- Email: [email protected]
Integrated analysis of the prognostic role of the lymph node ratio in node-positive gastric cancer: A meta-analysis
International Journal of Surgery 57 (2018) 76–83
Contents lists available at ScienceDirect
International Journal of Surgery journal homepage: www.elsevier.com/locate/ijsu
Review
Integrated analysis of the prognostic role of the lymph node ratio in nodepositive gastric cancer: A meta-analysis
T
Jiang Zhua,1, Zhao Xueb,1, Shumei Zhanga, Xinxin Guob, Laihui Zhaib, Shipeng Shanga, Yan Zhanga,∗∗, Haibo Lub,∗ a b
College of Bioinformatics Science and Technology, Harbin Medical University, Harbin, 150081, China Department of Gastrointestinal Medical Oncology, Harbin Medical University Cancer Hospital, Harbin, 150040, China
A R T I C LE I N FO
A B S T R A C T
Keywords: Lymph node positive ratio Meta-analysis Overall survival
Background: The lymph node ratio (LNR) as a prognostic parameter for gastric cancer has yet to be fully validated in the current tumor node metastasis staging system. We assessed the prognostic role of LNR in lymph node-positive gastric cancer through a meta-analysis. Materials and methods: PubMed and EMBASE were searched for relevant studies up until December 2016. The effect measure for meta-analysis of primary outcomes was the hazard ratio (HR) for overall survival. Pooled HRs and 95% confidence intervals were calculated using random effects models. The I2 statistic was used to measure heterogeneity. Subgroup analysis and meta-regression were chosen to illustrate the potential heterogeneity of the risk factors of outcomes. Publication bias was assessed using Egger's test and Begg's funnel plots. Sensitivity analysis was applied to evaluate the origin of the heterogeneity. Results: We included 27 studies in this meta-analysis. Higher LNRs were significantly associated with a shorter overall survival (OS). High heterogeneity among the studies was identified (I2 = 85.6), and the publication bias was moderate. Subgroup analysis showed similar results, and elevated LNR was associated with late-stage gastric cancer and indicative of a worse prognosis. Univariate meta-regression analysis of OS indicated that both treatment type and ethnicity may be causes of heterogeneity in patients with gastric cancer (p values were 0.005 and 0.008, respectively). Conclusion: LNR was associated with a significantly poorer OS and LNR was an independent predictor of survival in patients with gastric cancer. LNR should be added as one of the parameters to be used in future tumor staging classification systems.
1. Introduction Gastric cancer is the fourth most common cancer and the third leading cause of cancer-related death throughout the world [1]. Surgery is the main treatment in gastric cancer, and lymph node metastasis is the most effective predictor of postoperative survival [2]. Therefore, lymph node status is considered to be one of the key prognostic factors in gastric cancer. The AJCC (UICC TNM classification) staging system of malignant tumors showed that proper pathological lymph node staging (pN stage) requires at least 15 lymph nodes to be cleared [3]. Certain factors have led to an insufficient number of lymph nodes being dissected in clinical practice. Increasing the number of lymph nodes dissected has been
found to affect the number of metastases detected based on standard pN staging, which may result in a change in the TMN classification of a cancer that can affect the accuracy of the prognostic prediction [4–6]. The classic staging system for gastric cancer is the TNM staging system. However, the TNM staging system is not used when checking for tumor-free lymph nodes. Thus, the lymph node ratio (LNR) is considered to be an essential prognostic factor and a suitable staging method for patients with positive lymph nodes [7]. However, its use as a prognostic factor is controversial because of conflicting LNR results and differences in study design and sample size used in previous studies. Previous systematic reviews suggested that LNR can be used as a prognostic factor for gastric cancer and colorectal cancer [8,9]. Nevertheless, to date there has been no formal meta-analysis focused on
∗
Corresponding author. Corresponding author. E-mail addresses: [email protected] (Y. Zhang), [email protected] (H. Lu). 1 These authors contributed equally to this work. ∗∗
https://doi.org/10.1016/j.ijsu.2018.08.002 Received 14 May 2018; Received in revised form 25 July 2018; Accepted 6 August 2018 Available online 10 August 2018 1743-9191/ © 2018 Published by Elsevier Ltd on behalf of IJS Publishing Group Ltd.
International Journal of Surgery 57 (2018) 76–83
J. Zhu et al.
calculating the heterogeneity in each situation in which a single study was removed in turn in order to evaluate the effect of a single study on the overall outcome. Subgroup analysis was performed according to each parameter including year, race, treatment, sample size, ratio of patients with stage I/II and stage III/IV gastric cancer, and number of lymph nodes. We compared the pooled HR estimates from different subgroups using an interaction test. A meta-regression model was developed to explore the potential impact of different factors on heterogeneity, and to assess the effect of year, race, treatment, and other risk factors or potential confounding factors on outcome. Finally, Begg's test and Egger's test were used to assess publication bias. In addition, a bias risk test chart was made to assess the risk of various biases and the bias of each article using Revman 5.2 software [12,13]. Statistical analysis was performed using STATA12.0 software, according to the Cochrane Collaboration Organization and the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-analyses) guidelines. A p value < 0.05 indicated a statistically significant difference.
the prognostic significance of LNR in gastric cancer. In this study, the first comprehensive systematic review was conducted to investigate the prognostic role of LNR in patients with lymph node-positive gastric cancer, and the results demonstrated that an increased LNR correlated with poor overall survival (OS) in gastric cancer patients. 2. Materials and Methods 2.1. Search strategy A comprehensive, computerized literature search was conducted in PubMed and Embase for relevant studies up to December 2016. The search was performed using the following text words and corresponding medical subject heading terms: ((gastric cancer) OR (stomach cancer) OR (gastric carcinoma) OR (stomach carcinoma) OR (gastric neoplasm) OR (stomach neoplasm) OR (lymph node ratio)) AND ((LNR) OR (lymph positive node ratio) OR (lymph metastatic node ratio)). The search strategy was repeatedly performed until no new relevant articles were found. In addition, we reviewed references in the retrieved articles to search for additional relevant studies. All articles were evaluated by two authors based on the eligibility criteria we designed.
3. Results 3.1. Study selection and characteristics A flow diagram of the literature selection process used in this study is shown in Fig. 1. A total of 1208 articles were initially retrieved from PubMed and EMBASE according to particular keywords. After manually screening and filtering these, 27 articles were eventually chosen that included a total of 11,441 patients. Characteristics of the included studies are summarized in Table S1 and Table S2. All patients underwent radical surgery, and the median age of patients in the studies ranged from 54 to 72 years old. Of these studies, 17 involved Asian patients and 10 involved non-Asian patients. Lymph node data and survival time of patients were also retrieved. Median follow-up time in the studies ranged from 52 to 75.3 months. HRs and 95% CIs obtained from all articles are listed in Table 1. HRs and 95% CIs of OS could be directly obtained from 12 studies. The other 15 studies did not provide HRs and 95% CIs directly; therefore, we used relative data such as Kaplan–Meier curves and the total number of survivors to calculate HRs and 95% CIs [14].
2.2. Study selection First, we checked titles and abstracts of articles that were searched using keywords to exclude irrelevant articles. Next, all retained studies were screened according to inclusion and exclusion criteria. The following conditions were used to evaluate whether the study was included: (a) all patients were diagnosed with gastric cancer using pathology; (b) patients underwent radical surgery (R0 resection); (c) the outcome of interest was disease-free survival (DFS) and OS; and (d) hazard ratios (HRs) and 95% confidence intervals (CIs) used to evaluate prognosis could be extracted from the original literature, including direct acquisition or approximation by calculation. Exclusion criteria were defined as follows: (a) gastric cancer patients with other tumors or patients with distant metastases (TNM IV staging); (b) articles including patients undergoing tumor-related neoadjuvant chemotherapy before surgery; (c) articles consisting of letters, meeting summaries, commentary articles, and posters; and (d) studies could not provide outcome data and did not calculate the necessary results.
3.2. Primary outcomes and sensitivity analysis Using a random effects model, pooled results of HR and OS statistics from 27 studies indicated that there was a significant association between the LNR and OS in patients with gastric cancer. The survival rate was significantly higher in patients with lower LNRs (HR = 1.99; 95% CI 1.74–2.27; p < 0.001) (Fig. 2). The results also showed that there was high heterogeneity among studies (I2 = 85.6%; p < 0.001). Subsequently, sensitivity analysis was performed to evaluate the stability of the model by omitting each individual study and calculating new HRs. Results showed that HRs were relatively stable and that study heterogeneity was still apparent (Fig. 3).
