had sufficient data for the calculation of a summary sensitivity and specificity value [14, 15, 21–32], and were included in the meta-analysis. Of the included studies, 92.9 % (13/14) were published in 2000–2007, and 7.1 % (1/14) were published in 1995 (Table 1). The majority of studies (78.6 %) were performed in the Japan (11/14), whereas 14.3 % (2/14) were performed in China (mainland and Taiwan), and 7.1 % (1/ 14) in Turkey (Table 1). Seven and eight studies fulfilled the inclusion criteria for CEA protein and mRNA determination, including 635 and 849 patients, respectively (Tables 2, 3). In all studies included in the meta-analysis, the diagnoses of patients with peritoneal recurrence were made based on clinical findings, supported by ultrasonography, computed tomography or magnetic resonance imaging. Radiometric immunoassay, enzyme immunoassay or sphere turbidimetric assay were used to detect CEA protein in seven studies [14, 15, 21, 27, 28, 31, 32], whereas CEA mRNA in peritoneal lavage was measured by
reverse-transcriptase polymerase chain reaction (RT-PCR), real-time RT-PCR or transcription–reverse transcription concerted reaction (TRC) in eight studies [22–26, 29–31]. The clinical characteristics of these studies, along with STARD and QUADAS scores, were outlined in Table 1. In all, the quality of study design and reporting diagnostic accuracy of most studies were good, as all studies had higher QUADAS scores ([10), and only two studies had lower STARD scores (\13). Diagnostic accuracy of CEA protein Seven studies documented the sensitivity and specificity of CEA protein in prediction of the peritoneal recurrence of gastric cancer (Table 2) [14, 15, 21, 27, 28, 31, 32]. As shown in Fig. 1, the sensitivity ranged from 0.50 to 1.00 (pooled 0.77, 95 % CI 0.69–0.84), whereas specificity ranged from 0.73 to 0.95 (pooled 0.89, 95 % CI 0.86–0.92); it was noted that PLR was 6.89 (95 % CI 4.00–11.85), NLR was 0.29 (95 % CI 0.16–0.54) and DOR was 29.71 (95 % CI 10.27–85.92). The summary receiver operating characteristic (SROC) curve showed that the maximum joint sensitivity and specificity was 0.8699 (SEM 0.038); whereas the area under the curve (AUC) was 0.934 (SEM 0.031), indicating that the level of overall accuracy of CEA protein was high (Fig. 2). In addition, five included studies compared the sensitivity and specificity of CEA protein and PLC in predicting the peritoneal recurrence of gastric cancer [15, 21, 27, 28, 31]. Analysis of the pooled data revealed that the sensitivity of CEA protein and PLC was 0.72 (95 % CI 0.60–0.82) and 0.45 (95 % CI 0.33–0.57), respectively, whereas their specificity was 0.88 (95 % CI 0.84–0.92) and 0.92 (95 % CI 0.89–0.95). Diagnostic accuracy of CEA mRNA Eight studies reported the sensitivity and specificity of CEA mRNA in the prediction of peritoneal recurrence of gastric cancer (Table 3) [22–26, 29–31]. As shown in Fig. 1, the sensitivity ranged from 0.65 to 1.00 (pooled 0.82, 95 % CI 0.75–0.88), whereas specificity ranged from 0.65 to 0.90 (pooled 0.82, 95 % CI 0.79–0.85); meanwhile, the PLR was 4.59 (95 % CI 3.30–6.37), NLR was 0.22 (95 % CI 0.12–0.40) and DOR was 22.97 (95 % CI 10.90–48.41). The SROC curve showed that the maximum joint sensitivity and specificity was 0.8221 (SEM 0.024); whereas the area under the curve (AUC) was 0.891 (SEM 0.023), indicating the high level of overall accuracy of CEA mRNA (Fig. 2). In the included studies, five articles compared the sensitivity and specificity of CEA mRNA and PLC in the prediction of peritoneal recurrence of gastric cancer [22, 26, 29–31]. The pooled analysis
