DOI: https://doi.org/10.21203/rs.3.rs-38420/v1
Background:Eastern Asia is a region with a high incidence of gastric cancer (GC), and the detection rate of early gastric cancer (EGC) has been rising in recent decades.2With the development of endoscopic submucosal dissection (ESD) and the scope of its indications gradually expanded,the treatment for EGC is becoming a hot topic, which involves accurate preoperative evaluation.3The purpose of this study was to assess overall accuracy of multidetector computed tomography (MDCT) in N staging of GC in Eastern Asia.4
Methods: A comprehensive search was conducted to collect all relevant published studies about MDCT in assessing N staging of GC in Eastern Asia, within the PubMed, Cochrane library and Web of science databases from January 01, 2002 to April 01, 2020.5 The raw data from the included studies calculated the pooled sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) for evaluating the performance of MDCT.6 The studies were grouped according to different criteria to assess the effect of possible factors on the sensitivity of MDCT.7
Results: fourteen studies (5119 patients) were included in this analysis.8 In Eastern Asia, MDCT had a pooled accuracy of 67% (95% confidence interval [CI] 0.62-0.72, I = 90.1%) for N staging. By subgroup analysis, reference diameter of lymph node metastasis (LNM) and degree of tumor invasion (DTI) were correlated with the sensitivity of N staging.9 And the pooled sensitivity for N+ of EGC was relatively low.10
Conclusions: In Eastern Asia, as the influence of reference diameter of LNM and DTI, the sensitivity of MDCT in N staging of GC was insufficient.11 This may lead to incorrect preoperative staging, so that patients with GC may be over treated or inadequate lymph node (LN)dissection.12 The hierarchical assessment strategy may significantly improve the sensitivity for N+ of EGC to make appropriate methods of treatment.13
Gastric cancer (GC) is the fifth most frequently diagnosed cancer and the third leading cause of cancer death worldwide [1].14 GC causes a particularly high proportion of cancer deaths in China and it has been the most common malignancy in Japan [2, 3].15 It was estimated that Eastern Asia had the highest mortality rate and the incidence of GC in the world in 2008 was about 1 million, half of which occurred in Eastern Asia.16 The detection rate of early gastric cancer (EGC) has been gradually increasing in recent decades, which may be related to the popularization of gastroscopy screening and the improvement of people's health concept.17 Therefore, endoscopic submucosal dissection (ESD) has gradually entered the middle of people's eyes and is committed to removing EGC within the scope of indications.18 Compared with laparoscopic and laparotomy, ESD has many advantages, such as less trauma and faster recovery, but ESD cannot effectively remove positive lymph nodes (LNs).19 The indication of ESD is mainly GC T1a staging without lymph node metastasis (LNM).20 Through clinical comparative tests, the reliability and safety of ESD's extended indication has been verified [4].21 For patients who have been treated with ESD, based on the potential risk of residual tumors in the stomach and / or LNM, gastrectomy + LN dissection was performed.22 The actual incidence of LNM in patients with additional gastrectomy for incomplete tumor removal is about 5%-10% [5–8].23 This means that the vast majority of patients were overtreated.24 Therefore, accurate preoperative staging of GC, especially N staging of EGC, is essential for the selection of appropriate treatment.25 A variety of tests have been used to assess LN status in GC such as multidetector computed tomography (MDCT), endoscopic ultrasonography (EUS), magnetic resonance imaging (MRI), and positron emission tomography/computed tomography (PET/CT) [9].26 MDCT has been used as a hot tool for continuous research to more accurately assess the preoperative staging of GC since its invention [10, 11].27 With the development of the times and technological innovation, MDCT has been greatly improved and is currently considered as one of the most reliable preoperative staging techniques for GC [12–15].28 When using MDCT alone, its accuracy has been reported to reach 89% [16, 17].29 There were also many conclusions about the accuracy of MDCT in evaluating the N staging of GC.30 Fujikawa et al. [18] showed that the main method for preoperative evaluation of N staging of GC with MDCT was to detect the size of LN diameter, but the relatively low sensitivity of MDCT led to non-negligible false negative results.31 Cimavilla et al. [19] concluded that the pooled sensitivity and accuracy of MDCT for N staging were 55% and 64%, respectively, through the comparison between MDCT and EUS.32 However, few studies have studied the accuracy of MDCT in N staging of GC in Eastern Asia with high morbidity and mortality.33 The purpose of this study was to conduct a systematic review and meta-analysis of studies on the diagnostic accuracy of MDCT in the evaluation of LNM of GC in Eastern Asia.34