2.3. Data extraction Data were extracted independently by two researchers and any discrepancies in the data were settled by consensus. If necessary, a third researcher was expected to participate in the discussion and make a decision. The main data extracted from each study included: first author, publication year, number of patients, country of the study population, study design, duration of follow-up, patient age, gender, number of checked nodes, type of study, cut-off value of the LNR and the definition of stratification, and HRs and 95% CIs. The primary goal of this meta-analysis was to compare the predictive effect of LNR for survival time in gastric cancer patients.
3.3. Subgroup and meta-regression analyses
2.4. Data synthesis and statistical analysis
To identify factors involved in the heterogeneity, we used metaregression and subgroup analyses. Subgroup analysis was conducted using variables that included race (Asian vs. non-Asian), treatment (R0 surgery + adjuvant therapy (AT) vs. R0 surgery), sample size, ratio of patients with stages I/II and III/IV gastric cancer, and lymph node numbers. Data in Table S3 shows that in nearly all subgroup analyses, higher LNRs also correlated with poor OS in gastric cancer patients. Taking treatment as an example, R0 surgery + AT results were similar to those studies in which treatment was only R0 surgery (HR = 2.15; 95% CI 1.64–2.83; p < 0.001 vs. HR = 1.82; 95% CI 1.59–2.09; p < 0.001). The heterogeneity I2s of the two groups were 79.5% and 79.5%, respectively. In Asian patients, HR = 1.88 (95% CI 1.61–2.20;
This meta-analysis was undertaken to obtain the association of survival time with LNR in gastric cancer patients. It was performed using pooled HRs and 95% CIs to assess prognosis in gastric cancer. We used a more conservative random effects model to balance the random effects of different studies rather than a fixed effects model [10]. We chose I2 and Q statistics to evaluate statistical heterogeneity. I2 was calculated to obtain the difference in total variance between the observed trials, in which I2 < 25% was considered low heterogeneity, while I2 > 75% was considered high heterogeneity [11]. Sensitivity analysis was used to assess the robustness and stability of the results, 77
International Journal of Surgery 57 (2018) 76–83
J. Zhu et al.
Fig. 1. Flow diagram of selection process evidence searches and inclusion.
I2 = 88%, p < 0.001), and in the other group, HR = 2.22 (95% CI 1.75–2.82; I2 = 79.6%; p < 0.001). Meta-regression analysis was used to investigate potential reasons for the origin of the heterogeneity. We performed single covariate regression with variables containing treatment, region, year, and duration. Results of the regression analysis suggested that treatment and region were the main causes of heterogeneity, with p values of 0.007 and 0.008, respectively. Finally, three studies [8] [15,16] provided data on the relationship between TNM staging and LNR in patients with gastric cancer. We divided gastric cancer patients into two groups based on stage, those with
stages I/II and those with stages III/IV, to assess the effect of LNR using the Chi-square test and meta-analysis based on dichotomous variables. Patients with higher LNRs tended to suffer from advanced gastric cancer. The p value of the chi-square test was significant, regardless of thresholds. Three studies [16] provided data that were used to analyze the association between differentiation and LNR in gastric cancer patients. Merge OR for differentiation and LNR 5 confidence intervals are on both sides of 1 suggested that the tumors with a higher LNR had a poor degree of differentiation. Meanwhile, we also performed a similar analysis on blood vessel number and lymphatic infiltration in gastric
Table 1 Summary table of HR (95% CI). study
HR (95%CI)
cutoff value
LNR stratifcation
Statistical analysis
study design
Inoue et al. [8] Marchet et al. [9] Xu et al. [26] Liu et al. [27] Sakcak et al. [19] Komats et al. [11] Zhang et al. [28] Alatengbaolide et al. [29] Bando et al. [30] Persiani et al. [31] Huang et al. [13] Celen et al. [32] Zhang et al. [33] Lee et al. [34] Fukuda et al. [35] Chen et al. [36] Asoglu et al. [37] Taghizadeh et al. [38] Zhao et al. [12] Chen et al. [39] Sianesi et al. [40] Pedrazzani et al. [41] Nitti et al. [42] Wang et al. [43] Xiao et al. [44] Liu et al. [45] Qiu et al. [46]
1.69(1.30,2.20) 1.41(1.19,1.67) 1.63(1.27,2.09) 1.89(1.04,3.45) 2.00(1.32,3.03) 2.77(1.87,4.10) 2.48(2.23,2.75) 1.42(1.27,1.59) 1.76(1.26,2.47) 2.27(1.50,3.43) 13.06(7.14,23.90) 11.40(4.97,26.15) 1.55(1.29,1.85) 1.07(0.59,1.95) 2.30(1.40,3.76) 1.55(1.19,2.01) 2.91(1.48,5.74) 2.77(1.51,5.08) 1.41(1.16,1.72) 1.96(1.49,2.57) 7.43(2.70,20.40) 1.59(1.21,2.09) 2.54(1.59,4.07) 1.74(1.25,2.42) 1.33(1.15,1.54) 2.55(1.64,3.98) 1.46(1.29,1.66)
0.25/0.5 0/0.01/0.1/0.25 0/0.01/0.1/0.25 0/0.4/0.8 0/0.01/0.1/0.25 0/0.2/0.4 0/0.01/0.1/0.25 0/0.01/0.1/0.25 0/0.01/0.1/0.25 0.15/0.4 0.1/0.2/0.3 0.01/0.1 0/0.25/0.5 0/0.01/0.05/0.1/0.2/0.3 0/0.01/0.2 0/0.01/0.1/0.25 0/0.01/0.1/0.25 0/0.33 0/0.15/0.4 0/0.13/0.4 0/0.01/0.1/0.25 0/0.01/0.25 0/0.01/0.1/0.25 0/0.3/0.5 0/0.01/0.3/0.5 0/0.1/0.4 0/0.3/0.6
Literature data Log rank analysis Log rank analysis Log rank analysis Literature data Log rank analysis Log rank analysis Log rank analysis NA Log rank analysis Log rank analysis Log rank analysis NA ROC curve analysis mean ROC curve analysis Log rank analysis median Log rank analysis NA Literature data Literature data Log rank analysis Log rank analysis ROC curve analysis Log rank analysis Log rank analysis
MA MA MA MA MA MA MA MA MA MA MA MA MA MA MA MA MA MA MA MA UA MA MA MA MA MA MA
R R R R R R R R R R R R R R R R R R R R R R R R R R R
R: retrospective; NA: not available; MA: multivariate statistical analysis. UA: univariate statistical analysis. 78
International Journal of Surgery 57 (2018) 76–83
J. Zhu et al.
Fig. 2. Forest plots and pooled estimates of the effect for meta-analysis of the association between LNR and overall survival in patients with gastric cancer. The effect size for meta-analysis of primary outcomes was the hazard ratio (HR) for overall survival.
Fig. 3. Sensitivity analysis of the association between LNR and overall survival in gastric cancer patients. Every horizontal line representing combined HR and the range of 95% CI after omitting study of the included studies one by one. 79
International Journal of Surgery 57 (2018) 76–83
J. Zhu et al.
Fig. 4. Begg's and egger's plots show bias between included studies. (A) Funnel plots showing association of LNR and OS in gastric cancer. Begg funnel plot suggest substantial asymmetry, the point represents the included study. The horizontal axis is the standard error, the ordinate axis is logHR. (B) Egger's regression asymmetry publication bias plot tend to suggest the presence of publication bias. The Egger test detects funnel plot asymmetry by determining whether the intercept deviates significantly from zero in a regression of the standardized effect estimates against their precision. (C) Risk of bias graph. The authors assess risk of bias item presented as percentages across all included studies, including high risk, low risk and unclear risk. (D) Risk of bias summary plot shows the risk for each of the included studies. The green point represents low risk, white point represents unclear risk, and the red is high risk. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)
4. Discussion
cancer patients. Results indicated that high LNRs suggested a higher degree of lymphatic vessels and vascular infiltration.