123
Ir J Med Sci Table 1 Characteristics of included studies Author
Year
Country
Duration
Design
Quality score STARD
QUADAS 12
Nishiyama
1995
Japan
NA
Prospective
11
Tsutsumi
2000
Japan
1991–1996
Prospective
12
11
Abe
2001
Japan
1994
Prospective
16
13
Fujii
2002
Japan
1995–1998
Prospective
15
13
Ueno
2003
Japan
1998–2001
Prospective
17
13
Tokuda
2003
Japan
1997–1999
Prospective
15
13
Ishii Wang
2004 2005
Japan Taiwan
1999–2002 2001–2002
Prospective Prospective
14 17
14 12
Ito
2005
Japan
1995–1999
Prospective
14
14
Cetin
2005
Turkey
1998–2001
Prospective
16
12
Li
2005
China
1995–1997
Prospective
17
12
Kodera
2006
Japan
1995–1999
Prospective
17
13
Katsuragi
2007
Japan
NA
Prospective
14
12
Yamamoto
2007
Japan
1996–2001
Prospective
14
12
STARD standards for reporting diagnostic accuracy, QUADAS quality assessment for studies of diagnostic accuracy, NA non-available
Table 2 Summary of included studies of CEA protein in predicting peritoneal recurrence of gastric cancer Author
No. of patients
Cut-off
CEA protein TP
Nishiyama Tsutsumi
Cytology
FP
FN
TN
TP
FP
FN
TN
118
100 ng/g
10
5
4
99
2
1
12
103
60
100 ng/g
18
3
0
39
NA
NA
NA
NA
Abe
54
210 ng/g
6
3
1
44
3
0
4
47
Cetin Li
70 64
10 ng/g 100 ng/g
11 18
14 8
8 1
37 37
9 14
17 1
10 5
34 44
40
200 ng/g
6
2
6
26
4
2
8
26
229
0.4 ng/ml
29
21
9
170
NA
NA
NA
NA
FP
FN
TN
Wang Yamamoto
TP true positive, FP false positive, FN false negative, TN true negative, NA non-available Table 3 Summary of included studies of CEA mRNA in predicting peritoneal recurrence of gastric cancer Author
No. of patients
Cut-off
CEA mRNA
Cytology
TP
FP
FN
TN
TP
Fujii
49
10 cells/S
15
12
0
22
5
1
10
33
Ueno
78
27.11a
10
21
3
44
3
1
10
64
10 cells/S
15
15
1
105
5
0
11
120
9
5
0
46
NA
NA
NA
NA
8 11
3 9
4 2
25 64
4 NA
2 NA
8 NA
26 NA
32
46
4
202
17
8
19
240
24
14
13
65
NA
NA
NA
NA
Tokuda
136
Ishii
60
1.2
Wang Ito
40 86
10 cells/S 0.1 cells/S
Kodera
284
0.1 cells/S
Katsuragi
116
2.490 910
-2b
TP true positive, FP false positive, FN false negative, TN true negative, S sample, NA non-available a
Crossing-point value
b
CEA/GAPDH ratio
123
Ir J Med Sci
Fig. 1 Forest plot of the sensitivity and specificity of CEA protein or mRNA in the prediction of peritoneal recurrence of gastric cancer. The point estimates of sensitivity and specificity from each study were shown as solid circles. Error bars indicate 95 % confidence intervals
Fig. 2 Summary receiver operating characteristic curves for CEA protein or mRNA. Each study included in the meta-analysis was represented by the solid circle, with the size indicated by the size of the solid circle
revealed that the sensitivity of CEA mRNA and PLC was 0.87 (95 % CI 0.78–0.93) and 0.37 (95 % CI 0.27–0.48), respectively, whereas their specificity was 0.80 (95 % CI 0.77–0.84) and 0.98 (95 % CI 0.96–0.99).