Search strategy: Two authors searched systematically from the PubMed, Cochrane library and Web of science databases for all relevant articles published from January 01, 2002 to April 01, 2020, that about MDCT for preoperative N staging of GC in Eastern Asia.35 The following keywords were used in the search article: “(stomach neoplasm) OR (gastric neoplasm) OR (stomach cancers) OR (gastric cancer) OR (cancer, gastric) OR (gastric cancers) OR (stomach cancer)” and “(lymph node metastasis) OR (lymph node involvement) OR (lymph node staging) OR (lymphatic metastasis) OR (nodal involvement) OR (lymph node status) OR (N staging)” and “(computed tomography) OR (multidetector computed tomography) OR (CT) OR (MDCT)”.36 No language restrictions, and also manually checked whether the reference list of the related studies met the inclusion criteria.37
Study inclusion and exclusion criteria: inclusion criteria:(1) Number of samples n ≥ 30 cases, no radiotherapy and chemotherapy before surgery, postoperative pathology confirmed GC; (2) Histopathological results after gastrectomy as the criterion for the evaluation of LNM; (3) Only studies contain original data that the sensitivity specificity, positive predictive value (PPV), and negative predictive value (NPV) of N0, N1, N2 and N3 can be obtained or calculated respectively; (4) only studies in eastern Asian population.38 exclusion criteria:(1) Animal or in vitro studies;(2) Review articles, case reports, abstracts, meta-analyses, correspondence and meeting records. 39
Data Extraction, Synthesis and Analysis
If the two authors encounter dispute in the inclusion or exclusion of articles, the dispute shall be settled through consultation with the third author.40 If no agreement can be reached, the article shall be removed to achieve the quality control of the included articles.41 The sensitivity, specificity, PPV and NPV of each N stage were calculated respectively from the raw data or known results with the bivariate random effects model.42 In addition, we divided studies into groups by reference diameter of LNM and the proportion of degree of tumor invasion (DTI) to evaluate possible factors on the accuracy of MDCT.43 P < 0.05 or I²>50% indicated heterogeneity in statistical results.44 Endnote-x9, noteexpress-3.2, and statase-64 were used to organize articles, process data, and make graphics.45
A systematic search was conducted to initially identify 4333 articles for the first review.46 After deleting duplicates and screening related titles and abstracts, the remaining 233 articles were reviewed for full text, and a total of 14 studies [20–33] were finally adopted in accordance with the inclusion criteria (Fig. 1).47 The relevant characteristics of these studies are listed in table 1.48 A total of 5119 patients with GC in Eastern Asia received detailed staging of MDCT before surgery.49
N staging
The overall diagnostic accuracy of MDCT for N staging varied between 43% and 86.3%, and the pooled accuracy was 67% (95% CI 0.62–0.72, I = 90.1%, P < 0.05) in Eastern Asia.50 The range of individual study sensitivity, specificity, PPV, NPV are presented in Table 2, the results of each study vary greatly for N + .51 For N0, the pooled sensitivity, specificity, PPV and NPV were 0.77 (95%CI 0.69–0.84), 0.77 (0.67–0.86), 0.75 (0.71–0.78), and 0.82 (0.78–0.86) respectively (Fig. 2).52 For N1, the results were 0.55 (0.43–0.66), 0.83 (0.78–0.88), 0.47 (0.38–0.56) and 0.86(0.84–0.89) respectively (Fig. 3).53 The pooled sensitivity, specificity, PPV and NPV of MDCT for N2 were 0.43 (0.35–0.52), 0.9 (0.85–0.94), 0.51 (0.38–0.63) and 0.89 (0.86–0.91) respectively (Fig. 4), and were 0.42 (0.31–0.53), 0.99 (0.98-1.0), 0.56 (0.34–0.77) and 0.92 (0.88–0.96) for N3 respectively (Fig. 5).54 The calculated I² value for above results were almost all > 50% and the P value was often < 0.05, indicated that there was considerable heterogeneity among the included studies.55 In terms of grouping, the pooled sensitivity