In this meta-analysis study, data collected from public databases were used to investigate the relationship between LNR and OS in gastric cancer. The survival rate in patients with low LNRs was significantly higher than that in patients with higher LNRs (HR = 1.99; 95% CI 1.74–2.27; p < 0.001). The results showed that there was a high heterogeneity detected between studies (I2 = 85.6%). Data from one independent study conducted by Huang et al. [17] demonstrated that there was significant heterogeneity detected among studies using sensitivity analysis. When we performed a subgroup analysis of some important clinicopathological features, the results were still similar. In addition, both sensitivity and regression analyses supported the stability of our meta-analysis. In large studies of prognostic factors in gastric cancer, it is generally
3.4. Publication bias Begg's and Egger's linear regression tests were conducted to assess publication bias in the meta-analysis. Results of Begg's and Egger's test showed significant publication bias (p value = 0.000 and 0.006, respectively) (See Fig. 4A and B). Next, we used the Cochrane risk bias assessment tool and evaluated the degree of bias risk including high risk, low risk, and unclear risk for each study. In addition, biased risk graphs were plotted to visually display the various types of bias in percentage form, as well as the proportion of various bias risks, respectively (Fig. 4C and D). 80
International Journal of Surgery 57 (2018) 76–83
J. Zhu et al.
majority (70%) of lymph node dissections involved less than 15 lymph nodes. Therefore, during the clinical evaluation, it is still important to retrieve a sufficient number of lymph nodes from surgical specimens. This study also has some limitations. First, we calculated HRs using raw data reported in the studies, rather than pooling adjusted HRs as reported in the individual studies, which would have been adjusted for some potential underlying differences between study patients. Second, although most of the included studies were published in high impact journals, there were study features could carry potential risk of bias. All articles included retrospective studies, and there was significant heterogeneity among studies. Unfortunately, in the meta-regression analysis, we did not find the cause of the heterogeneity. Other variables such as different experimental designs, individual responses to treatment, patients with different lifestyles, and other factors may also be reasons for the heterogeneity; however, we were not able to assess the heterogeneity of these factors due to a lack of related data. In this metaanalysis, the study of neoadjuvant therapy was excluded because the total number of detected lymph nodes and the number of positive lymph nodes were reported to decrease after preoperative radiotherapy [4,25], and this could also be related to the heterogeneity. Third, because this meta-analysis was based on the overall data retrieved from the literature and not on information obtained about each patient in every study, it may have introduced some additional errors. Furthermore, the cut-off values defined for each of the included studies were different, which may have generated more heterogeneity. Current evidence is still unanimous about which cut-off value is most reliable in predicting prognosis in gastric cancer patients. Wang et al. [4] studied a large cohort of gastric cancer patients using a median number of 10 lymph nodes, and concluded that 10 (1/15, 3/10, and 7/ 10) was the optimal cut-off value. However, this cut-off value may still not be the optimal value for other studies, and, therefore, it may be necessary to modify the cut-off value. Therefore, a larger cohort study may be required with LNR data to assess the prognosis of patients with gastric cancer and to identify differences between prognostic outcomes. Fourth, the current N-stage system in the AJCC international gastric cancer TNM staging (eighth edition of 2016) is still determined by the number of metastatic lymph nodes, but the effect of the number of anatomical lymph nodes on the number of metastatic lymph nodes was ignored. Therefore, lymph node metastasis classification of gastric cancer is controversy. The phenomenon of “staged migration” was observed when D2 lymph node dissection was incomplete, the accuracy of evaluating the prognosis of patients was affected. LNR can minimize the effects of stage migration caused by an increase in total resected lymph nodes; therefore LNR is increasingly considered to be an important additional prognostic factor for other malignancies such as esophageal, colon, and pancreatic cancers. This meta-analysis demonstrate that LNR was associated with prognosis in gastric cancer. A higher LNR was significantly related to a shorter OS. It is difficult to provide a gold standard for TNM staging; therefore, we expect LNR to be used as an independent prognostic indicator in gastric cancer patients and should be considered as a reliable prognostic parameter in future staging systems to help guide the clinical development of treatment plans.
considered that prognosis is related to clinicopathological factors (including tumor location, depth of invasion, lymph node metastasis, and others) and treatment (such as surgery), as well as lymph node dissection. Clinical and lymph node staging could help the clinician to accurately evaluate the course of the disease, develop an individualized comprehensive treatment program, and assess treatment and prognosis. At present, lymph node staging standards in gastric cancer are still in the initial phases. UICC and AJCC/TNM staging systems based on the number of positive lymph nodes dissected are used as predictors of gastric cancer prognosis [18]. But this classification has been shown to predict long-term outcomes [19]. In recent years, the rate of metastasis to the lymph nodes as a prognostic factor in gastric cancer has become a research focus. This is because of the simplicity in obtaining the lymph node metastasis rate (number of metastatic lymph nodes/number of resected lymph nodes), the reproducibility, and the reduced phase transfer phenomenon. Many studies have shown that the lymph node metastasis rate used in staging has a more predictive prognostic value than the AJCC, UICC/TNM, and JGCA staging systems, which can not only determine the number of metastatic lymph nodes, but also shed light on treatment efficacy. Patients with high LNR have a worse prognosis than patients with low LNR. The potential prognostic significance of LNR has also been evaluated in several studies of gastric cancer, suggesting that high MLNR is a poorer indicator of survival. LNR instead of the number of metastatic lymph nodes is considered as an independent prognostic factor for gastric cancer. We hypothesize that the ratio of metastatic lymph nodes can provide more accurate information on lymph node metastasis and tumor staging. Therefore, using LNR as a parameter to predict prognosis is more sensible. Some research studies have confirmed that LNR can be used as a prognostic factor in a variety of malignant tumors, especially in gastrointestinal cancer [2,20–22]. Zhao et al. [16] and a study of US SEER data validated the superiority of lymph node ratio as a staging factor. In this study, patients with the same pN stage had different numbers of lymph nodes and their overall survival was significantly different (P < 0.05). But in the same Nr staging patients, the overall survival rate is independent of the number of lymph nodes (P > 0.05) [4]. This finding suggests that the proposed Nr staging system can more accurately analyze the prognosis of patients with limited numbers of lymph node dissection than the existing AJCC TNM system. In clinical work, we found that for patients with lymph node metastasis, the same number of positive lymph node metastases, both TNM staging, such as 1/5 (+) (20%), the prognosis is worse than 1/20 (+) (5%). However, there was no difference in prognosis between different lymph node metastases. For example, 1/5 (+) (20%) and 4/20 (+) (20%) have no significant difference in prognosis. Our research has certain advantages. First, this is the first complete meta-analysis to quantify the role of LNR in the prognosis of gastric cancer. Second, this meta-analysis included a large number of primary studies (27 papers) and patients (11,441 patients with positive lymph nodes), which allowed for a more robust statistical analysis. Finally, our findings have demonstrated the importance of LNR in the prognosis of gastric cancer. Therefore, we recommend LNR as a prognostic parameter that should be included in a future gastric cancer staging system. It is noteworthy that the extent of the lymph node dissection can affect the LNR. For example, the number of lymph nodes removed and the depth of the surgical resection can affect the determination of cancer stage. Some studies have shown that when less than 15 lymph nodes are removed and lymph node dissection is limited to D1, the staging system will incur 5%–15% of the staging bias, and as a result, the pN staging system is not applicable [23,24]. Wang et al. [4] reported on the prognosis of 18,000 patients with gastric cancer in the Surveillance, Epidemiology and End Results (SEER) database, suggesting that there was significantly less bias (12%) using LNR staging compared with 57% bias resulting from TNM staging. The reason for the large bias using TNM staging was that the
5. Conclusions In conclusion, results from this meta-analysis of 27 studies suggest that a higher LNR was significantly related to a shorter OS in gastric cancer patients, even when subgroup analysis was performed using all the different factors. This study provides additional evidence that LNR could be an independent prognostic indicator in gastric cancer patients and used as a parameter in future staging systems. Ethical approval No. 81
International Journal of Surgery 57 (2018) 76–83
J. Zhu et al.
Sources of funding
References
Study Abroad Returned Science Foundation of Hei Longjiang Province [grand number LC2013C27].