Publication bias Evaluation of publication bias for CEA protein and mRNA showed that the Egger test was not significant (P = 0.193,
123
Ir J Med Sci
Fig. 3 Funnel graph for the assessment of potential publication bias in CEA protein or mRNA determination. Each study included in the meta-analysis was represented by the solid circle. Funnel graph plots
the log of the diagnostic odds ratio (DOR) against the SE of the log of the DOR (an indicator of sample size). The line in the center indicates the summary DOR
P = 0.07, respectively). The funnel plots for publication bias did not show obvious asymmetry (Fig. 3). In addition, meta-regression analysis of the STARD and QUADAS scores indicated no effects of study quality on relative DOR of CEA protein (P = 0.645 and P = 0.251) or CEA mRNA (P = 0.492 and P = 0.227) in predicting the peritoneal recurrence of gastric cancer.
of gastric cancer. In 1991, Asao et al. [37] first reported that CEA protein levels in peritoneal washings can be a useful guide to peritoneal recurrence of gastric cancer. A series of studies demonstrated that CEA protein levels in peritoneal washings could be more responsive to the presence of peritoneal dissemination than PLC, with low falsepositive rate [14, 15, 21, 28]. In 64 patients with gastric cancer, all the 15 patients with PLC positive findings were CEA protein positive, while 11 of 49 patients with PLC negative findings were also CEA positive [27]. We further analyzed five studies comparing the sensitivity and specificity between CEA protein and PLC [22, 26, 29–31], and the results indicated that CEA protein determination was more sensitive (0.72 versus 0.45) but less specific (0.88 versus 0.92) than PLC assay, suggesting that CEA protein level in peritoneal washings is of potential value as an indicator of peritoneal recurrence [38]. As a more sensitive method than conventional PLC, the polymerase chain reaction (PCR) method has recently been developed to detect CEA in peritoneal washing fluid [39]. In general, almost all gastric cancer cells are considered to have and secrete CEA mRNA [40]. Nakanishi et al. [40] reported a higher sensitivity of CEA RT-PCR assay (31 %) over routine PLC (10 %) in the prediction of peritoneal recurrence; combining PLC and CEA RT-PCR assays improved the sensitivity rate up to 57 % [41]. In another study, 93.7 % patients (15/16) with peritoneal recurrence after surgery were positive for CEA mRNA, but only 31.2 % (5/16) were positive by PLC examination [29]. Thus, CEA RT-PCR assay may serve as a useful method for predicting the peritoneal recurrence of gastric cancer patients. Quantitative measurement of CEA mRNA with subsequent automatic data analysis can be completed well within 3 h [42]. Nakanishi et al. [43] also reported that the results obtained from real-time RT-PCR were essentially
Discussion Peritoneal lavage cytology (PLC) is the gold standard for assessing the presence or absence of peritoneal dissemination of gastric cancer. However, its sensitivity is relatively low in gastric cancer involving the serosa [33, 34]. Some patients with cytologically negative peritoneal washing develop peritoneal recurrence of gastric cancer after surgery [35]. In recent years, novel diagnostic tools, including immunohistochemical and molecular biological methods, have been developed to detect free cancer cells [33, 36]. Although several investigators have reported that immunohistochemistry with antibody panels can be an aid to conventional cytology [33, 36], the sensitivity still needs to be increased. In addition, these diagnostic techniques are time-consuming and laborious compared to conventional PLC [16]. Therefore, there is a need for a more accurate test to early detect the peritoneal metastasis of gastric cancer. In the present meta-analysis, our results indicate that determining CEA protein levels in peritoneal lavage is sensitive (0.77, 95 % CI 0.69–0.84), suggesting that CEA determination is feasible to predict the peritoneal recurrence of gastric cancer after surgery. Specificity was also high (0.89, 95 % CI 0.86–0.92), and this was very helpful in confirming peritoneal recurrence after curative resection