for N0 and N + were 0.66 and 0.88 in 6 studies[20, 22, 26, 28, 32, 33] which the reference diameter of LNM in MDCT was < 8 mm, and another group [21, 23, 25, 27, 30, 31] of the reference was ≥ 8 mm, the pooled sensitivity for N0 and N + were 0.91 and 0.52 respectively.56 The result may instruct that appropriate reduction of the reference diameter of LNM could increase MDCT sensitivity for LNM in Eastern Asia.57 The other grouping was about DTI, which was divided into two groups according to the proportion of T3 and T4 in T stage by pathological results, < 50% (the majority of patients with EGC) [23, 25, 27, 31] or ≥ 50% ( the majority of patients with advanced gastric cancer(AGC)) [21, 24, 26, 28–30].58 The pooled sensitivity for N0 and N + were 0.92 and 0.49 in the proportion < 50%, and were 0.6 and 0.84 in the proportion ≥ 50% respectively.59 It was possible to confirm that MDCT staging sensitivity significantly varied in EGC and AGC in Eastern Asia.60 For N +, there may be better sensitivity for preoperative diagnosis of LNM of AGC by MDCT.61
Table 1. Characteristics of included studies
Table2. The range of individual study sensitivity, specificity, PPV,NPV.
N staging |
sensitivity |
specificity |
PPV |
NPV |
N0 |
0.5-0.97 |
0.45-0.99 |
0.6-0.93 |
0.6-0.98 |
N1 |
0.18-0.88 |
0.68-1.0 |
0.16-1.0 |
0.73-0.99 |
N2 |
0.14-0.74 |
0.72-0.99 |
0.14-0.8 |
0.79-0.94 |
N3 |
0-1.0 |
0.9-1.0 |
0-1.0 |
0.84-1.0 |
This study evaluated data from multiple studies (number of patients n = 5119) to assess the diagnostic performance of MDCT in N staging of GC in Eastern Asia.62 Meta-analytical statistical methods were used to integrate and subgroup the data and it was found that the preoperative diagnostic performance of MDCT was not very reliable and the pooled accuracy of N staging was only 67% (95% CI 0.62–0.72).63 And the sensitivity of MDCT in the clinical diagnosis of GC (especially EGC) N staging was not satisfactory.64
The data were grouped according to the reference diameter of LNM and the proportion of DTI.65 The subgroup analysis results indicated that when the reference diameter of LNM was < 8 mm or ≥ 8 mm, the sensitivity of MDCT for N + staging was significantly different, and the sensitivity of the former was significantly greater than that of the latter.66 Preoperative LNM can be determined from the size of the LN, morphology, strengthening methods to comprehensive analysis.67 The American Joint Committee for Cancer (AJCC) and the Japanese Gastric Cancer Association (JGCA) are currently widely recognized TNM staging criteria for GC.68 However, the above two staging systems did not provide a reference standard for the preoperative determination of positive LNs by MDCT.69 Therefore, the preoperative diagnosis of LNM in GC has always been the focus of research.70 Generally, the diameter of LN is ≥ 8 mm, and LNs are considered to be involved.71 For the grouping of DTI, MDCT in patients with mostly AGC was more sensitive for N + staging than that in patients with mostly EGC, suggesting that the proportion of DTI was an important variable affecting the sensitivity, while MDCT was poor in determining N staging with EGC.72 In AGC with LNM, large or clustered LNs are usually found around the primary lesion [34].73 However, EGC with few LNM usually has very small metastatic LNs and is less typical in MDCT imaging than in AGC [35].74 Microscopic metastases found in the LNs of patients with EGC are often classed as negative because the diameters do not meet the criteria for LNM, which makes accurate N staging more difficult in EGC than in AGC [12].75 This means that MDCT is challenging in the proper clinical evaluation of LN involvement in EGC.76 Early reports showed that the accuracy of preoperative N staging for GC ranged from 51–76%, and the sensitivity of N staging ranged from 48–91% [12].77 Fukuya et al. [36] took > 5 mm as the standard, with N staging sensitivity of 75% and specificity of 42%.78 D 'elia et al. [37] took the shorter diameter of perigastric LN > 6 mm and the shorter diameter of peripheral LN > 8 mm as the standard, and the sensitivity and specificity of N staging were 97.2% and 65.7%, respectively.79 Research reports from different periods have shown that due to the effect of DTI on LN status not considered and the lack of reliable preoperative reference diameter criteria for the determination of positive LNs, the results fluctuated widely.80 Therefore, whether in Eastern Asia or elsewhere, in the preoperative evaluation of GC with MDCT, we need to establish more refined and specific criteria for LNM to detect LN status under different conditions.81