[1] A. Ferro, et al., Worldwide trends in gastric cancer mortality (1980-2011), with predictions to 2015, and incidence by subtype, Eur. J. Canc. 50 (7) (2014) 1330–1344. [2] O.C. Kutlu, M. Watchell, S. Dissanaike, Metastatic lymph node ratio successfully predicts prognosis in western gastric cancer patients, Surg Oncol 24 (2) (2015) 84–88. [3] W. Attaallah, et al., Prognostic impact of the metastatic lymph node ratio on survival in gastric cancer, Indian J Surg Oncol 7 (1) (2016) 67–72. [4] J. Wang, et al., Comparison of a lymph node ratio-based staging system with the 7th AJCC system for gastric cancer: analysis of 18,043 patients from the SEER database, Ann. Surg. 255 (3) (2012) 478–485. [5] R. Zhou, et al., Comparison of three lymph node classifications for survival prediction in distant metastatic gastric cancer, Int. J. Surg. 35 (2016) 165–171. [6] J.W. Lee, et al., Different lymph node staging systems in patients with gastric cancer from Korean: what is the best prognostic assessment tool? Medicine (Baltim.) 95 (25) (2016) e3860. [7] N. McGauran, et al., Reporting bias in medical research - a narrative review, Trials 11 (2010) 37. [8] K. Inoue, et al., The superiority of ratio-based lymph node staging in gastric carcinoma, Ann. Surg Oncol. 9 (1) (2002) 27–34. [9] A. Marchet, et al., The prognostic value of N-ratio in patients with gastric cancer: validation in a large, multicenter series, Eur. J. Surg. Oncol. 34 (2) (2008) 159–165. [10] R. DerSimonian, N. Laird, Meta-analysis in clinical trials revisited, Contemp. Clin. Trials 45 (Pt A) (2015) 139–145. [11] J.P. Higgins, et al., Measuring inconsistency in meta-analyses, BMJ 327 (7414) (2003) 557–560. [12] C.B. Begg, M. Mazumdar, Operating characteristics of a rank correlation test for publication bias, Biometrics 50 (4) (1994) 1088–1101. [13] M. Egger, et al., Bias in meta-analysis detected by a simple, graphical test, BMJ 315 (7109) (1997) 629–634. [14] M.K. Parmar, L.A. Stewart, D.G. Altman, Meta-analyses of randomised trials: when the whole is more than just the sum of the parts, Br. J. Canc. 74 (4) (1996) 496–501. [15] S. Komatsu, et al., Evaluation of prognostic value and stage migration effect using positive lymph node ratio in gastric cancer, Eur. J. Surg. Oncol. 43 (1) (2017) 203–209. [16] L.Y. Zhao, et al., Superiority of lymph node ratio-based staging system for prognostic prediction in 2575 patients with gastric cancer: validation analysis in a large single center, Oncotarget 7 (32) (2016) 51069–51081. [17] C.M. Huang, et al., Prognostic impact of metastatic lymph node ratio in advanced gastric cancer from cardia and fundus, World J. Gastroenterol. 14 (27) (2008) 4383–4388. [18] C.J. Yu, et al., Prognostic value of metastatic lymph node ratio for patients with gastric cancer, Zhonghua Wei Chang Wai Ke Za Zhi 14 (7) (2011) 516–519. [19] J. Deng, et al., Suitability of 7th UICC N stage for predicting the overall survival of gastric cancer patients after curative resection in China, Ann. Surg Oncol. 17 (5) (2010) 1259–1266. [20] J. Zhou, et al., Lymph node ratio may predict the benefit of postoperative radiotherapy in node-positive cervical cancer, Oncotarget 7 (20) (2016) 29420–29428. [21] A. Isik, et al., A new prognostic strategy for gastric carcinoma: albumin level and metastatic lymph node ratio, Minerva Chir. 69 (3) (2014) 147–153. [22] G. Sun, et al., Lymph node ratio is a prognostic factor for non-small cell lung cancer, Oncotarget 6 (32) (2015) 33912–33918. [23] I. Sakcak, et al., Does N ratio affect survival in D1 and D2 lymph node dissection for gastric cancer? World J. Gastroenterol. 17 (35) (2011) 4007–4012. [24] Z. Sun, et al., The impact of N-ratio in minimizing stage migration phenomenon in gastric cancer patients with insufficient number or level of lymph node retrieved: results from a Chinese mono-institutional study in 2159 patients, Ann. Oncol. 20 (5) (2009) 897–905. [25] S. Ito, et al., Neoadjuvant chemotherapy followed by surgery in gastric cancer patients with extensive lymph node metastasis, World J. Clin. Oncol. 6 (6) (2015) 291–294. [26] D.Z. Xu, et al., Positive lymph node ratio is an independent prognostic factor in gastric cancer after d2 resection regardless of the examined number of lymph nodes, Ann. Surg Oncol. 16 (2) (2009) 319–326. [27] C. Liu, et al., Clinical implications of metastatic lymph node ratio in gastric cancer, BMC Canc. 7 (2007) 200. [28] B.Y. Zhang, et al., Evaluation of the prognostic value of the metastatic lymph node ratio for gastric cancer, Am. J. Surg. 207 (4) (2014) 555–565. [29] Alatengbaolide, et al., Lymph node ratio is an independent prognostic factor in gastric cancer after curative resection (R0) regardless of the examined number of lymph nodes, Am. J. Clin. Oncol. 36 (4) (2013) 325–330. [30] E. Bando, et al., Outcome of ratio of lymph node metastasis in gastric carcinoma, Ann. Surg Oncol. 9 (8) (2002) 775–784. [31] R. Persiani, et al., Metastatic lymph node ratio: a new staging system for gastric cancer, World J. Surg. 33 (10) (2009) 2106–2111. [32] O. Celen, E. Yildirim, U. Berberoglu, Prognostic impact of positive lymph node ratio in gastric carcinoma, J. Surg. Oncol. 96 (2) (2007) 95–101. [33] M. Zhang, et al., Prognostic significance of metastatic lymph nodes ratio in patients with gastric adenocarcinoma after curative gastrectomy, Chin Med J (Engl) 127 (10) (2014) 1874–1878. [34] S.R. Lee, et al., Prognostic significance of the metastatic lymph node ratio in patients with gastric cancer, World J. Surg. 36 (5) (2012) 1096–1101. [35] N. Fukuda, et al., Prognostic significance of the metastatic lymph node ratio in
Author contribution Specific author contributions: Yan Zhang and Haibo Lu provided experimental design; Jiang Zhu and Zhao Xue collected and analyzed the data; Jiang Zhu drafted the manuscript; Yan Zhang and Haibo Lu contributed to revision the manuscript; All authors commented on drafts of the paper and final approval of the manuscript. Conflicts of interest No. Trial registry number reviewregistry497. Guarantor Guarantors of the article: Jiang Zhu, Zhao Xue, Yan Zhang, Haibo Lu. Provenance and peer review Not commissioned, externally peer reviewed. Acknowledgments We thank Mark Abramovitz, PhD, from Liwen Bianji, Edanz Group China (www.liwenbianji.cn/ac), for editing the English text of a draft of this manuscript. Abbreviation In alphabetical order AJCC UICC TNM classification AT adjuvant therapy CI confidence interval DFS disease-free survival HR hazard ratio LNR lymph node ratio MA multivariate statistical analysis NA not available No. of nodes (N+) total number of lymph nodes harvested (number of positive lymph nodes) NR not report OS overall survival PRISMA Preferred reporting items for systematic reviews and metaanalyses TNM tumor node metastasis R retrospective SEER Surverillance, Epidmiology and End Results UA univariate statistical analysis Appendix A. Supplementary data Supplementary data related to this article can be found at https:// doi.org/10.1016/j.ijsu.2018.08.002. 82
International Journal of Surgery 57 (2018) 76–83
J. Zhu et al.
gastric cancer patients, World J. Surg. 33 (11) (2009) 2378–2382. [36] C. Jian-Hui, et al., Prognostic value of three different lymph node staging systems in the survival of patients with gastric cancer following D2 lymphadenectomy, Tumour Biol 37 (8) (2016) 11105–11113. [37] O. Asoglu, et al., Metastatic lymph node ratio is an independent prognostic factor in gastric cancer, Hepato-Gastroenterology 56 (91–92) (2009) 908–913. [38] A. Taghizadeh-Kermani, et al., Prognostic significance of metastatic lymph node ratio in patients with gastric cancer: an evaluation in north-East of Iran, Iran. J. Cancer Prev. 7 (2) (2014) 73–79. [39] S. Chen, et al., The prognostic value of harvested lymph nodes and the metastatic lymph node ratio for gastric cancer patients: results of a study of 1,101 patients, PLoS One 7 (11) (2012) e49424. [40] M. Sianesi, et al., The node ratio as prognostic factor after curative resection for gastric cancer, J. Gastrointest. Surg. 14 (4) (2010) 614–619. [41] C. Pedrazzani, et al., Ratio between metastatic and examined lymph nodes (N ratio) may have low clinical utility in gastric cancer patients treated by limited
[42]
[43]
[44] [45] [46]
83
lymphadenectomy: results from a single-center experience of 526 patients, World J. Surg. 34 (1) (2010) 85–91. D. Nitti, et al., Ratio between metastatic and examined lymph nodes is an independent prognostic factor after D2 resection for gastric cancer: analysis of a large European monoinstitutional experience, Ann. Surg Oncol. 10 (9) (2003) 1077–1085. X. Wang, et al., Prognostic value of the ratio of metastatic lymph nodes in gastric cancer: an analysis based on a Chinese population, J. Surg. Oncol. 99 (6) (2009) 329–334. L.B. Xiao, et al., Superiority of metastatic lymph node ratio to the 7th edition UICC N staging in gastric cancer, World J. Gastroenterol. 17 (46) (2011) 5123–5130. H. Liu, et al., The RML of lymph node metastasis was superior to the LODDS for evaluating the prognosis of gastric cancer, Int. J. Surg. 11 (5) (2013) 419–424. M.Z. Qiu, et al., The tumor-log odds of positive lymph nodes-metastasis staging system, a promising new staging system for gastric cancer after D2 resection in China, PLoS One 7 (2) (2012) e31736.