123
Ir J Med Sci
the same as those obtained by conventional RT-PCR. However, some CEA secreting cells found in peritoneal washing fluids, such as macrophages, leukocytes and necrotic cancer cells, may alter the sensitivity of PCR results, resulting in false positivity [42, 44]. In this metaanalysis, we noted the sensitivity (0.82, 95 % CI 0.75–0.88) and specificity (0.82, 95 % CI 0.79–0.85) of CEA mRNA determination were suitable for the prediction of peritoneal recurrence of gastric cancer. Similarly, higher sensitivity (0.87 versus 0.37) and lower specificity (0.80 versus 0.98) were noted for CEA mRNA determination than PLC in five included studies [22, 26, 29–31]. Based on the comprehensive search strategy, including screening, study selection and quality assessment, the quality of study design and reporting diagnostic accuracy of most studies were good. To present a global summary of test performance, we further applied the SROC curve to show the trade-off between sensitivity and specificity. The results showed that the maximum joint sensitivity and specificity of CEA protein and mRNA determination was 0.8699 and 0.8221, respectively, with AUC being 0.934 and 0.891. In addition, as an indicator of test accuracy that combines data from sensitivity and specificity into a single number [45], DOR was applied to indicate better discriminatory test performance. We found that the DOR values of CEA protein and mRNA determination were 29.71 and 22.97, respectively, further conforming that their roles in the prediction of peritoneal recurrence of gastric cancer. The present meta-analysis had some limitations. First, although obvious publication bias was not noted by funnel plot and Egger test, the conference abstracts and nonEnglish language studies were excluded, resulting in potential publication bias. Second, due to the lack of sufficient data in the original publications, it was not possible to compare the accuracy of CEA protein and mRNA determination in the prediction of peritoneal recurrence of gastric cancer. Third, there were different methods to measure CEA protein and CEA mRNA. Due to the limited study number, it was not possible to analyze the effects of laboratory infrastructure, expertise with assay technology and patient spectrum. Thus, a series of studies with larger case number and standard methods are necessary to further explore the accuracy of CEA protein and mRNA. In conclusion, CEA protein and mRNA levels in peritoneal lavage show a high diagnostic accuracy and may help accurately predict the peritoneal recurrence after curative resection of gastric cancer. The sensitivity and diagnostic odds ratio of CEA protein or mRNA are higher than those of PLC, while higher specificity of PLC is noted. Combining determination of CEA and PLC may further improve the diagnostic accuracy for predicting the peritoneal recurrence of gastric cancer.
Acknowledgments This study was funded by Natural Science Fund Project of Hubei China (No. 2011CDB391). Conflict of interest authors.
There are no conflicts of interest for any of the
References 1. GASTRIC (Global Advanced/Adjuvant Stomach Tumor Research International Collaboration) Group, Paoletti X, Oba K, Burzykowski T et al (2010) Benefit of adjuvant chemotherapy for resectable gastric cancer: a meta-analysis. JAMA 303(17):1729– 1737 2. Gunderson LL, Sosin H (1982) Adenocarcinoma of the stomach: areas of failure in a re-operation series (second or symptomatic look) clinicopathologic correlation and implications for adjuvant therapy. Int J Radiat Oncol Biol Phys 8(1):1–11 3. Kaibara N, Iitsuka Y, Kimura A et al (1987) Relationship between area of serosal invasion and prognosis in patient with gastric carcinoma. Cancer 60(1):136–139 4. Maehara Y, Moriguchi S, Kakeji Y et al (1991) Pertinent risk factors and gastric carcinoma with synchronous peritoneal dissemination or liver metastasis. Surgery 110(5):820–823 5. Noda S, Soejima K, Inokuchi K (1980) Clinicopathological analysis of the intestinal type and diffuse