The effect of LNs at different locations on the accuracy of MDCT was also widely discussed.82 Pan et al. [38] showed that the results of LNs evaluation by MDCT at different sites showed different sensitivities compared with those of pathology, in addition, among the various metastatic LNs, MDCT showed better sensitivity in the splenic hilum, splenic artery and the sites with less curvature.83 The results of Jia et al. [22] indicated that the examination of LNs on CT depended on the contrast of surrounding structures, and clear anatomical structure, close to the large blood vessels, more fat in the abdominal cavity and LNs in the stomach area were easier to detect.84 In their study, group location diagnosis was adopted.85 For MDCT, according to the fifth edition of the AJCC TNM staging, the sensitivity of LNs in groups 1, 3, 7–9, and 11 was high, and the sensitivity of LNs in groups 4 and 5 was poor.86 Therefore, the lack of one-to-one correspondence between the LNs evaluated by MDCT before operation and those detected by pathology after operation may make the accuracy of MDCT unreliable.87
As mentioned above, the existing accuracy of MDCT is not sufficient for the preoperative evaluation of EGC.88 In order to select appropriate surgical methods (ESD, gastrectomy + perigastric LN resection or enlarged resection) for patients to achieve precision medicine, hierarchical assessment may be effective according to the result of the subgroup analysis.89 In order to achieve the hierarchical assessment and standard development, imaging technicians should first assess the T staging of GC, and then stratify the patients according to the T staging: EGC (T1,T2) with < 8 mm(such as 6 mm) as the reference diameter of LNM, or AGC (T3,T4) with ≥ 8 mm(such as 8 mm or 10 mm) as the reference diameter of LNM. This hierarchical assessment can take into account both the effect of DTI on LNs and the effect of reference diameter of LNM on the sensitivity.90 Such a diagnostic process may greatly improve the accuracy of preoperative staging of EGC, avoid patients from being overtreated or timely adjust the surgical method to expand the scope of surgery.91 About the LNs in different locations, a criterion for clinical diagnosis of N stages is needed.92 MDCT in the preoperative assessment of LNs status should be accurate to mark specific location of the suspected LNs, surgeons should accurately remove and mark the corresponding LNs to achieve a one-to-one relationship with the pathology.93 This not only enables a real preoperative evaluation of N staging with MDCT, but also enables a more accurate understanding of the biological behavior of LNM in GC.94
Through the above presentation and discussion of the results, this study indicated that the overall accuracy of MDCT for N staging of GC is relatively low in Eastern Asia, especially the sensitivity for N + of EGC, and its sensitivity may be related to the reference diameter of LNM, DTI and the location of LN.95 In order to address the effect of LN differences between EGC and AGC on MDCT, a concept of hierarchical assessment was proposed, which may greatly improve the accuracy of MDCT in the diagnosis of preoperative N staging of EGC to select appropriate surgical procedures for patients.96
gastric cancer
early gastric cancer
endoscopic submucosal dissection
lymph node
lymph node metastasis
multidetector computed tomography
endoscopic ultrasonography
magnetic resonance imaging
positron emission tomography/computed tomography
positive predictive value
negative predictive value
degree of tumor invasion
advanced gastric cancer
the American Joint Committee for Cancer
the Japanese Gastric Cancer Association
Our study did not require an ethical board approval because the included studies were published articles and there are no experiments involving humans or animals.
Consent for publication:
not applicable.
Availability of data and materials:
The datasets generated during and/or analysed during the current study are available in the [NCBI-PubMed] for English articles,
no potential conflict of interests.
not applicable.
XL and GX searched and screened the articles from the database, XL and ZJP reviewed and confirmed the included articles, and XL wrote and modified the article. All authors' read and approved the final manuscript.
Thanks to the surgeons of the Second Affiliated Hospital of Nanjing Medical University for their comments on this article.