Contents lists available at ScienceDirect
International Journal of Surgery journal homepage: www.elsevier.com/locate/ijsu
Review
Integrated analysis of the prognostic role of the lymph node ratio in nodepositive gastric cancer: A meta-analysis
T
Jiang Zhua,1, Zhao Xueb,1, Shumei Zhanga, Xinxin Guob, Laihui Zhaib, Shipeng Shanga, Yan Zhanga,∗∗, Haibo Lub,∗ a b
College of Bioinformatics Science and Technology, Harbin Medical University, Harbin, 150081, China Department of Gastrointestinal Medical Oncology, Harbin Medical University Cancer Hospital, Harbin, 150040, China
A R T I C LE I N FO
A B S T R A C T
Keywords: Lymph node positive ratio Meta-analysis Overall survival
Background: The lymph node ratio (LNR) as a prognostic parameter for gastric cancer has yet to be fully validated in the current tumor node metastasis staging system. We assessed the prognostic role of LNR in lymph node-positive gastric cancer through a meta-analysis. Materials and methods: PubMed and EMBASE were searched for relevant studies up until December 2016. The effect measure for meta-analysis of primary outcomes was the hazard ratio (HR) for overall survival. Pooled HRs and 95% confidence intervals were calculated using random effects models. The I2 statistic was used to measure heterogeneity. Subgroup analysis and meta-regression were chosen to illustrate the potential heterogeneity of the risk factors of outcomes. Publication bias was assessed using Egger's test and Begg's funnel plots. Sensitivity analysis was applied to evaluate the origin of the heterogeneity. Results: We included 27 studies in this meta-analysis. Higher LNRs were significantly associated with a shorter overall survival (OS). High heterogeneity among the studies was identified (I2 = 85.6), and the publication bias was moderate. Subgroup analysis showed similar results, and elevated LNR was associated with late-stage gastric cancer and indicative of a worse prognosis. Univariate meta-regression analysis of OS indicated that both treatment type and ethnicity may be causes of heterogeneity in patients with gastric cancer (p values were 0.005 and 0.008, respectively). Conclusion: LNR was associated with a significantly poorer OS and LNR was an independent predictor of survival in patients with gastric cancer. LNR should be added as one of the parameters to be used in future tumor staging classification systems.
1. Introduction Gastric cancer is the fourth most common cancer and the third leading cause of cancer-related death throughout the world [1]. Surgery is the main treatment in gastric cancer, and lymph node metastasis is the most effective predictor of postoperative survival [2]. Therefore, lymph node status is considered to be one of the key prognostic factors in gastric cancer. The AJCC (UICC TNM classification) staging system of malignant tumors showed that proper pathological lymph node staging (pN stage) requires at least 15 lymph nodes to be cleared [3]. Certain factors have led to an insufficient number of lymph nodes being dissected in clinical practice. Increasing the number of lymph nodes dissected has been
found to affect the number of metastases detected based on standard pN staging, which may result in a change in the TMN classification of a cancer that can affect the accuracy of the prognostic prediction [4–6]. The classic staging system for gastric cancer is the TNM staging system. However, the TNM staging system is not used when checking for tumor-free lymph nodes. Thus, the lymph node ratio (LNR) is considered to be an essential prognostic factor and a suitable staging method for patients with positive lymph nodes [7]. However, its use as a prognostic factor is controversial because of conflicting LNR results and differences in study design and sample size used in previous studies. Previous systematic reviews suggested that LNR can be used as a prognostic factor for gastric cancer and colorectal cancer [8,9]. Nevertheless, to date there has been no formal meta-analysis focused on
∗
Corresponding author. Corresponding author. E-mail addresses: [email protected] (Y. Zhang), [email protected] (H. Lu). 1 These authors contributed equally to this work. ∗∗
https://doi.org/10.1016/j.ijsu.2018.08.002 Received 14 May 2018; Received in revised form 25 July 2018; Accepted 6 August 2018 Available online 10 August 2018 1743-9191/ © 2018 Published by Elsevier Ltd on behalf of IJS Publishing Group Ltd.
International Journal of Surgery 57 (2018) 76–83
J. Zhu et al.
calculating the heterogeneity in each situation in which a single study was removed in turn in order to evaluate the effect of a single study on the overall outcome. Subgroup analysis was performed according to each parameter including year, race, treatment, sample size, ratio of patients with stage I/II and stage III/IV gastric cancer, and number of lymph nodes. We compared the pooled HR estimates from different subgroups using an interaction test. A meta-regression model was developed to explore the potential impact of different factors on heterogeneity, and to assess the effect of year, race, treatment, and other risk factors or potential confounding factors on outcome. Finally, Begg's test and Egger's test were used to assess publication bias. In addition, a bias risk test chart was made to assess the risk of various biases and the bias of each article using Revman 5.2 software [12,13]. Statistical analysis was performed using STATA12.0 software, according to the Cochrane Collaboration Organization and the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-analyses) guidelines. A p value < 0.05 indicated a statistically significant difference.
the prognostic significance of LNR in gastric cancer. In this study, the first comprehensive systematic review was conducted to investigate the prognostic role of LNR in patients with lymph node-positive gastric cancer, and the results demonstrated that an increased LNR correlated with poor overall survival (OS) in gastric cancer patients. 2. Materials and Methods 2.1. Search strategy A comprehensive, computerized literature search was conducted in PubMed and Embase for relevant studies up to December 2016. The search was performed using the following text words and corresponding medical subject heading terms: ((gastric cancer) OR (stomach cancer) OR (gastric carcinoma) OR (stomach carcinoma) OR (gastric neoplasm) OR (stomach neoplasm) OR (lymph node ratio)) AND ((LNR) OR (lymph positive node ratio) OR (lymph metastatic node ratio)). The search strategy was repeatedly performed until no new relevant articles were found. In addition, we reviewed references in the retrieved articles to search for additional relevant studies. All articles were evaluated by two authors based on the eligibility criteria we designed.
3. Results 3.1. Study selection and characteristics A flow diagram of the literature selection process used in this study is shown in Fig. 1. A total of 1208 articles were initially retrieved from PubMed and EMBASE according to particular keywords. After manually screening and filtering these, 27 articles were eventually chosen that included a total of 11,441 patients. Characteristics of the included studies are summarized in Table S1 and Table S2. All patients underwent radical surgery, and the median age of patients in the studies ranged from 54 to 72 years old. Of these studies, 17 involved Asian patients and 10 involved non-Asian patients. Lymph node data and survival time of patients were also retrieved. Median follow-up time in the studies ranged from 52 to 75.3 months. HRs and 95% CIs obtained from all articles are listed in Table 1. HRs and 95% CIs of OS could be directly obtained from 12 studies. The other 15 studies did not provide HRs and 95% CIs directly; therefore, we used relative data such as Kaplan–Meier curves and the total number of survivors to calculate HRs and 95% CIs [14].
2.2. Study selection First, we checked titles and abstracts of articles that were searched using keywords to exclude irrelevant articles. Next, all retained studies were screened according to inclusion and exclusion criteria. The following conditions were used to evaluate whether the study was included: (a) all patients were diagnosed with gastric cancer using pathology; (b) patients underwent radical surgery (R0 resection); (c) the outcome of interest was disease-free survival (DFS) and OS; and (d) hazard ratios (HRs) and 95% confidence intervals (CIs) used to evaluate prognosis could be extracted from the original literature, including direct acquisition or approximation by calculation. Exclusion criteria were defined as follows: (a) gastric cancer patients with other tumors or patients with distant metastases (TNM IV staging); (b) articles including patients undergoing tumor-related neoadjuvant chemotherapy before surgery; (c) articles consisting of letters, meeting summaries, commentary articles, and posters; and (d) studies could not provide outcome data and did not calculate the necessary results.
3.2. Primary outcomes and sensitivity analysis Using a random effects model, pooled results of HR and OS statistics from 27 studies indicated that there was a significant association between the LNR and OS in patients with gastric cancer. The survival rate was significantly higher in patients with lower LNRs (HR = 1.99; 95% CI 1.74–2.27; p < 0.001) (Fig. 2). The results also showed that there was high heterogeneity among studies (I2 = 85.6%; p < 0.001). Subsequently, sensitivity analysis was performed to evaluate the stability of the model by omitting each individual study and calculating new HRs. Results showed that HRs were relatively stable and that study heterogeneity was still apparent (Fig. 3).