type of gastric carcinoma. Jpn J Surg 10(4):277–283 6. Saito H, Tsujitani S, Maeda Y et al (2001) Combined resection of invaded organs in patients with T4 gastric carcinoma. Gastric Cancer 4(4):206–211 7. To EM, Chan WY, Chow C et al (2003) Gastric cancer cell detection in peritoneal washing: cytology versus RT-PCR for CEA transcripts. Diagn Mol Pathol 12(2):88–95 8. Boku T, Nakane Y, Minoura T et al (1990) Prognostic significance of serosal invasion and free intraperitoneal cancer cells in gastric cancer. Br J Surg 77(4):436–439 9. Bonenkamp JJ, Songun I, Hermans J et al (1996) Prognostic value of positive cytology findings from abdominal washings in patients with gastric cancer. Br J Surg 83(5):672–674 10. Schott A, Vogel I, Krueger U et al (1998) Isolated tumor cells are frequently detectable in the peritoneal cavity of gastric and colorectal cancer patients and serve as a new prognostic marker. Ann Surg 227(3):372–379 11. Bando E, Yonemura Y, Takeshita Y et al (1999) Intraoperative lavage for cytological examination in 1,297 patients with gastric carcinoma. Am J Surg 178(3):256–262 12. Suzuki T, Ochiai T, Hayashi H et al (1999) Peritoneal lavage cytology findings as prognostic factor for gastric cancer. Semin Surg Oncol 17(2):103–107 13. Mori K, Aoyagi K, Ueda T et al (2004) Highly specific marker genes for detecting minimal gastric cancer cells in cytology negative peritoneal washings. Biochem Biophys Res Commun 313(4):931–937 14. Tsutsumi S, Asao T, Shimura T et al (2000) A novel rapid colorimetric assay of carcinoembryonic antigen levels in the abdominal cavity to detect peritoneal micrometastasis during gastric cancer surgery. Cancer Lett 149(1–2):1–5 15. Abe N, Watanabe T, Toda H et al (2001) Prognostic significance of carcinoembryonic antigen levels in peritoneal washes in patients with gastric cancer. Am J Surg 181(4):356–361 16. Kodera Y, Nakanishi H, Ito S et al (2002) Quantitative detection of disseminated free cancer cells in peritoneal washes with realtime reverse transcriptase-polymerase chain reaction: a sensitive predictor of outcome for patients with gastric carcinoma. Ann Surg 235(4):499–506
123
Ir J Med Sci 17. Bossuyt PM, Reitsma JB, Bruns DE et al (2003) Standards for reporting of diagnostic accuracy towards complete and accurate reporting of studies of diagnostic accuracy: the STARD initiative. BMJ 326(7379):41–44 18. Whiting P, Rutjes AW, Reitsma JB et al (2003) The development of QUADAS: a tool for the quality assessment of studies of diagnostic accuracy included in systematic reviews. BMC Med Res Methodol 3:25 19. Deville´ WL, Buntinx F, Bouter LM et al (2002) Conducting systematic reviews of diagnostic studies: didactic guidelines. BMC Med Res Methodol 2:9 20. Egger M, Davey Smith G, Schneider M et al (1997) Bias in metaanalysis detected by a simple, graphical test. BMJ 315(7109):629–634 21. Cetin B, Atalay C, Aslan S et al (2005) Peritoneal carcinoembryonic antigen level for predicting locoregional and distant spread of gastric cancer. Surg Today 35(11):919–924 22. Fujii S, Kitayama J, Kaisaki S et al (2002) Carcinoembryonic antigen mRNA in abdominal cavity as a useful predictor of peritoneal recurrence of gastric cancer with serosal exposure. J Exp Clin Cancer Res 21(4):547–553 23. Ishii T, Fujiwara Y, Ohnaka S et al (2004) Rapid genetic diagnosis with the transcription-reverse transcription concerted reaction system for cancer micrometastasis. Ann Surg Oncol 11(8):778–785 24. Ito S, Nakanishi H, Kodera Y et al (2005) Prospective validation of quantitative CEA mRNA detection in peritoneal washes in gastric carcinoma patients. Br J Cancer 93(9):986–992 25. Katsuragi K, Yashiro M, Sawada T et al (2007) Prognostic impact of PCR-based identification of isolated tumor cells in the peritoneal lavage fluid of gastric cancer patients who underwent a curative R0 resection. Br J Cancer 97(4):550–556 26. Kodera Y, Nakanishi H, Ito S et al (2006) Prognostic significance of intraperitoneal cancer cells in gastric carcinoma: analysis of real time reverse transcriptase-polymerase chain reaction after 5 years of followup. J