2.3. Data extraction Data were extracted independently by two researchers and any discrepancies in the data were settled by consensus. If necessary, a third researcher was expected to participate in the discussion and make a decision. The main data extracted from each study included: first author, publication year, number of patients, country of the study population, study design, duration of follow-up, patient age, gender, number of checked nodes, type of study, cut-off value of the LNR and the definition of stratification, and HRs and 95% CIs. The primary goal of this meta-analysis was to compare the predictive effect of LNR for survival time in gastric cancer patients.
3.3. Subgroup and meta-regression analyses
2.4. Data synthesis and statistical analysis
To identify factors involved in the heterogeneity, we used metaregression and subgroup analyses. Subgroup analysis was conducted using variables that included race (Asian vs. non-Asian), treatment (R0 surgery + adjuvant therapy (AT) vs. R0 surgery), sample size, ratio of patients with stages I/II and III/IV gastric cancer, and lymph node numbers. Data in Table S3 shows that in nearly all subgroup analyses, higher LNRs also correlated with poor OS in gastric cancer patients. Taking treatment as an example, R0 surgery + AT results were similar to those studies in which treatment was only R0 surgery (HR = 2.15; 95% CI 1.64–2.83; p < 0.001 vs. HR = 1.82; 95% CI 1.59–2.09; p < 0.001). The heterogeneity I2s of the two groups were 79.5% and 79.5%, respectively. In Asian patients, HR = 1.88 (95% CI 1.61–2.20;
This meta-analysis was undertaken to obtain the association of survival time with LNR in gastric cancer patients. It was performed using pooled HRs and 95% CIs to assess prognosis in gastric cancer. We used a more conservative random effects model to balance the random effects of different studies rather than a fixed effects model [10]. We chose I2 and Q statistics to evaluate statistical heterogeneity. I2 was calculated to obtain the difference in total variance between the observed trials, in which I2 < 25% was considered low heterogeneity, while I2 > 75% was considered high heterogeneity [11]. Sensitivity analysis was used to assess the robustness and stability of the results, 77
International Journal of Surgery 57 (2018) 76–83
J. Zhu et al.
Fig. 1. Flow diagram of selection process evidence searches and inclusion.
I2 = 88%, p < 0.001), and in the other group, HR = 2.22 (95% CI 1.75–2.82; I2 = 79.6%; p < 0.001). Meta-regression analysis was used to investigate potential reasons for the origin of the heterogeneity. We performed single covariate regression with variables containing treatment, region, year, and duration. Results of the regression analysis suggested that treatment and region were the main causes of heterogeneity, with p values of 0.007 and 0.008, respectively. Finally, three studies [8] [15,16] provided data on the relationship between TNM staging and LNR in patients with gastric cancer. We divided gastric cancer patients into two groups based on stage, those with
stages I/II and those with stages III/IV, to assess the effect of LNR using the Chi-square test and meta-analysis based on dichotomous variables. Patients with higher LNRs tended to suffer from advanced gastric cancer. The p value of the chi-square test was significant, regardless of thresholds. Three studies [16] provided data that were used to analyze the association between differentiation and LNR in gastric cancer patients. Merge OR for differentiation and LNR 5 confidence intervals are on both sides of 1 suggested that the tumors with a higher LNR had a poor degree of differentiation. Meanwhile, we also performed a similar analysis on blood vessel number and lymphatic infiltration in gastric
Table 1 Summary table of HR (95% CI). study
HR (95%CI)
cutoff value
LNR stratifcation
Statistical analysis
study design
Inoue et al. [8] Marchet et al. [9] Xu et al. [26] Liu et al. [27] Sakcak et al. [19] Komats et al. [11] Zhang et al. [28] Alatengbaolide et al. [29] Bando et al. [30] Persiani et al. [31] Huang et al. [13] Celen et al. [32] Zhang et al. [33] Lee et al. [34] Fukuda et al. [35] Chen et al. [36] Asoglu et al. [37] Taghizadeh et al. [38] Zhao et al. [12] Chen et al. [39] Sianesi et al. [40] Pedrazzani et al. [41] Nitti et al. [42] Wang et al. [43] Xiao et al. [44] Liu et al. [45] Qiu et al. [46]
1.69(1.30,2.20) 1.41(1.19,1.67) 1.63(1.27,2.09) 1.89(1.04,3.45) 2.00(1.32,3.03) 2.77(1.87,4.10) 2.48(2.23,2.75) 1.42(1.27,1.59) 1.76(1.26,2.47) 2.27(1.50,3.43) 13.06(7.14,23.90) 11.40(4.97,26.15) 1.55(1.29,1.85) 1.07(0.59,1.95) 2.30(1.40,3.76) 1.55(1.19,2.01) 2.91(1.48,5.74) 2.77(1.51,5.08) 1.41(1.16,1.72) 1.96(1.49,2.57) 7.43(2.70,20.40) 1.59(1.21,2.09) 2.54(1.59,4.07) 1.74(1.25,2.42) 1.33(1.15,1.54) 2.55(1.64,3.98) 1.46(1.29,1.66)
0.25/0.5 0/0.01/0.1/0.25 0/0.01/0.1/0.25 0/0.4/0.8 0/0.01/0.1/0.25 0/0.2/0.4 0/0.01/0.1/0.25 0/0.01/0.1/0.25 0/0.01/0.1/0.25 0.15/0.4 0.1/0.2/0.3 0.01/0.1 0/0.25/0.5 0/0.01/0.05/0.1/0.2/0.3 0/0.01/0.2 0/0.01/0.1/0.25 0/0.01/0.1/0.25 0/0.33 0/0.15/0.4 0/0.13/0.4 0/0.01/0.1/0.25 0/0.01/0.25 0/0.01/0.1/0.25 0/0.3/0.5 0/0.01/0.3/0.5 0/0.1/0.4 0/0.3/0.6
Literature data Log rank analysis Log rank analysis Log rank analysis Literature data Log rank analysis Log rank analysis Log rank analysis NA Log rank analysis Log rank analysis Log rank analysis NA ROC curve analysis mean ROC curve analysis Log rank analysis median Log rank analysis NA Literature data Literature data Log rank analysis Log rank analysis ROC curve analysis Log rank analysis Log rank analysis
MA MA MA MA MA MA MA MA MA MA MA MA MA MA MA MA MA MA MA MA UA MA MA MA MA MA MA
R R R R R R R R R R R R R R R R R R R R R R R R R R R
R: retrospective; NA: not available; MA: multivariate statistical analysis. UA: univariate statistical analysis. 78
International Journal of Surgery 57 (2018) 76–83
J. Zhu et al.
Fig. 2. Forest plots and pooled estimates of the effect for meta-analysis of the association between LNR and overall survival in patients with gastric cancer. The effect size for meta-analysis of primary outcomes was the hazard ratio (HR) for overall survival.
Fig. 3. Sensitivity analysis of the association between LNR and overall survival in gastric cancer patients. Every horizontal line representing combined HR and the range of 95% CI after omitting study of the included studies one by one. 79
International Journal of Surgery 57 (2018) 76–83
J. Zhu et al.
Fig. 4. Begg's and egger's plots show bias between included studies. (A) Funnel plots showing association of LNR and OS in gastric cancer. Begg funnel plot suggest substantial asymmetry, the point represents the included study. The horizontal axis is the standard error, the ordinate axis is logHR. (B) Egger's regression asymmetry publication bias plot tend to suggest the presence of publication bias. The Egger test detects funnel plot asymmetry by determining whether the intercept deviates significantly from zero in a regression of the standardized effect estimates against their precision. (C) Risk of bias graph. The authors assess risk of bias item presented as percentages across all included studies, including high risk, low risk and unclear risk. (D) Risk of bias summary plot shows the risk for each of the included studies. The green point represents low risk, white point represents unclear risk, and the red is high risk. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)
4. Discussion
cancer patients. Results indicated that high LNRs suggested a higher degree of lymphatic vessels and vascular infiltration.
In this meta-analysis study, data collected from public databases were used to investigate the relationship between LNR and OS in gastric cancer. The survival rate in patients with low LNRs was significantly higher than that in patients with higher LNRs (HR = 1.99; 95% CI 1.74–2.27; p < 0.001). The results showed that there was a high heterogeneity detected between studies (I2 = 85.6%). Data from one independent study conducted by Huang et al. [17] demonstrated that there was significant heterogeneity detected among studies using sensitivity analysis. When we performed a subgroup analysis of some important clinicopathological features, the results were still similar. In addition, both sensitivity and regression analyses supported the stability of our meta-analysis. In large studies of prognostic factors in gastric cancer, it is generally
3.4. Publication bias Begg's and Egger's linear regression tests were conducted to assess publication bias in the meta-analysis. Results of Begg's and Egger's test showed significant publication bias (p value = 0.000 and 0.006, respectively) (See Fig. 4A and B). Next, we used the Cochrane risk bias assessment tool and evaluated the degree of bias risk including high risk, low risk, and unclear risk for each study. In addition, biased risk graphs were plotted to visually display the various types of bias in percentage form, as well as the proportion of various bias risks, respectively (Fig. 4C and D). 80
International Journal of Surgery 57 (2018) 76–83
J. Zhu et al.
majority (70%) of lymph node dissections involved less than 15 lymph nodes. Therefore, during the clinical evaluation, it is still important to retrieve a sufficient number of lymph nodes from surgical specimens. This study also has some limitations. First, we calculated HRs using raw data reported in the studies, rather than pooling adjusted HRs as reported in the individual studies, which would have been adjusted for some potential underlying differences between study patients. Second, although most of the included studies were published in high impact journals, there were study features could carry potential risk of bias. All articles included retrospective studies, and there was significant heterogeneity among studies. Unfortunately, in the meta-regression analysis, we did not find the cause of the heterogeneity. Other variables such as different experimental designs, individual responses to treatment, patients with different lifestyles, and other factors may also be reasons for the heterogeneity; however, we were not able to assess the heterogeneity of these factors due to a lack of related data. In this metaanalysis, the study of neoadjuvant therapy was excluded because the total number of detected lymph nodes and the number of positive lymph nodes were reported to decrease after preoperative radiotherapy [4,25], and this could also be related to the heterogeneity. Third, because this meta-analysis was based on the overall data retrieved from the literature and not on information obtained about each patient in every study, it may have introduced some additional errors. Furthermore, the cut-off values defined for each of the included studies were different, which may have generated more heterogeneity. Current evidence is still unanimous about which cut-off value is most reliable in predicting prognosis in gastric cancer patients. Wang et al. [4] studied a large cohort of gastric cancer patients using a median number of 10 lymph nodes, and concluded that 10 (1/15, 3/10, and 7/ 10) was the optimal cut-off value. However, this cut-off value may still not be the optimal value for other studies, and, therefore, it may be necessary to modify the cut-off value. Therefore, a larger cohort study may be required with LNR data to assess the prognosis of patients with gastric cancer and to identify differences between prognostic outcomes. Fourth, the current N-stage system in the AJCC international gastric cancer TNM staging (eighth edition of 2016) is still determined by the number of metastatic lymph nodes, but the effect of the number of anatomical lymph nodes on the number of metastatic lymph nodes was ignored. Therefore, lymph node metastasis classification of gastric cancer is controversy. The phenomenon of “staged migration” was observed when D2 lymph node dissection was incomplete, the accuracy of evaluating the prognosis of patients was affected. LNR can minimize the effects of stage migration caused by an increase in total resected lymph nodes; therefore LNR is increasingly considered to be an important additional prognostic factor for other malignancies such as esophageal, colon, and pancreatic cancers. This meta-analysis demonstrate that LNR was associated with prognosis in gastric cancer. A higher LNR was significantly related to a shorter OS. It is difficult to provide a gold standard for TNM staging; therefore, we expect LNR to be used as an independent prognostic indicator in gastric cancer patients and should be considered as a reliable prognostic parameter in future staging systems to help guide the clinical development of treatment plans.
considered that prognosis is related to clinicopathological factors (including tumor location, depth of invasion, lymph node metastasis, and others) and treatment (such as surgery), as well as lymph node dissection. Clinical and lymph node staging could help the clinician to accurately evaluate the course of the disease, develop an individualized comprehensive treatment program, and assess treatment and prognosis. At present, lymph node staging standards in gastric cancer are still in the initial phases. UICC and AJCC/TNM staging systems based on the number of positive lymph nodes dissected are used as predictors of gastric cancer prognosis [18]. But this classification has been shown to predict long-term outcomes [19]. In recent years, the rate of metastasis to the lymph nodes as a prognostic factor in gastric cancer has become a research focus. This is because of the simplicity in obtaining the lymph node metastasis rate (number of metastatic lymph nodes/number of resected lymph nodes), the reproducibility, and the reduced phase transfer phenomenon. Many studies have shown that the lymph node metastasis rate used in staging has a more predictive prognostic value than the AJCC, UICC/TNM, and JGCA staging systems, which can not only determine the number of metastatic lymph nodes, but also shed light on treatment efficacy. Patients with high LNR have a worse prognosis than patients with low LNR. The potential prognostic significance of LNR has also been evaluated in several studies of gastric cancer, suggesting that high MLNR is a poorer indicator of survival. LNR instead of the number of metastatic lymph nodes is considered as an independent prognostic factor for gastric cancer. We hypothesize that the ratio of metastatic lymph nodes can provide more accurate information on lymph node metastasis and tumor staging. Therefore, using LNR as a parameter to predict prognosis is more sensible. Some research studies have confirmed that LNR can be used as a prognostic factor in a variety of malignant tumors, especially in gastrointestinal cancer [2,20–22]. Zhao et al. [16] and a study of US SEER data validated the superiority of lymph node ratio as a staging factor. In this study, patients with the same pN stage had different numbers of lymph nodes and their overall survival was significantly different (P < 0.05). But in the same Nr staging patients, the overall survival rate is independent of the number of lymph nodes (P > 0.05) [4]. This finding suggests that the proposed Nr staging system can more accurately analyze the prognosis of patients with limited numbers of lymph node dissection than the existing AJCC TNM system. In clinical work, we found that for patients with lymph node metastasis, the same number of positive lymph node metastases, both TNM staging, such as 1/5 (+) (20%), the prognosis is worse than 1/20 (+) (5%). However, there was no difference in prognosis between different lymph node metastases. For example, 1/5 (+) (20%) and 4/20 (+) (20%) have no significant difference in prognosis. Our research has certain advantages. First, this is the first complete meta-analysis to quantify the role of LNR in the prognosis of gastric cancer. Second, this meta-analysis included a large number of primary studies (27 papers) and patients (11,441 patients with positive lymph nodes), which allowed for a more robust statistical analysis. Finally, our findings have demonstrated the importance of LNR in the prognosis of gastric cancer. Therefore, we recommend LNR as a prognostic parameter that should be included in a future gastric cancer staging system. It is noteworthy that the extent of the lymph node dissection can affect the LNR. For example, the number of lymph nodes removed and the depth of the surgical resection can affect the determination of cancer stage. Some studies have shown that when less than 15 lymph nodes are removed and lymph node dissection is limited to D1, the staging system will incur 5%–15% of the staging bias, and as a result, the pN staging system is not applicable [23,24]. Wang et al. [4] reported on the prognosis of 18,000 patients with gastric cancer in the Surveillance, Epidemiology and End Results (SEER) database, suggesting that there was significantly less bias (12%) using LNR staging compared with 57% bias resulting from TNM staging. The reason for the large bias using TNM staging was that the
5. Conclusions In conclusion, results from this meta-analysis of 27 studies suggest that a higher LNR was significantly related to a shorter OS in gastric cancer patients, even when subgroup analysis was performed using all the different factors. This study provides additional evidence that LNR could be an independent prognostic indicator in gastric cancer patients and used as a parameter in future staging systems. Ethical approval No. 81
International Journal of Surgery 57 (2018) 76–83
J. Zhu et al.
Sources of funding
References
Study Abroad Returned Science Foundation of Hei Longjiang Province [grand number LC2013C27].