Am Coll Surg 202(2):231–236 27. Li JK, Zheng M, Miao CW et al (2005) Peritoneal lavage cytology and carcinoembryonic antigen determination in predicting peritoneal metastasis and prognosis of gastric cancer. World J Gastroenterol 11(46):7374–7377 28. Nishiyama M, Takashima I, Tanaka T et al (1995) Carcinoembryonic antigen levels in the peritoneal cavity: useful guide to peritoneal recurrence and prognosis for gastric cancer. World J Surg 19(1):133–137 29. Tokuda K, Natsugoe S, Nakajo A et al (2003) Clinical significance of CEA-mRNA expression in peritoneal lavage fluid from patients with gastric cancer. Int J Mol Med 11(1):79–84 30. Ueno H, Yoshida K, Hirai T et al (2003) Quantitative detection of carcinoembryonic antigen messenger RNA in the peritoneal cavity of gastric cancer patients by real-time quantitative reverse transcription polymerase chain reaction. Anticancer Res 23(2C): 1701–1708 31. Wang JY, Lin SR, Lu CY et al (2005) Gastric cancer cell detection in peritoneal lavage: RT-PCR for carcinoembryonic antigen transcripts versus the combined cytology with peritoneal carcinoembryonic antigen levels. Cancer Lett 223(1):129–135
123
32. Yamamoto M, Baba H, Toh Y et al (2007) Peritoneal lavage CEA/CA125 is a prognostic factor for gastric cancer patients. J Cancer Res Clin Oncol 133(7):471–476 33. Benevolo M, Mottolese M, Cosimelli M et al (1998) Diagnostic and prognostic value of peritoneal immunocytology in gastric cancer. J Clin Oncol 16(10):3406–3411 34. Wu CC, Chen JT, Chang MC et al (1997) Optimal surgical strategy for potentially curable serosa-involved gastric carcinoma with intraperitoneal free cancer cells. J Am Coll Surg 184(6): 611–617 35. Abe S, Yoshimura H, Tabara H et al (1995) Curative resection of gastric cancer: limitation of peritoneal lavage cytology in predicting the outcome. J Surg Oncol 59(4):226–229 36. Juhl H, Stritzel M, Wroblewski A et al (1994) Immunocytological detection of micrometastatic cells: comparative evaluation of findings in the peritoneal cavity and the bone marrow of gastric, colorectal and pancreatic cancer patients. Int J Cancer 57(3):330–335 37. Asao T, Fukuda T, Yazawa S et al (1991) Carcinoembryonic antigen levels in peritoneal washings can predict peritoneal recurrence after curative resection of gastric cancer. Cancer 68(1):44–47 38. Korenaga D, Okuyama T, Orita H et al (1994) Role of intraoperative assessment of lymph node metastasis and serosal invasion in patients with gastric cancer. J Surg Oncol 55(4):250–254 39. Kodera Y, Nakanishi H, Yamamura Y et al (1998) Prognostic value and clinical implications of disseminated cancer cells in the peritoneal cavity detected by reverse transcriptase-polymerase chain reaction and cytology. Int J Cancer 79(4):429–433 40. Nakanishi H, Kodera Y, Torii A et al (1997) Detection of carcinoembryonic antigen-expressing free tumor cells in peritoneal washes from patients with gastric carcinoma by polymerase chain reaction. Jpn J Cancer Res 88(7):687–692 41. Yonemura Y, Endou Y, Fujimura T et al (2001) Diagnostic value of preoperative RT-PCR-based screening method to detect carcinoembryonic antigen-expressing free cancer cells in the peritoneal cavity from patients with gastric cancer. ANZ J Surg 71(9):521–528 42. Nakanishi H, Kodera Y, Yamamura Y et al (2000) Rapid quantitative detection of carcinoembryonic antigen-expressing free tumor cells in the peritoneal cavity of gastric-cancer patients with real-time RT-PCR on the lightcycler. Int J Cancer 89(5):411–417 43. Nakanishi H, Kodera Y, Yamamura Y et al (1999) Molecular diagnostic detection of free cancer cells in the peritoneal cavity of patients with gastrointestinal and gynecologic malignancies. Cancer Chemother Pharmacol 43:S32–S36 44. Kojima N, Kunieda K, Matsui K et al (2003) Evaluation of carcinoembryonic antigen mRNA in living, necrotic, and apoptotic gastric cancer cells by reverse transcriptase-polymerase chain reaction. Surg Today 33(11):839–846 45. Glas AS, Lijmer JG, Prins MH et al (2003) The diagnostic odds ratio: a single indicator of test performance. J Clin Epidemiol 56(11):1129–1135