[1] A. Ferro, et al., Worldwide trends in gastric cancer mortality (1980-2011), with predictions to 2015, and incidence by subtype, Eur. J. Canc. 50 (7) (2014) 1330–1344. [2] O.C. Kutlu, M. Watchell, S. Dissanaike, Metastatic lymph node ratio successfully predicts prognosis in western gastric cancer patients, Surg Oncol 24 (2) (2015) 84–88. [3] W. Attaallah, et al., Prognostic impact of the metastatic lymph node ratio on survival in gastric cancer, Indian J Surg Oncol 7 (1) (2016) 67–72. [4] J. Wang, et al., Comparison of a lymph node ratio-based staging system with the 7th AJCC system for gastric cancer: analysis of 18,043 patients from the SEER database, Ann. Surg. 255 (3) (2012) 478–485. [5] R. Zhou, et al., Comparison of three lymph node classifications for survival prediction in distant metastatic gastric cancer, Int. J. Surg. 35 (2016) 165–171. [6] J.W. Lee, et al., Different lymph node staging systems in patients with gastric cancer from Korean: what is the best prognostic assessment tool? Medicine (Baltim.) 95 (25) (2016) e3860. [7] N. McGauran, et al., Reporting bias in medical research - a narrative review, Trials 11 (2010) 37. [8] K. Inoue, et al., The superiority of ratio-based lymph node staging in gastric carcinoma, Ann. Surg Oncol. 9 (1) (2002) 27–34. [9] A. Marchet, et al., The prognostic value of N-ratio in patients with gastric cancer: validation in a large, multicenter series, Eur. J. Surg. Oncol. 34 (2) (2008) 159–165. [10] R. DerSimonian, N. Laird, Meta-analysis in clinical trials revisited, Contemp. Clin. Trials 45 (Pt A) (2015) 139–145. [11] J.P. Higgins, et al., Measuring inconsistency in meta-analyses, BMJ 327 (7414) (2003) 557–560. [12] C.B. Begg, M. Mazumdar, Operating characteristics of a rank correlation test for publication bias, Biometrics 50 (4) (1994) 1088–1101. [13] M. Egger, et al., Bias in meta-analysis detected by a simple, graphical test, BMJ 315 (7109) (1997) 629–634. [14] M.K. Parmar, L.A. Stewart, D.G. Altman, Meta-analyses of randomised trials: when the whole is more than just the sum of the parts, Br. J. Canc. 74 (4) (1996) 496–501. [15] S. Komatsu, et al., Evaluation of prognostic value and stage migration effect using positive lymph node ratio in gastric cancer, Eur. J. Surg. Oncol. 43 (1) (2017) 203–209. [16] L.Y. Zhao, et al., Superiority of lymph node ratio-based staging system for prognostic prediction in 2575 patients with gastric cancer: validation analysis in a large single center, Oncotarget 7 (32) (2016) 51069–51081. [17] C.M. Huang, et al., Prognostic impact of metastatic lymph node ratio in advanced gastric cancer from cardia and fundus, World J. Gastroenterol. 14 (27) (2008) 4383–4388. [18] C.J. Yu, et al., Prognostic value of metastatic lymph node ratio for patients with gastric cancer, Zhonghua Wei Chang Wai Ke Za Zhi 14 (7) (2011) 516–519. [19] J. Deng, et al., Suitability of 7th UICC N stage for predicting the overall survival of gastric cancer patients after curative resection in China, Ann. Surg Oncol. 17 (5) (2010) 1259–1266. [20] J. Zhou, et al., Lymph node ratio may predict the benefit of postoperative radiotherapy in node-positive cervical cancer, Oncotarget 7 (20) (2016) 29420–29428. [21] A. Isik, et al., A new prognostic strategy for gastric carcinoma: albumin level and metastatic lymph node ratio, Minerva Chir. 69 (3) (2014) 147–153. [22] G. Sun, et al., Lymph node ratio is a prognostic factor for non-small cell lung cancer, Oncotarget 6 (32) (2015) 33912–33918. [23] I. Sakcak, et al., Does N ratio affect survival in D1 and D2 lymph node dissection for gastric cancer? World J. Gastroenterol. 17 (35) (2011) 4007–4012. [24] Z. Sun, et al., The impact of N-ratio in minimizing stage migration phenomenon in gastric cancer patients with insufficient number or level of lymph node retrieved: results from a Chinese mono-institutional study in 2159 patients, Ann. Oncol. 20 (5) (2009) 897–905. [25] S. Ito, et al., Neoadjuvant chemotherapy followed by surgery in gastric cancer patients with extensive lymph node metastasis, World J. Clin. Oncol. 6 (6) (2015) 291–294. [26] D.Z. Xu, et al., Positive lymph node ratio is an independent prognostic factor in gastric cancer after d2 resection regardless of the examined number of lymph nodes, Ann. Surg Oncol. 16 (2) (2009) 319–326. [27] C. Liu, et al., Clinical implications of metastatic lymph node ratio in gastric cancer, BMC Canc. 7 (2007) 200. [28] B.Y. Zhang, et al., Evaluation of the prognostic value of the metastatic lymph node ratio for gastric cancer, Am. J. Surg. 207 (4) (2014) 555–565. [29] Alatengbaolide, et al., Lymph node ratio is an independent prognostic factor in gastric cancer after curative resection (R0) regardless of the examined number of lymph nodes, Am. J. Clin. Oncol. 36 (4) (2013) 325–330. [30] E. Bando, et al., Outcome of ratio of lymph node metastasis in gastric carcinoma, Ann. Surg Oncol. 9 (8) (2002) 775–784. [31] R. Persiani, et al., Metastatic lymph node ratio: a new staging system for gastric cancer, World J. Surg. 33 (10) (2009) 2106–2111. [32] O. Celen, E. Yildirim, U. Berberoglu, Prognostic impact of positive lymph node ratio in gastric carcinoma, J. Surg. Oncol. 96 (2) (2007) 95–101. [33] M. Zhang, et al., Prognostic significance of metastatic lymph nodes ratio in patients with gastric adenocarcinoma after curative gastrectomy, Chin Med J (Engl) 127 (10) (2014) 1874–1878. [34] S.R. Lee, et al., Prognostic significance of the metastatic lymph node ratio in patients with gastric cancer, World J. Surg. 36 (5) (2012) 1096–1101. [35] N. Fukuda, et al., Prognostic significance of the metastatic lymph node ratio in
Author contribution Specific author contributions: Yan Zhang and Haibo Lu provided experimental design; Jiang Zhu and Zhao Xue collected and analyzed the data; Jiang Zhu drafted the manuscript; Yan Zhang and Haibo Lu contributed to revision the manuscript; All authors commented on drafts of the paper and final approval of the manuscript. Conflicts of interest No. Trial registry number reviewregistry497. Guarantor Guarantors of the article: Jiang Zhu, Zhao Xue, Yan Zhang, Haibo Lu. Provenance and peer review Not commissioned, externally peer reviewed. Acknowledgments We thank Mark Abramovitz, PhD, from Liwen Bianji, Edanz Group China (www.liwenbianji.cn/ac), for editing the English text of a draft of this manuscript. Abbreviation In alphabetical order AJCC UICC TNM classification AT adjuvant therapy CI confidence interval DFS disease-free survival HR hazard ratio LNR lymph node ratio MA multivariate statistical analysis NA not available No. of nodes (N+) total number of lymph nodes harvested (number of positive lymph nodes) NR not report OS overall survival PRISMA Preferred reporting items for systematic reviews and metaanalyses TNM tumor node metastasis R retrospective SEER Surverillance, Epidmiology and End Results UA univariate statistical analysis Appendix A. Supplementary data Supplementary data related to this article can be found at https:// doi.org/10.1016/j.ijsu.2018.08.002. 82
International Journal of Surgery 57 (2018) 76–83
J. Zhu et al.
gastric cancer patients, World J. Surg. 33 (11) (2009) 2378–2382. [36] C. Jian-Hui, et al., Prognostic value of three different lymph node staging systems in the survival of patients with gastric cancer following D2 lymphadenectomy, Tumour Biol 37 (8) (2016) 11105–11113. [37] O. Asoglu, et al., Metastatic lymph node ratio is an independent prognostic factor in gastric cancer, Hepato-Gastroenterology 56 (91–92) (2009) 908–913. [38] A. Taghizadeh-Kermani, et al., Prognostic significance of metastatic lymph node ratio in patients with gastric cancer: an evaluation in north-East of Iran, Iran. J. Cancer Prev. 7 (2) (2014) 73–79. [39] S. Chen, et al., The prognostic value of harvested lymph nodes and the metastatic lymph node ratio for gastric cancer patients: results of a study of 1,101 patients, PLoS One 7 (11) (2012) e49424. [40] M. Sianesi, et al., The node ratio as prognostic factor after curative resection for gastric cancer, J. Gastrointest. Surg. 14 (4) (2010) 614–619. [41] C. Pedrazzani, et al., Ratio between metastatic and examined lymph nodes (N ratio) may have low clinical utility in gastric cancer patients treated by limited
[42]
[43]
[44] [45] [46]
83
lymphadenectomy: results from a single-center experience of 526 patients, World J. Surg. 34 (1) (2010) 85–91. D. Nitti, et al., Ratio between metastatic and examined lymph nodes is an independent prognostic factor after D2 resection for gastric cancer: analysis of a large European monoinstitutional experience, Ann. Surg Oncol. 10 (9) (2003) 1077–1085. X. Wang, et al., Prognostic value of the ratio of metastatic lymph nodes in gastric cancer: an analysis based on a Chinese population, J. Surg. Oncol. 99 (6) (2009) 329–334. L.B. Xiao, et al., Superiority of metastatic lymph node ratio to the 7th edition UICC N staging in gastric cancer, World J. Gastroenterol. 17 (46) (2011) 5123–5130. H. Liu, et al., The RML of lymph node metastasis was superior to the LODDS for evaluating the prognosis of gastric cancer, Int. J. Surg. 11 (5) (2013) 419–424. M.Z. Qiu, et al., The tumor-log odds of positive lymph nodes-metastasis staging system, a promising new staging system for gastric cancer after D2 resection in China, PLoS One 7 (2) (2012) e31736.